From 04b18c15f48a0b2b0996132f68e7186a5195986d Mon Sep 17 00:00:00 2001
From: Kerry McLaughlin <kerry.mclaughlin@arm.com>
Date: Fri, 3 Nov 2023 14:28:46 +0000
Subject: [PATCH 1/5] [Clang][SVE2] Add builtins for moving multi-vectors
 to/from ZA

Adds the following SME2 builtins:
 - svread_hor/ver,
 - svwrite_hor/ver,
 - svread_za64,
 - svwrite_za64

See https://github.com/ARM-software/acle/pull/217
---
 clang/include/clang/Basic/arm_sme.td          |   45 +
 .../aarch64-sme2-intrinsics/acle_sme2_read.c  | 1422 +++++++++++++++++
 .../aarch64-sme2-intrinsics/acle_sme2_write.c | 1097 +++++++++++++
 .../aarch64-sme2-intrinsics/acle_sme2_imm.cpp |   51 +
 4 files changed, 2615 insertions(+)
 create mode 100644 clang/test/CodeGen/aarch64-sme2-intrinsics/acle_sme2_read.c
 create mode 100644 clang/test/CodeGen/aarch64-sme2-intrinsics/acle_sme2_write.c

diff --git a/clang/include/clang/Basic/arm_sme.td b/clang/include/clang/Basic/arm_sme.td
index 34dbfff6c4c85..b97629d943334 100644
--- a/clang/include/clang/Basic/arm_sme.td
+++ b/clang/include/clang/Basic/arm_sme.td
@@ -299,6 +299,51 @@ multiclass ZAAddSub<string n_suffix> {
 defm SVADD : ZAAddSub<"add">;
 defm SVSUB : ZAAddSub<"sub">;
 
+// SME2 - MOVA
+
+//
+// 2 and 4 vector-group read/write intrinsics.
+//
+
+multiclass WriteHV_VG<string n, string t, string i, list<ImmCheck> checks> {
+  let TargetGuard = "sme2" in {
+    def NAME # _VG2_H : Inst<"svwrite_hor_" # n # "_vg2", "vim2", t, MergeNone, i # "_hor_vg2", [IsSharedZA, IsStreaming], checks>;
+    def NAME # _VG2_V : Inst<"svwrite_ver_" # n # "_vg2", "vim2", t, MergeNone, i # "_ver_vg2", [IsSharedZA, IsStreaming], checks>;
+    def NAME # _VG4_H : Inst<"svwrite_hor_" # n # "_vg4", "vim4", t, MergeNone, i # "_hor_vg4", [IsSharedZA, IsStreaming], checks>;
+    def NAME # _VG4_V : Inst<"svwrite_ver_" # n # "_vg4", "vim4", t, MergeNone, i # "_ver_vg4", [IsSharedZA, IsStreaming], checks>;
+  }
+}
+
+defm SVWRITE_ZA8  : WriteHV_VG<"za8[_{d}]",  "cUc",   "aarch64_sme_write", [ImmCheck<0, ImmCheck0_0>]>;
+defm SVWRITE_ZA16 : WriteHV_VG<"za16[_{d}]", "sUshb", "aarch64_sme_write", [ImmCheck<0, ImmCheck0_1>]>;
+defm SVWRITE_ZA32 : WriteHV_VG<"za32[_{d}]", "iUif",  "aarch64_sme_write", [ImmCheck<0, ImmCheck0_3>]>;
+defm SVWRITE_ZA64 : WriteHV_VG<"za64[_{d}]", "lUld",  "aarch64_sme_write", [ImmCheck<0, ImmCheck0_7>]>;
+
+multiclass ReadHV_VG<string n, string t, string i, list<ImmCheck> checks> {
+  let TargetGuard = "sme2" in {
+    def NAME # _VG2_H : Inst<"svread_hor_" # n # "_vg2", "2im", t, MergeNone, i # "_hor_vg2", [IsSharedZA, IsPreservesZA, IsStreaming], checks>;
+    def NAME # _VG2_V : Inst<"svread_ver_" # n # "_vg2", "2im", t, MergeNone, i # "_ver_vg2", [IsSharedZA, IsPreservesZA, IsStreaming], checks>;
+    def NAME # _VG4_H : Inst<"svread_hor_" # n # "_vg4", "4im", t, MergeNone, i # "_hor_vg4", [IsSharedZA, IsPreservesZA, IsStreaming], checks>;
+    def NAME # _VG4_V : Inst<"svread_ver_" # n # "_vg4", "4im", t, MergeNone, i # "_ver_vg4", [IsSharedZA, IsPreservesZA, IsStreaming], checks>;
+  }
+}
+
+defm SVREAD_ZA8  : ReadHV_VG<"za8_{d}",  "cUc",   "aarch64_sme_read", [ImmCheck<0, ImmCheck0_0>]>;
+defm SVREAD_ZA16 : ReadHV_VG<"za16_{d}", "sUshb", "aarch64_sme_read", [ImmCheck<0, ImmCheck0_1>]>;
+defm SVREAD_ZA32 : ReadHV_VG<"za32_{d}", "iUif",  "aarch64_sme_read", [ImmCheck<0, ImmCheck0_3>]>;
+defm SVREAD_ZA64 : ReadHV_VG<"za64_{d}", "lUld",  "aarch64_sme_read", [ImmCheck<0, ImmCheck0_7>]>;
+
+//
+// Single vector-group read/write intrinsics.
+//
+
+let TargetGuard = "sme2" in {
+  def SVWRITE_ZA64_VG1x2 : Inst<"svwrite_za64[_{d}]_vg1x2", "vm2", "lUld", MergeNone, "aarch64_sme_write_vg1x2", [IsSharedZA, IsStreaming], []>;
+  def SVWRITE_ZA64_VG1x4 : Inst<"svwrite_za64[_{d}]_vg1x4", "vm4", "lUld", MergeNone, "aarch64_sme_write_vg1x4", [IsSharedZA, IsStreaming], []>;
+  def SVREAD_ZA64_VG1x2  : Inst<"svread_za64_{d}_vg1x2",    "2m",  "lUld", MergeNone, "aarch64_sme_read_vg1x2",  [IsSharedZA, IsPreservesZA, IsStreaming], []>;
+  def SVREAD_ZA64_VG1x4  : Inst<"svread_za64_{d}_vg1x4",    "4m",  "lUld", MergeNone, "aarch64_sme_read_vg1x4",  [IsSharedZA, IsPreservesZA, IsStreaming], []>;
+}
+
 //
 // Outer product and accumulate/subtract
 //
diff --git a/clang/test/CodeGen/aarch64-sme2-intrinsics/acle_sme2_read.c b/clang/test/CodeGen/aarch64-sme2-intrinsics/acle_sme2_read.c
new file mode 100644
index 0000000000000..706403e5180ad
--- /dev/null
+++ b/clang/test/CodeGen/aarch64-sme2-intrinsics/acle_sme2_read.c
@@ -0,0 +1,1422 @@
+// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py
+// REQUIRES: aarch64-registered-target
+
+// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sme2 -target-feature +sve -S -disable-O0-optnone -Werror -Wall -emit-llvm -o - %s | opt -S -p mem2reg,instcombine,tailcallelim | FileCheck %s
+// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sme2 -target-feature +sve -S -disable-O0-optnone -Werror -Wall -emit-llvm -o - -x c++ %s | opt -S -p mem2reg,instcombine,tailcallelim | FileCheck %s -check-prefix=CPP-CHECK
+// RUN: %clang_cc1 -DSVE_OVERLOADED_FORMS -triple aarch64-none-linux-gnu -target-feature +sme2 -target-feature +sve -S -disable-O0-optnone -Werror -Wall -emit-llvm -o - %s | opt -S -p mem2reg,instcombine,tailcallelim | FileCheck %s
+// RUN: %clang_cc1 -DSVE_OVERLOADED_FORMS -triple aarch64-none-linux-gnu -target-feature +sme2 -target-feature +sve -S -disable-O0-optnone -Werror -Wall -emit-llvm -o - -x c++ %s | opt -S -p mem2reg,instcombine,tailcallelim | FileCheck %s -check-prefix=CPP-CHECK
+// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sme2 -target-feature +sve -S -disable-O0-optnone -Werror -Wall -o /dev/null %s
+
+#include <arm_sme_draft_spec_subject_to_change.h>
+
+#ifdef SVE_OVERLOADED_FORMS
+// A simple used,unused... macro, long enough to represent any SVE builtin.
+#define SVE_ACLE_FUNC(A1,A2_UNUSED,A3,A4_UNUSED) A1##A3
+#else
+#define SVE_ACLE_FUNC(A1,A2,A3,A4) A1##A2##A3##A4
+#endif
+
+// CHECK-LABEL: @test_svread_ver_za8_u8_vg2(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call { <vscale x 16 x i8>, <vscale x 16 x i8> } @llvm.aarch64.sme.read.ver.vg2.nxv16i8(i32 0, i32 [[BASE:%.*]])
+// CHECK-NEXT:    [[TMP1:%.*]] = extractvalue { <vscale x 16 x i8>, <vscale x 16 x i8> } [[TMP0]], 0
+// CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 32 x i8> @llvm.vector.insert.nxv32i8.nxv16i8(<vscale x 32 x i8> poison, <vscale x 16 x i8> [[TMP1]], i64 0)
+// CHECK-NEXT:    [[TMP3:%.*]] = extractvalue { <vscale x 16 x i8>, <vscale x 16 x i8> } [[TMP0]], 1
+// CHECK-NEXT:    [[TMP4:%.*]] = tail call <vscale x 32 x i8> @llvm.vector.insert.nxv32i8.nxv16i8(<vscale x 32 x i8> [[TMP2]], <vscale x 16 x i8> [[TMP3]], i64 16)
+// CHECK-NEXT:    ret <vscale x 32 x i8> [[TMP4]]
+//
+// CPP-CHECK-LABEL: @_Z26test_svread_ver_za8_u8_vg2j(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call { <vscale x 16 x i8>, <vscale x 16 x i8> } @llvm.aarch64.sme.read.ver.vg2.nxv16i8(i32 0, i32 [[BASE:%.*]])
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = extractvalue { <vscale x 16 x i8>, <vscale x 16 x i8> } [[TMP0]], 0
+// CPP-CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 32 x i8> @llvm.vector.insert.nxv32i8.nxv16i8(<vscale x 32 x i8> poison, <vscale x 16 x i8> [[TMP1]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP3:%.*]] = extractvalue { <vscale x 16 x i8>, <vscale x 16 x i8> } [[TMP0]], 1
+// CPP-CHECK-NEXT:    [[TMP4:%.*]] = tail call <vscale x 32 x i8> @llvm.vector.insert.nxv32i8.nxv16i8(<vscale x 32 x i8> [[TMP2]], <vscale x 16 x i8> [[TMP3]], i64 16)
+// CPP-CHECK-NEXT:    ret <vscale x 32 x i8> [[TMP4]]
+//
+svuint8x2_t test_svread_ver_za8_u8_vg2(uint32_t base) {
+  return svread_ver_za8_u8_vg2(0, base);
+}
+
+// CHECK-LABEL: @test_svread_ver_za8_s8_vg2(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call { <vscale x 16 x i8>, <vscale x 16 x i8> } @llvm.aarch64.sme.read.ver.vg2.nxv16i8(i32 0, i32 [[BASE:%.*]])
+// CHECK-NEXT:    [[TMP1:%.*]] = extractvalue { <vscale x 16 x i8>, <vscale x 16 x i8> } [[TMP0]], 0
+// CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 32 x i8> @llvm.vector.insert.nxv32i8.nxv16i8(<vscale x 32 x i8> poison, <vscale x 16 x i8> [[TMP1]], i64 0)
+// CHECK-NEXT:    [[TMP3:%.*]] = extractvalue { <vscale x 16 x i8>, <vscale x 16 x i8> } [[TMP0]], 1
+// CHECK-NEXT:    [[TMP4:%.*]] = tail call <vscale x 32 x i8> @llvm.vector.insert.nxv32i8.nxv16i8(<vscale x 32 x i8> [[TMP2]], <vscale x 16 x i8> [[TMP3]], i64 16)
+// CHECK-NEXT:    ret <vscale x 32 x i8> [[TMP4]]
+//
+// CPP-CHECK-LABEL: @_Z26test_svread_ver_za8_s8_vg2j(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call { <vscale x 16 x i8>, <vscale x 16 x i8> } @llvm.aarch64.sme.read.ver.vg2.nxv16i8(i32 0, i32 [[BASE:%.*]])
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = extractvalue { <vscale x 16 x i8>, <vscale x 16 x i8> } [[TMP0]], 0
+// CPP-CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 32 x i8> @llvm.vector.insert.nxv32i8.nxv16i8(<vscale x 32 x i8> poison, <vscale x 16 x i8> [[TMP1]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP3:%.*]] = extractvalue { <vscale x 16 x i8>, <vscale x 16 x i8> } [[TMP0]], 1
+// CPP-CHECK-NEXT:    [[TMP4:%.*]] = tail call <vscale x 32 x i8> @llvm.vector.insert.nxv32i8.nxv16i8(<vscale x 32 x i8> [[TMP2]], <vscale x 16 x i8> [[TMP3]], i64 16)
+// CPP-CHECK-NEXT:    ret <vscale x 32 x i8> [[TMP4]]
+//
+svint8x2_t test_svread_ver_za8_s8_vg2(uint32_t base) {
+  return svread_ver_za8_s8_vg2(0, base);
+}
+
+// CHECK-LABEL: @test_svread_hor_za8_u8_vg2(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call { <vscale x 16 x i8>, <vscale x 16 x i8> } @llvm.aarch64.sme.read.hor.vg2.nxv16i8(i32 0, i32 [[BASE:%.*]])
+// CHECK-NEXT:    [[TMP1:%.*]] = extractvalue { <vscale x 16 x i8>, <vscale x 16 x i8> } [[TMP0]], 0
+// CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 32 x i8> @llvm.vector.insert.nxv32i8.nxv16i8(<vscale x 32 x i8> poison, <vscale x 16 x i8> [[TMP1]], i64 0)
+// CHECK-NEXT:    [[TMP3:%.*]] = extractvalue { <vscale x 16 x i8>, <vscale x 16 x i8> } [[TMP0]], 1
+// CHECK-NEXT:    [[TMP4:%.*]] = tail call <vscale x 32 x i8> @llvm.vector.insert.nxv32i8.nxv16i8(<vscale x 32 x i8> [[TMP2]], <vscale x 16 x i8> [[TMP3]], i64 16)
+// CHECK-NEXT:    ret <vscale x 32 x i8> [[TMP4]]
+//
+// CPP-CHECK-LABEL: @_Z26test_svread_hor_za8_u8_vg2j(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call { <vscale x 16 x i8>, <vscale x 16 x i8> } @llvm.aarch64.sme.read.hor.vg2.nxv16i8(i32 0, i32 [[BASE:%.*]])
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = extractvalue { <vscale x 16 x i8>, <vscale x 16 x i8> } [[TMP0]], 0
+// CPP-CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 32 x i8> @llvm.vector.insert.nxv32i8.nxv16i8(<vscale x 32 x i8> poison, <vscale x 16 x i8> [[TMP1]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP3:%.*]] = extractvalue { <vscale x 16 x i8>, <vscale x 16 x i8> } [[TMP0]], 1
+// CPP-CHECK-NEXT:    [[TMP4:%.*]] = tail call <vscale x 32 x i8> @llvm.vector.insert.nxv32i8.nxv16i8(<vscale x 32 x i8> [[TMP2]], <vscale x 16 x i8> [[TMP3]], i64 16)
+// CPP-CHECK-NEXT:    ret <vscale x 32 x i8> [[TMP4]]
+//
+svuint8x2_t test_svread_hor_za8_u8_vg2(uint32_t base) {
+  return svread_hor_za8_u8_vg2(0, base);
+}
+
+// CHECK-LABEL: @test_svread_hor_za8_s8_vg2(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call { <vscale x 16 x i8>, <vscale x 16 x i8> } @llvm.aarch64.sme.read.hor.vg2.nxv16i8(i32 0, i32 [[BASE:%.*]])
+// CHECK-NEXT:    [[TMP1:%.*]] = extractvalue { <vscale x 16 x i8>, <vscale x 16 x i8> } [[TMP0]], 0
+// CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 32 x i8> @llvm.vector.insert.nxv32i8.nxv16i8(<vscale x 32 x i8> poison, <vscale x 16 x i8> [[TMP1]], i64 0)
+// CHECK-NEXT:    [[TMP3:%.*]] = extractvalue { <vscale x 16 x i8>, <vscale x 16 x i8> } [[TMP0]], 1
+// CHECK-NEXT:    [[TMP4:%.*]] = tail call <vscale x 32 x i8> @llvm.vector.insert.nxv32i8.nxv16i8(<vscale x 32 x i8> [[TMP2]], <vscale x 16 x i8> [[TMP3]], i64 16)
+// CHECK-NEXT:    ret <vscale x 32 x i8> [[TMP4]]
+//
+// CPP-CHECK-LABEL: @_Z26test_svread_hor_za8_s8_vg2j(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call { <vscale x 16 x i8>, <vscale x 16 x i8> } @llvm.aarch64.sme.read.hor.vg2.nxv16i8(i32 0, i32 [[BASE:%.*]])
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = extractvalue { <vscale x 16 x i8>, <vscale x 16 x i8> } [[TMP0]], 0
+// CPP-CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 32 x i8> @llvm.vector.insert.nxv32i8.nxv16i8(<vscale x 32 x i8> poison, <vscale x 16 x i8> [[TMP1]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP3:%.*]] = extractvalue { <vscale x 16 x i8>, <vscale x 16 x i8> } [[TMP0]], 1
+// CPP-CHECK-NEXT:    [[TMP4:%.*]] = tail call <vscale x 32 x i8> @llvm.vector.insert.nxv32i8.nxv16i8(<vscale x 32 x i8> [[TMP2]], <vscale x 16 x i8> [[TMP3]], i64 16)
+// CPP-CHECK-NEXT:    ret <vscale x 32 x i8> [[TMP4]]
+//
+svint8x2_t test_svread_hor_za8_s8_vg2(uint32_t base) {
+  return svread_hor_za8_s8_vg2(0, base);
+}
+
+// CHECK-LABEL: @test_svread_hor_za8_u8_vg4(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call { <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8> } @llvm.aarch64.sme.read.hor.vg4.nxv16i8(i32 0, i32 [[BASE:%.*]])
+// CHECK-NEXT:    [[TMP1:%.*]] = extractvalue { <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8> } [[TMP0]], 0
+// CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 64 x i8> @llvm.vector.insert.nxv64i8.nxv16i8(<vscale x 64 x i8> poison, <vscale x 16 x i8> [[TMP1]], i64 0)
+// CHECK-NEXT:    [[TMP3:%.*]] = extractvalue { <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8> } [[TMP0]], 1
+// CHECK-NEXT:    [[TMP4:%.*]] = tail call <vscale x 64 x i8> @llvm.vector.insert.nxv64i8.nxv16i8(<vscale x 64 x i8> [[TMP2]], <vscale x 16 x i8> [[TMP3]], i64 16)
+// CHECK-NEXT:    [[TMP5:%.*]] = extractvalue { <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8> } [[TMP0]], 2
+// CHECK-NEXT:    [[TMP6:%.*]] = tail call <vscale x 64 x i8> @llvm.vector.insert.nxv64i8.nxv16i8(<vscale x 64 x i8> [[TMP4]], <vscale x 16 x i8> [[TMP5]], i64 32)
+// CHECK-NEXT:    [[TMP7:%.*]] = extractvalue { <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8> } [[TMP0]], 3
+// CHECK-NEXT:    [[TMP8:%.*]] = tail call <vscale x 64 x i8> @llvm.vector.insert.nxv64i8.nxv16i8(<vscale x 64 x i8> [[TMP6]], <vscale x 16 x i8> [[TMP7]], i64 48)
+// CHECK-NEXT:    ret <vscale x 64 x i8> [[TMP8]]
+//
+// CPP-CHECK-LABEL: @_Z26test_svread_hor_za8_u8_vg4j(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call { <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8> } @llvm.aarch64.sme.read.hor.vg4.nxv16i8(i32 0, i32 [[BASE:%.*]])
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = extractvalue { <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8> } [[TMP0]], 0
+// CPP-CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 64 x i8> @llvm.vector.insert.nxv64i8.nxv16i8(<vscale x 64 x i8> poison, <vscale x 16 x i8> [[TMP1]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP3:%.*]] = extractvalue { <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8> } [[TMP0]], 1
+// CPP-CHECK-NEXT:    [[TMP4:%.*]] = tail call <vscale x 64 x i8> @llvm.vector.insert.nxv64i8.nxv16i8(<vscale x 64 x i8> [[TMP2]], <vscale x 16 x i8> [[TMP3]], i64 16)
+// CPP-CHECK-NEXT:    [[TMP5:%.*]] = extractvalue { <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8> } [[TMP0]], 2
+// CPP-CHECK-NEXT:    [[TMP6:%.*]] = tail call <vscale x 64 x i8> @llvm.vector.insert.nxv64i8.nxv16i8(<vscale x 64 x i8> [[TMP4]], <vscale x 16 x i8> [[TMP5]], i64 32)
+// CPP-CHECK-NEXT:    [[TMP7:%.*]] = extractvalue { <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8> } [[TMP0]], 3
+// CPP-CHECK-NEXT:    [[TMP8:%.*]] = tail call <vscale x 64 x i8> @llvm.vector.insert.nxv64i8.nxv16i8(<vscale x 64 x i8> [[TMP6]], <vscale x 16 x i8> [[TMP7]], i64 48)
+// CPP-CHECK-NEXT:    ret <vscale x 64 x i8> [[TMP8]]
+//
+svuint8x4_t test_svread_hor_za8_u8_vg4(uint32_t base) {
+  return svread_hor_za8_u8_vg4(0, base);
+}
+
+// CHECK-LABEL: @test_svread_hor_za8_s8_vg4(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call { <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8> } @llvm.aarch64.sme.read.hor.vg4.nxv16i8(i32 0, i32 [[BASE:%.*]])
+// CHECK-NEXT:    [[TMP1:%.*]] = extractvalue { <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8> } [[TMP0]], 0
+// CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 64 x i8> @llvm.vector.insert.nxv64i8.nxv16i8(<vscale x 64 x i8> poison, <vscale x 16 x i8> [[TMP1]], i64 0)
+// CHECK-NEXT:    [[TMP3:%.*]] = extractvalue { <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8> } [[TMP0]], 1
+// CHECK-NEXT:    [[TMP4:%.*]] = tail call <vscale x 64 x i8> @llvm.vector.insert.nxv64i8.nxv16i8(<vscale x 64 x i8> [[TMP2]], <vscale x 16 x i8> [[TMP3]], i64 16)
+// CHECK-NEXT:    [[TMP5:%.*]] = extractvalue { <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8> } [[TMP0]], 2
+// CHECK-NEXT:    [[TMP6:%.*]] = tail call <vscale x 64 x i8> @llvm.vector.insert.nxv64i8.nxv16i8(<vscale x 64 x i8> [[TMP4]], <vscale x 16 x i8> [[TMP5]], i64 32)
+// CHECK-NEXT:    [[TMP7:%.*]] = extractvalue { <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8> } [[TMP0]], 3
+// CHECK-NEXT:    [[TMP8:%.*]] = tail call <vscale x 64 x i8> @llvm.vector.insert.nxv64i8.nxv16i8(<vscale x 64 x i8> [[TMP6]], <vscale x 16 x i8> [[TMP7]], i64 48)
+// CHECK-NEXT:    ret <vscale x 64 x i8> [[TMP8]]
+//
+// CPP-CHECK-LABEL: @_Z26test_svread_hor_za8_s8_vg4j(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call { <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8> } @llvm.aarch64.sme.read.hor.vg4.nxv16i8(i32 0, i32 [[BASE:%.*]])
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = extractvalue { <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8> } [[TMP0]], 0
+// CPP-CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 64 x i8> @llvm.vector.insert.nxv64i8.nxv16i8(<vscale x 64 x i8> poison, <vscale x 16 x i8> [[TMP1]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP3:%.*]] = extractvalue { <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8> } [[TMP0]], 1
+// CPP-CHECK-NEXT:    [[TMP4:%.*]] = tail call <vscale x 64 x i8> @llvm.vector.insert.nxv64i8.nxv16i8(<vscale x 64 x i8> [[TMP2]], <vscale x 16 x i8> [[TMP3]], i64 16)
+// CPP-CHECK-NEXT:    [[TMP5:%.*]] = extractvalue { <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8> } [[TMP0]], 2
+// CPP-CHECK-NEXT:    [[TMP6:%.*]] = tail call <vscale x 64 x i8> @llvm.vector.insert.nxv64i8.nxv16i8(<vscale x 64 x i8> [[TMP4]], <vscale x 16 x i8> [[TMP5]], i64 32)
+// CPP-CHECK-NEXT:    [[TMP7:%.*]] = extractvalue { <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8> } [[TMP0]], 3
+// CPP-CHECK-NEXT:    [[TMP8:%.*]] = tail call <vscale x 64 x i8> @llvm.vector.insert.nxv64i8.nxv16i8(<vscale x 64 x i8> [[TMP6]], <vscale x 16 x i8> [[TMP7]], i64 48)
+// CPP-CHECK-NEXT:    ret <vscale x 64 x i8> [[TMP8]]
+//
+svint8x4_t test_svread_hor_za8_s8_vg4(uint32_t base) {
+  return svread_hor_za8_s8_vg4(0, base);
+}
+
+// CHECK-LABEL: @test_svread_ver_za8_u8_vg4(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call { <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8> } @llvm.aarch64.sme.read.ver.vg4.nxv16i8(i32 0, i32 [[BASE:%.*]])
+// CHECK-NEXT:    [[TMP1:%.*]] = extractvalue { <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8> } [[TMP0]], 0
+// CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 64 x i8> @llvm.vector.insert.nxv64i8.nxv16i8(<vscale x 64 x i8> poison, <vscale x 16 x i8> [[TMP1]], i64 0)
+// CHECK-NEXT:    [[TMP3:%.*]] = extractvalue { <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8> } [[TMP0]], 1
+// CHECK-NEXT:    [[TMP4:%.*]] = tail call <vscale x 64 x i8> @llvm.vector.insert.nxv64i8.nxv16i8(<vscale x 64 x i8> [[TMP2]], <vscale x 16 x i8> [[TMP3]], i64 16)
+// CHECK-NEXT:    [[TMP5:%.*]] = extractvalue { <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8> } [[TMP0]], 2
+// CHECK-NEXT:    [[TMP6:%.*]] = tail call <vscale x 64 x i8> @llvm.vector.insert.nxv64i8.nxv16i8(<vscale x 64 x i8> [[TMP4]], <vscale x 16 x i8> [[TMP5]], i64 32)
+// CHECK-NEXT:    [[TMP7:%.*]] = extractvalue { <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8> } [[TMP0]], 3
+// CHECK-NEXT:    [[TMP8:%.*]] = tail call <vscale x 64 x i8> @llvm.vector.insert.nxv64i8.nxv16i8(<vscale x 64 x i8> [[TMP6]], <vscale x 16 x i8> [[TMP7]], i64 48)
+// CHECK-NEXT:    ret <vscale x 64 x i8> [[TMP8]]
+//
+// CPP-CHECK-LABEL: @_Z26test_svread_ver_za8_u8_vg4j(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call { <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8> } @llvm.aarch64.sme.read.ver.vg4.nxv16i8(i32 0, i32 [[BASE:%.*]])
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = extractvalue { <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8> } [[TMP0]], 0
+// CPP-CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 64 x i8> @llvm.vector.insert.nxv64i8.nxv16i8(<vscale x 64 x i8> poison, <vscale x 16 x i8> [[TMP1]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP3:%.*]] = extractvalue { <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8> } [[TMP0]], 1
+// CPP-CHECK-NEXT:    [[TMP4:%.*]] = tail call <vscale x 64 x i8> @llvm.vector.insert.nxv64i8.nxv16i8(<vscale x 64 x i8> [[TMP2]], <vscale x 16 x i8> [[TMP3]], i64 16)
+// CPP-CHECK-NEXT:    [[TMP5:%.*]] = extractvalue { <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8> } [[TMP0]], 2
+// CPP-CHECK-NEXT:    [[TMP6:%.*]] = tail call <vscale x 64 x i8> @llvm.vector.insert.nxv64i8.nxv16i8(<vscale x 64 x i8> [[TMP4]], <vscale x 16 x i8> [[TMP5]], i64 32)
+// CPP-CHECK-NEXT:    [[TMP7:%.*]] = extractvalue { <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8> } [[TMP0]], 3
+// CPP-CHECK-NEXT:    [[TMP8:%.*]] = tail call <vscale x 64 x i8> @llvm.vector.insert.nxv64i8.nxv16i8(<vscale x 64 x i8> [[TMP6]], <vscale x 16 x i8> [[TMP7]], i64 48)
+// CPP-CHECK-NEXT:    ret <vscale x 64 x i8> [[TMP8]]
+//
+svuint8x4_t test_svread_ver_za8_u8_vg4(uint32_t base) {
+  return svread_ver_za8_u8_vg4(0, base);
+}
+
+// CHECK-LABEL: @test_svread_ver_za8_s8_vg4(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call { <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8> } @llvm.aarch64.sme.read.ver.vg4.nxv16i8(i32 0, i32 [[BASE:%.*]])
+// CHECK-NEXT:    [[TMP1:%.*]] = extractvalue { <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8> } [[TMP0]], 0
+// CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 64 x i8> @llvm.vector.insert.nxv64i8.nxv16i8(<vscale x 64 x i8> poison, <vscale x 16 x i8> [[TMP1]], i64 0)
+// CHECK-NEXT:    [[TMP3:%.*]] = extractvalue { <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8> } [[TMP0]], 1
+// CHECK-NEXT:    [[TMP4:%.*]] = tail call <vscale x 64 x i8> @llvm.vector.insert.nxv64i8.nxv16i8(<vscale x 64 x i8> [[TMP2]], <vscale x 16 x i8> [[TMP3]], i64 16)
+// CHECK-NEXT:    [[TMP5:%.*]] = extractvalue { <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8> } [[TMP0]], 2
+// CHECK-NEXT:    [[TMP6:%.*]] = tail call <vscale x 64 x i8> @llvm.vector.insert.nxv64i8.nxv16i8(<vscale x 64 x i8> [[TMP4]], <vscale x 16 x i8> [[TMP5]], i64 32)
+// CHECK-NEXT:    [[TMP7:%.*]] = extractvalue { <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8> } [[TMP0]], 3
+// CHECK-NEXT:    [[TMP8:%.*]] = tail call <vscale x 64 x i8> @llvm.vector.insert.nxv64i8.nxv16i8(<vscale x 64 x i8> [[TMP6]], <vscale x 16 x i8> [[TMP7]], i64 48)
+// CHECK-NEXT:    ret <vscale x 64 x i8> [[TMP8]]
+//
+// CPP-CHECK-LABEL: @_Z26test_svread_ver_za8_s8_vg4j(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call { <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8> } @llvm.aarch64.sme.read.ver.vg4.nxv16i8(i32 0, i32 [[BASE:%.*]])
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = extractvalue { <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8> } [[TMP0]], 0
+// CPP-CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 64 x i8> @llvm.vector.insert.nxv64i8.nxv16i8(<vscale x 64 x i8> poison, <vscale x 16 x i8> [[TMP1]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP3:%.*]] = extractvalue { <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8> } [[TMP0]], 1
+// CPP-CHECK-NEXT:    [[TMP4:%.*]] = tail call <vscale x 64 x i8> @llvm.vector.insert.nxv64i8.nxv16i8(<vscale x 64 x i8> [[TMP2]], <vscale x 16 x i8> [[TMP3]], i64 16)
+// CPP-CHECK-NEXT:    [[TMP5:%.*]] = extractvalue { <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8> } [[TMP0]], 2
+// CPP-CHECK-NEXT:    [[TMP6:%.*]] = tail call <vscale x 64 x i8> @llvm.vector.insert.nxv64i8.nxv16i8(<vscale x 64 x i8> [[TMP4]], <vscale x 16 x i8> [[TMP5]], i64 32)
+// CPP-CHECK-NEXT:    [[TMP7:%.*]] = extractvalue { <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8> } [[TMP0]], 3
+// CPP-CHECK-NEXT:    [[TMP8:%.*]] = tail call <vscale x 64 x i8> @llvm.vector.insert.nxv64i8.nxv16i8(<vscale x 64 x i8> [[TMP6]], <vscale x 16 x i8> [[TMP7]], i64 48)
+// CPP-CHECK-NEXT:    ret <vscale x 64 x i8> [[TMP8]]
+//
+svint8x4_t test_svread_ver_za8_s8_vg4(uint32_t base) {
+  return svread_ver_za8_s8_vg4(0, base);
+}
+
+// CHECK-LABEL: @test_svread_hor_za16_u16_vg2(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call { <vscale x 8 x i16>, <vscale x 8 x i16> } @llvm.aarch64.sme.read.hor.vg2.nxv8i16(i32 1, i32 [[BASE:%.*]])
+// CHECK-NEXT:    [[TMP1:%.*]] = extractvalue { <vscale x 8 x i16>, <vscale x 8 x i16> } [[TMP0]], 0
+// CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 16 x i16> @llvm.vector.insert.nxv16i16.nxv8i16(<vscale x 16 x i16> poison, <vscale x 8 x i16> [[TMP1]], i64 0)
+// CHECK-NEXT:    [[TMP3:%.*]] = extractvalue { <vscale x 8 x i16>, <vscale x 8 x i16> } [[TMP0]], 1
+// CHECK-NEXT:    [[TMP4:%.*]] = tail call <vscale x 16 x i16> @llvm.vector.insert.nxv16i16.nxv8i16(<vscale x 16 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP3]], i64 8)
+// CHECK-NEXT:    ret <vscale x 16 x i16> [[TMP4]]
+//
+// CPP-CHECK-LABEL: @_Z28test_svread_hor_za16_u16_vg2j(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call { <vscale x 8 x i16>, <vscale x 8 x i16> } @llvm.aarch64.sme.read.hor.vg2.nxv8i16(i32 1, i32 [[BASE:%.*]])
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = extractvalue { <vscale x 8 x i16>, <vscale x 8 x i16> } [[TMP0]], 0
+// CPP-CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 16 x i16> @llvm.vector.insert.nxv16i16.nxv8i16(<vscale x 16 x i16> poison, <vscale x 8 x i16> [[TMP1]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP3:%.*]] = extractvalue { <vscale x 8 x i16>, <vscale x 8 x i16> } [[TMP0]], 1
+// CPP-CHECK-NEXT:    [[TMP4:%.*]] = tail call <vscale x 16 x i16> @llvm.vector.insert.nxv16i16.nxv8i16(<vscale x 16 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP3]], i64 8)
+// CPP-CHECK-NEXT:    ret <vscale x 16 x i16> [[TMP4]]
+//
+svuint16x2_t test_svread_hor_za16_u16_vg2(uint32_t base) {
+  return svread_hor_za16_u16_vg2(1, base);
+}
+
+// CHECK-LABEL: @test_svread_hor_za16_bf16_vg2(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call { <vscale x 8 x bfloat>, <vscale x 8 x bfloat> } @llvm.aarch64.sme.read.hor.vg2.nxv8bf16(i32 1, i32 [[BASE:%.*]])
+// CHECK-NEXT:    [[TMP1:%.*]] = extractvalue { <vscale x 8 x bfloat>, <vscale x 8 x bfloat> } [[TMP0]], 0
+// CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 16 x bfloat> @llvm.vector.insert.nxv16bf16.nxv8bf16(<vscale x 16 x bfloat> poison, <vscale x 8 x bfloat> [[TMP1]], i64 0)
+// CHECK-NEXT:    [[TMP3:%.*]] = extractvalue { <vscale x 8 x bfloat>, <vscale x 8 x bfloat> } [[TMP0]], 1
+// CHECK-NEXT:    [[TMP4:%.*]] = tail call <vscale x 16 x bfloat> @llvm.vector.insert.nxv16bf16.nxv8bf16(<vscale x 16 x bfloat> [[TMP2]], <vscale x 8 x bfloat> [[TMP3]], i64 8)
+// CHECK-NEXT:    ret <vscale x 16 x bfloat> [[TMP4]]
+//
+// CPP-CHECK-LABEL: @_Z29test_svread_hor_za16_bf16_vg2j(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call { <vscale x 8 x bfloat>, <vscale x 8 x bfloat> } @llvm.aarch64.sme.read.hor.vg2.nxv8bf16(i32 1, i32 [[BASE:%.*]])
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = extractvalue { <vscale x 8 x bfloat>, <vscale x 8 x bfloat> } [[TMP0]], 0
+// CPP-CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 16 x bfloat> @llvm.vector.insert.nxv16bf16.nxv8bf16(<vscale x 16 x bfloat> poison, <vscale x 8 x bfloat> [[TMP1]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP3:%.*]] = extractvalue { <vscale x 8 x bfloat>, <vscale x 8 x bfloat> } [[TMP0]], 1
+// CPP-CHECK-NEXT:    [[TMP4:%.*]] = tail call <vscale x 16 x bfloat> @llvm.vector.insert.nxv16bf16.nxv8bf16(<vscale x 16 x bfloat> [[TMP2]], <vscale x 8 x bfloat> [[TMP3]], i64 8)
+// CPP-CHECK-NEXT:    ret <vscale x 16 x bfloat> [[TMP4]]
+//
+svbfloat16x2_t test_svread_hor_za16_bf16_vg2(uint32_t base) {
+  return svread_hor_za16_bf16_vg2(1, base);
+}
+
+// CHECK-LABEL: @test_svread_hor_za16_f16_vg2(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call { <vscale x 8 x half>, <vscale x 8 x half> } @llvm.aarch64.sme.read.hor.vg2.nxv8f16(i32 1, i32 [[BASE:%.*]])
+// CHECK-NEXT:    [[TMP1:%.*]] = extractvalue { <vscale x 8 x half>, <vscale x 8 x half> } [[TMP0]], 0
+// CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 16 x half> @llvm.vector.insert.nxv16f16.nxv8f16(<vscale x 16 x half> poison, <vscale x 8 x half> [[TMP1]], i64 0)
+// CHECK-NEXT:    [[TMP3:%.*]] = extractvalue { <vscale x 8 x half>, <vscale x 8 x half> } [[TMP0]], 1
+// CHECK-NEXT:    [[TMP4:%.*]] = tail call <vscale x 16 x half> @llvm.vector.insert.nxv16f16.nxv8f16(<vscale x 16 x half> [[TMP2]], <vscale x 8 x half> [[TMP3]], i64 8)
+// CHECK-NEXT:    ret <vscale x 16 x half> [[TMP4]]
+//
+// CPP-CHECK-LABEL: @_Z28test_svread_hor_za16_f16_vg2j(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call { <vscale x 8 x half>, <vscale x 8 x half> } @llvm.aarch64.sme.read.hor.vg2.nxv8f16(i32 1, i32 [[BASE:%.*]])
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = extractvalue { <vscale x 8 x half>, <vscale x 8 x half> } [[TMP0]], 0
+// CPP-CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 16 x half> @llvm.vector.insert.nxv16f16.nxv8f16(<vscale x 16 x half> poison, <vscale x 8 x half> [[TMP1]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP3:%.*]] = extractvalue { <vscale x 8 x half>, <vscale x 8 x half> } [[TMP0]], 1
+// CPP-CHECK-NEXT:    [[TMP4:%.*]] = tail call <vscale x 16 x half> @llvm.vector.insert.nxv16f16.nxv8f16(<vscale x 16 x half> [[TMP2]], <vscale x 8 x half> [[TMP3]], i64 8)
+// CPP-CHECK-NEXT:    ret <vscale x 16 x half> [[TMP4]]
+//
+svfloat16x2_t test_svread_hor_za16_f16_vg2(uint32_t base) {
+  return svread_hor_za16_f16_vg2(1, base);
+}
+
+// CHECK-LABEL: @test_svread_hor_za16_s16_vg2(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call { <vscale x 8 x i16>, <vscale x 8 x i16> } @llvm.aarch64.sme.read.hor.vg2.nxv8i16(i32 1, i32 [[BASE:%.*]])
+// CHECK-NEXT:    [[TMP1:%.*]] = extractvalue { <vscale x 8 x i16>, <vscale x 8 x i16> } [[TMP0]], 0
+// CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 16 x i16> @llvm.vector.insert.nxv16i16.nxv8i16(<vscale x 16 x i16> poison, <vscale x 8 x i16> [[TMP1]], i64 0)
+// CHECK-NEXT:    [[TMP3:%.*]] = extractvalue { <vscale x 8 x i16>, <vscale x 8 x i16> } [[TMP0]], 1
+// CHECK-NEXT:    [[TMP4:%.*]] = tail call <vscale x 16 x i16> @llvm.vector.insert.nxv16i16.nxv8i16(<vscale x 16 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP3]], i64 8)
+// CHECK-NEXT:    ret <vscale x 16 x i16> [[TMP4]]
+//
+// CPP-CHECK-LABEL: @_Z28test_svread_hor_za16_s16_vg2j(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call { <vscale x 8 x i16>, <vscale x 8 x i16> } @llvm.aarch64.sme.read.hor.vg2.nxv8i16(i32 1, i32 [[BASE:%.*]])
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = extractvalue { <vscale x 8 x i16>, <vscale x 8 x i16> } [[TMP0]], 0
+// CPP-CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 16 x i16> @llvm.vector.insert.nxv16i16.nxv8i16(<vscale x 16 x i16> poison, <vscale x 8 x i16> [[TMP1]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP3:%.*]] = extractvalue { <vscale x 8 x i16>, <vscale x 8 x i16> } [[TMP0]], 1
+// CPP-CHECK-NEXT:    [[TMP4:%.*]] = tail call <vscale x 16 x i16> @llvm.vector.insert.nxv16i16.nxv8i16(<vscale x 16 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP3]], i64 8)
+// CPP-CHECK-NEXT:    ret <vscale x 16 x i16> [[TMP4]]
+//
+svint16x2_t test_svread_hor_za16_s16_vg2(uint32_t base) {
+  return svread_hor_za16_s16_vg2(1, base);
+}
+
+// CHECK-LABEL: @test_svread_ver_za16_u16_vg2(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call { <vscale x 8 x i16>, <vscale x 8 x i16> } @llvm.aarch64.sme.read.ver.vg2.nxv8i16(i32 1, i32 [[BASE:%.*]])
+// CHECK-NEXT:    [[TMP1:%.*]] = extractvalue { <vscale x 8 x i16>, <vscale x 8 x i16> } [[TMP0]], 0
+// CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 16 x i16> @llvm.vector.insert.nxv16i16.nxv8i16(<vscale x 16 x i16> poison, <vscale x 8 x i16> [[TMP1]], i64 0)
+// CHECK-NEXT:    [[TMP3:%.*]] = extractvalue { <vscale x 8 x i16>, <vscale x 8 x i16> } [[TMP0]], 1
+// CHECK-NEXT:    [[TMP4:%.*]] = tail call <vscale x 16 x i16> @llvm.vector.insert.nxv16i16.nxv8i16(<vscale x 16 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP3]], i64 8)
+// CHECK-NEXT:    ret <vscale x 16 x i16> [[TMP4]]
+//
+// CPP-CHECK-LABEL: @_Z28test_svread_ver_za16_u16_vg2j(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call { <vscale x 8 x i16>, <vscale x 8 x i16> } @llvm.aarch64.sme.read.ver.vg2.nxv8i16(i32 1, i32 [[BASE:%.*]])
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = extractvalue { <vscale x 8 x i16>, <vscale x 8 x i16> } [[TMP0]], 0
+// CPP-CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 16 x i16> @llvm.vector.insert.nxv16i16.nxv8i16(<vscale x 16 x i16> poison, <vscale x 8 x i16> [[TMP1]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP3:%.*]] = extractvalue { <vscale x 8 x i16>, <vscale x 8 x i16> } [[TMP0]], 1
+// CPP-CHECK-NEXT:    [[TMP4:%.*]] = tail call <vscale x 16 x i16> @llvm.vector.insert.nxv16i16.nxv8i16(<vscale x 16 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP3]], i64 8)
+// CPP-CHECK-NEXT:    ret <vscale x 16 x i16> [[TMP4]]
+//
+svuint16x2_t test_svread_ver_za16_u16_vg2(uint32_t base) {
+  return svread_ver_za16_u16_vg2(1, base);
+}
+
+// CHECK-LABEL: @test_svread_ver_za16_bf16_vg2(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call { <vscale x 8 x bfloat>, <vscale x 8 x bfloat> } @llvm.aarch64.sme.read.ver.vg2.nxv8bf16(i32 1, i32 [[BASE:%.*]])
+// CHECK-NEXT:    [[TMP1:%.*]] = extractvalue { <vscale x 8 x bfloat>, <vscale x 8 x bfloat> } [[TMP0]], 0
+// CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 16 x bfloat> @llvm.vector.insert.nxv16bf16.nxv8bf16(<vscale x 16 x bfloat> poison, <vscale x 8 x bfloat> [[TMP1]], i64 0)
+// CHECK-NEXT:    [[TMP3:%.*]] = extractvalue { <vscale x 8 x bfloat>, <vscale x 8 x bfloat> } [[TMP0]], 1
+// CHECK-NEXT:    [[TMP4:%.*]] = tail call <vscale x 16 x bfloat> @llvm.vector.insert.nxv16bf16.nxv8bf16(<vscale x 16 x bfloat> [[TMP2]], <vscale x 8 x bfloat> [[TMP3]], i64 8)
+// CHECK-NEXT:    ret <vscale x 16 x bfloat> [[TMP4]]
+//
+// CPP-CHECK-LABEL: @_Z29test_svread_ver_za16_bf16_vg2j(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call { <vscale x 8 x bfloat>, <vscale x 8 x bfloat> } @llvm.aarch64.sme.read.ver.vg2.nxv8bf16(i32 1, i32 [[BASE:%.*]])
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = extractvalue { <vscale x 8 x bfloat>, <vscale x 8 x bfloat> } [[TMP0]], 0
+// CPP-CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 16 x bfloat> @llvm.vector.insert.nxv16bf16.nxv8bf16(<vscale x 16 x bfloat> poison, <vscale x 8 x bfloat> [[TMP1]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP3:%.*]] = extractvalue { <vscale x 8 x bfloat>, <vscale x 8 x bfloat> } [[TMP0]], 1
+// CPP-CHECK-NEXT:    [[TMP4:%.*]] = tail call <vscale x 16 x bfloat> @llvm.vector.insert.nxv16bf16.nxv8bf16(<vscale x 16 x bfloat> [[TMP2]], <vscale x 8 x bfloat> [[TMP3]], i64 8)
+// CPP-CHECK-NEXT:    ret <vscale x 16 x bfloat> [[TMP4]]
+//
+svbfloat16x2_t test_svread_ver_za16_bf16_vg2(uint32_t base) {
+  return svread_ver_za16_bf16_vg2(1, base);
+}
+
+// CHECK-LABEL: @test_svread_ver_za16_f16_vg2(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call { <vscale x 8 x half>, <vscale x 8 x half> } @llvm.aarch64.sme.read.ver.vg2.nxv8f16(i32 1, i32 [[BASE:%.*]])
+// CHECK-NEXT:    [[TMP1:%.*]] = extractvalue { <vscale x 8 x half>, <vscale x 8 x half> } [[TMP0]], 0
+// CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 16 x half> @llvm.vector.insert.nxv16f16.nxv8f16(<vscale x 16 x half> poison, <vscale x 8 x half> [[TMP1]], i64 0)
+// CHECK-NEXT:    [[TMP3:%.*]] = extractvalue { <vscale x 8 x half>, <vscale x 8 x half> } [[TMP0]], 1
+// CHECK-NEXT:    [[TMP4:%.*]] = tail call <vscale x 16 x half> @llvm.vector.insert.nxv16f16.nxv8f16(<vscale x 16 x half> [[TMP2]], <vscale x 8 x half> [[TMP3]], i64 8)
+// CHECK-NEXT:    ret <vscale x 16 x half> [[TMP4]]
+//
+// CPP-CHECK-LABEL: @_Z28test_svread_ver_za16_f16_vg2j(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call { <vscale x 8 x half>, <vscale x 8 x half> } @llvm.aarch64.sme.read.ver.vg2.nxv8f16(i32 1, i32 [[BASE:%.*]])
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = extractvalue { <vscale x 8 x half>, <vscale x 8 x half> } [[TMP0]], 0
+// CPP-CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 16 x half> @llvm.vector.insert.nxv16f16.nxv8f16(<vscale x 16 x half> poison, <vscale x 8 x half> [[TMP1]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP3:%.*]] = extractvalue { <vscale x 8 x half>, <vscale x 8 x half> } [[TMP0]], 1
+// CPP-CHECK-NEXT:    [[TMP4:%.*]] = tail call <vscale x 16 x half> @llvm.vector.insert.nxv16f16.nxv8f16(<vscale x 16 x half> [[TMP2]], <vscale x 8 x half> [[TMP3]], i64 8)
+// CPP-CHECK-NEXT:    ret <vscale x 16 x half> [[TMP4]]
+//
+svfloat16x2_t test_svread_ver_za16_f16_vg2(uint32_t base) {
+  return svread_ver_za16_f16_vg2(1, base);
+}
+
+// CHECK-LABEL: @test_svread_ver_za16_s16_vg2(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call { <vscale x 8 x i16>, <vscale x 8 x i16> } @llvm.aarch64.sme.read.ver.vg2.nxv8i16(i32 1, i32 [[BASE:%.*]])
+// CHECK-NEXT:    [[TMP1:%.*]] = extractvalue { <vscale x 8 x i16>, <vscale x 8 x i16> } [[TMP0]], 0
+// CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 16 x i16> @llvm.vector.insert.nxv16i16.nxv8i16(<vscale x 16 x i16> poison, <vscale x 8 x i16> [[TMP1]], i64 0)
+// CHECK-NEXT:    [[TMP3:%.*]] = extractvalue { <vscale x 8 x i16>, <vscale x 8 x i16> } [[TMP0]], 1
+// CHECK-NEXT:    [[TMP4:%.*]] = tail call <vscale x 16 x i16> @llvm.vector.insert.nxv16i16.nxv8i16(<vscale x 16 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP3]], i64 8)
+// CHECK-NEXT:    ret <vscale x 16 x i16> [[TMP4]]
+//
+// CPP-CHECK-LABEL: @_Z28test_svread_ver_za16_s16_vg2j(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call { <vscale x 8 x i16>, <vscale x 8 x i16> } @llvm.aarch64.sme.read.ver.vg2.nxv8i16(i32 1, i32 [[BASE:%.*]])
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = extractvalue { <vscale x 8 x i16>, <vscale x 8 x i16> } [[TMP0]], 0
+// CPP-CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 16 x i16> @llvm.vector.insert.nxv16i16.nxv8i16(<vscale x 16 x i16> poison, <vscale x 8 x i16> [[TMP1]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP3:%.*]] = extractvalue { <vscale x 8 x i16>, <vscale x 8 x i16> } [[TMP0]], 1
+// CPP-CHECK-NEXT:    [[TMP4:%.*]] = tail call <vscale x 16 x i16> @llvm.vector.insert.nxv16i16.nxv8i16(<vscale x 16 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP3]], i64 8)
+// CPP-CHECK-NEXT:    ret <vscale x 16 x i16> [[TMP4]]
+//
+svint16x2_t test_svread_ver_za16_s16_vg2(uint32_t base) {
+  return svread_ver_za16_s16_vg2(1, base);
+}
+
+// CHECK-LABEL: @test_svread_hor_za16_u16_vg4(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call { <vscale x 8 x i16>, <vscale x 8 x i16>, <vscale x 8 x i16>, <vscale x 8 x i16> } @llvm.aarch64.sme.read.hor.vg4.nxv8i16(i32 1, i32 [[BASE:%.*]])
+// CHECK-NEXT:    [[TMP1:%.*]] = extractvalue { <vscale x 8 x i16>, <vscale x 8 x i16>, <vscale x 8 x i16>, <vscale x 8 x i16> } [[TMP0]], 0
+// CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 32 x i16> @llvm.vector.insert.nxv32i16.nxv8i16(<vscale x 32 x i16> poison, <vscale x 8 x i16> [[TMP1]], i64 0)
+// CHECK-NEXT:    [[TMP3:%.*]] = extractvalue { <vscale x 8 x i16>, <vscale x 8 x i16>, <vscale x 8 x i16>, <vscale x 8 x i16> } [[TMP0]], 1
+// CHECK-NEXT:    [[TMP4:%.*]] = tail call <vscale x 32 x i16> @llvm.vector.insert.nxv32i16.nxv8i16(<vscale x 32 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP3]], i64 8)
+// CHECK-NEXT:    [[TMP5:%.*]] = extractvalue { <vscale x 8 x i16>, <vscale x 8 x i16>, <vscale x 8 x i16>, <vscale x 8 x i16> } [[TMP0]], 2
+// CHECK-NEXT:    [[TMP6:%.*]] = tail call <vscale x 32 x i16> @llvm.vector.insert.nxv32i16.nxv8i16(<vscale x 32 x i16> [[TMP4]], <vscale x 8 x i16> [[TMP5]], i64 16)
+// CHECK-NEXT:    [[TMP7:%.*]] = extractvalue { <vscale x 8 x i16>, <vscale x 8 x i16>, <vscale x 8 x i16>, <vscale x 8 x i16> } [[TMP0]], 3
+// CHECK-NEXT:    [[TMP8:%.*]] = tail call <vscale x 32 x i16> @llvm.vector.insert.nxv32i16.nxv8i16(<vscale x 32 x i16> [[TMP6]], <vscale x 8 x i16> [[TMP7]], i64 24)
+// CHECK-NEXT:    ret <vscale x 32 x i16> [[TMP8]]
+//
+// CPP-CHECK-LABEL: @_Z28test_svread_hor_za16_u16_vg4j(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call { <vscale x 8 x i16>, <vscale x 8 x i16>, <vscale x 8 x i16>, <vscale x 8 x i16> } @llvm.aarch64.sme.read.hor.vg4.nxv8i16(i32 1, i32 [[BASE:%.*]])
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = extractvalue { <vscale x 8 x i16>, <vscale x 8 x i16>, <vscale x 8 x i16>, <vscale x 8 x i16> } [[TMP0]], 0
+// CPP-CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 32 x i16> @llvm.vector.insert.nxv32i16.nxv8i16(<vscale x 32 x i16> poison, <vscale x 8 x i16> [[TMP1]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP3:%.*]] = extractvalue { <vscale x 8 x i16>, <vscale x 8 x i16>, <vscale x 8 x i16>, <vscale x 8 x i16> } [[TMP0]], 1
+// CPP-CHECK-NEXT:    [[TMP4:%.*]] = tail call <vscale x 32 x i16> @llvm.vector.insert.nxv32i16.nxv8i16(<vscale x 32 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP3]], i64 8)
+// CPP-CHECK-NEXT:    [[TMP5:%.*]] = extractvalue { <vscale x 8 x i16>, <vscale x 8 x i16>, <vscale x 8 x i16>, <vscale x 8 x i16> } [[TMP0]], 2
+// CPP-CHECK-NEXT:    [[TMP6:%.*]] = tail call <vscale x 32 x i16> @llvm.vector.insert.nxv32i16.nxv8i16(<vscale x 32 x i16> [[TMP4]], <vscale x 8 x i16> [[TMP5]], i64 16)
+// CPP-CHECK-NEXT:    [[TMP7:%.*]] = extractvalue { <vscale x 8 x i16>, <vscale x 8 x i16>, <vscale x 8 x i16>, <vscale x 8 x i16> } [[TMP0]], 3
+// CPP-CHECK-NEXT:    [[TMP8:%.*]] = tail call <vscale x 32 x i16> @llvm.vector.insert.nxv32i16.nxv8i16(<vscale x 32 x i16> [[TMP6]], <vscale x 8 x i16> [[TMP7]], i64 24)
+// CPP-CHECK-NEXT:    ret <vscale x 32 x i16> [[TMP8]]
+//
+svuint16x4_t test_svread_hor_za16_u16_vg4(uint32_t base) {
+  return svread_hor_za16_u16_vg4(1, base);
+}
+
+// CHECK-LABEL: @test_svread_hor_za16_bf16_vg4(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call { <vscale x 8 x bfloat>, <vscale x 8 x bfloat>, <vscale x 8 x bfloat>, <vscale x 8 x bfloat> } @llvm.aarch64.sme.read.hor.vg4.nxv8bf16(i32 1, i32 [[BASE:%.*]])
+// CHECK-NEXT:    [[TMP1:%.*]] = extractvalue { <vscale x 8 x bfloat>, <vscale x 8 x bfloat>, <vscale x 8 x bfloat>, <vscale x 8 x bfloat> } [[TMP0]], 0
+// CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 32 x bfloat> @llvm.vector.insert.nxv32bf16.nxv8bf16(<vscale x 32 x bfloat> poison, <vscale x 8 x bfloat> [[TMP1]], i64 0)
+// CHECK-NEXT:    [[TMP3:%.*]] = extractvalue { <vscale x 8 x bfloat>, <vscale x 8 x bfloat>, <vscale x 8 x bfloat>, <vscale x 8 x bfloat> } [[TMP0]], 1
+// CHECK-NEXT:    [[TMP4:%.*]] = tail call <vscale x 32 x bfloat> @llvm.vector.insert.nxv32bf16.nxv8bf16(<vscale x 32 x bfloat> [[TMP2]], <vscale x 8 x bfloat> [[TMP3]], i64 8)
+// CHECK-NEXT:    [[TMP5:%.*]] = extractvalue { <vscale x 8 x bfloat>, <vscale x 8 x bfloat>, <vscale x 8 x bfloat>, <vscale x 8 x bfloat> } [[TMP0]], 2
+// CHECK-NEXT:    [[TMP6:%.*]] = tail call <vscale x 32 x bfloat> @llvm.vector.insert.nxv32bf16.nxv8bf16(<vscale x 32 x bfloat> [[TMP4]], <vscale x 8 x bfloat> [[TMP5]], i64 16)
+// CHECK-NEXT:    [[TMP7:%.*]] = extractvalue { <vscale x 8 x bfloat>, <vscale x 8 x bfloat>, <vscale x 8 x bfloat>, <vscale x 8 x bfloat> } [[TMP0]], 3
+// CHECK-NEXT:    [[TMP8:%.*]] = tail call <vscale x 32 x bfloat> @llvm.vector.insert.nxv32bf16.nxv8bf16(<vscale x 32 x bfloat> [[TMP6]], <vscale x 8 x bfloat> [[TMP7]], i64 24)
+// CHECK-NEXT:    ret <vscale x 32 x bfloat> [[TMP8]]
+//
+// CPP-CHECK-LABEL: @_Z29test_svread_hor_za16_bf16_vg4j(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call { <vscale x 8 x bfloat>, <vscale x 8 x bfloat>, <vscale x 8 x bfloat>, <vscale x 8 x bfloat> } @llvm.aarch64.sme.read.hor.vg4.nxv8bf16(i32 1, i32 [[BASE:%.*]])
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = extractvalue { <vscale x 8 x bfloat>, <vscale x 8 x bfloat>, <vscale x 8 x bfloat>, <vscale x 8 x bfloat> } [[TMP0]], 0
+// CPP-CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 32 x bfloat> @llvm.vector.insert.nxv32bf16.nxv8bf16(<vscale x 32 x bfloat> poison, <vscale x 8 x bfloat> [[TMP1]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP3:%.*]] = extractvalue { <vscale x 8 x bfloat>, <vscale x 8 x bfloat>, <vscale x 8 x bfloat>, <vscale x 8 x bfloat> } [[TMP0]], 1
+// CPP-CHECK-NEXT:    [[TMP4:%.*]] = tail call <vscale x 32 x bfloat> @llvm.vector.insert.nxv32bf16.nxv8bf16(<vscale x 32 x bfloat> [[TMP2]], <vscale x 8 x bfloat> [[TMP3]], i64 8)
+// CPP-CHECK-NEXT:    [[TMP5:%.*]] = extractvalue { <vscale x 8 x bfloat>, <vscale x 8 x bfloat>, <vscale x 8 x bfloat>, <vscale x 8 x bfloat> } [[TMP0]], 2
+// CPP-CHECK-NEXT:    [[TMP6:%.*]] = tail call <vscale x 32 x bfloat> @llvm.vector.insert.nxv32bf16.nxv8bf16(<vscale x 32 x bfloat> [[TMP4]], <vscale x 8 x bfloat> [[TMP5]], i64 16)
+// CPP-CHECK-NEXT:    [[TMP7:%.*]] = extractvalue { <vscale x 8 x bfloat>, <vscale x 8 x bfloat>, <vscale x 8 x bfloat>, <vscale x 8 x bfloat> } [[TMP0]], 3
+// CPP-CHECK-NEXT:    [[TMP8:%.*]] = tail call <vscale x 32 x bfloat> @llvm.vector.insert.nxv32bf16.nxv8bf16(<vscale x 32 x bfloat> [[TMP6]], <vscale x 8 x bfloat> [[TMP7]], i64 24)
+// CPP-CHECK-NEXT:    ret <vscale x 32 x bfloat> [[TMP8]]
+//
+svbfloat16x4_t test_svread_hor_za16_bf16_vg4(uint32_t base) {
+  return svread_hor_za16_bf16_vg4(1, base);
+}
+
+// CHECK-LABEL: @test_svread_hor_za16_f16_vg4(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call { <vscale x 8 x half>, <vscale x 8 x half>, <vscale x 8 x half>, <vscale x 8 x half> } @llvm.aarch64.sme.read.hor.vg4.nxv8f16(i32 1, i32 [[BASE:%.*]])
+// CHECK-NEXT:    [[TMP1:%.*]] = extractvalue { <vscale x 8 x half>, <vscale x 8 x half>, <vscale x 8 x half>, <vscale x 8 x half> } [[TMP0]], 0
+// CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 32 x half> @llvm.vector.insert.nxv32f16.nxv8f16(<vscale x 32 x half> poison, <vscale x 8 x half> [[TMP1]], i64 0)
+// CHECK-NEXT:    [[TMP3:%.*]] = extractvalue { <vscale x 8 x half>, <vscale x 8 x half>, <vscale x 8 x half>, <vscale x 8 x half> } [[TMP0]], 1
+// CHECK-NEXT:    [[TMP4:%.*]] = tail call <vscale x 32 x half> @llvm.vector.insert.nxv32f16.nxv8f16(<vscale x 32 x half> [[TMP2]], <vscale x 8 x half> [[TMP3]], i64 8)
+// CHECK-NEXT:    [[TMP5:%.*]] = extractvalue { <vscale x 8 x half>, <vscale x 8 x half>, <vscale x 8 x half>, <vscale x 8 x half> } [[TMP0]], 2
+// CHECK-NEXT:    [[TMP6:%.*]] = tail call <vscale x 32 x half> @llvm.vector.insert.nxv32f16.nxv8f16(<vscale x 32 x half> [[TMP4]], <vscale x 8 x half> [[TMP5]], i64 16)
+// CHECK-NEXT:    [[TMP7:%.*]] = extractvalue { <vscale x 8 x half>, <vscale x 8 x half>, <vscale x 8 x half>, <vscale x 8 x half> } [[TMP0]], 3
+// CHECK-NEXT:    [[TMP8:%.*]] = tail call <vscale x 32 x half> @llvm.vector.insert.nxv32f16.nxv8f16(<vscale x 32 x half> [[TMP6]], <vscale x 8 x half> [[TMP7]], i64 24)
+// CHECK-NEXT:    ret <vscale x 32 x half> [[TMP8]]
+//
+// CPP-CHECK-LABEL: @_Z28test_svread_hor_za16_f16_vg4j(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call { <vscale x 8 x half>, <vscale x 8 x half>, <vscale x 8 x half>, <vscale x 8 x half> } @llvm.aarch64.sme.read.hor.vg4.nxv8f16(i32 1, i32 [[BASE:%.*]])
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = extractvalue { <vscale x 8 x half>, <vscale x 8 x half>, <vscale x 8 x half>, <vscale x 8 x half> } [[TMP0]], 0
+// CPP-CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 32 x half> @llvm.vector.insert.nxv32f16.nxv8f16(<vscale x 32 x half> poison, <vscale x 8 x half> [[TMP1]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP3:%.*]] = extractvalue { <vscale x 8 x half>, <vscale x 8 x half>, <vscale x 8 x half>, <vscale x 8 x half> } [[TMP0]], 1
+// CPP-CHECK-NEXT:    [[TMP4:%.*]] = tail call <vscale x 32 x half> @llvm.vector.insert.nxv32f16.nxv8f16(<vscale x 32 x half> [[TMP2]], <vscale x 8 x half> [[TMP3]], i64 8)
+// CPP-CHECK-NEXT:    [[TMP5:%.*]] = extractvalue { <vscale x 8 x half>, <vscale x 8 x half>, <vscale x 8 x half>, <vscale x 8 x half> } [[TMP0]], 2
+// CPP-CHECK-NEXT:    [[TMP6:%.*]] = tail call <vscale x 32 x half> @llvm.vector.insert.nxv32f16.nxv8f16(<vscale x 32 x half> [[TMP4]], <vscale x 8 x half> [[TMP5]], i64 16)
+// CPP-CHECK-NEXT:    [[TMP7:%.*]] = extractvalue { <vscale x 8 x half>, <vscale x 8 x half>, <vscale x 8 x half>, <vscale x 8 x half> } [[TMP0]], 3
+// CPP-CHECK-NEXT:    [[TMP8:%.*]] = tail call <vscale x 32 x half> @llvm.vector.insert.nxv32f16.nxv8f16(<vscale x 32 x half> [[TMP6]], <vscale x 8 x half> [[TMP7]], i64 24)
+// CPP-CHECK-NEXT:    ret <vscale x 32 x half> [[TMP8]]
+//
+svfloat16x4_t test_svread_hor_za16_f16_vg4(uint32_t base) {
+  return svread_hor_za16_f16_vg4(1, base);
+}
+
+// CHECK-LABEL: @test_svread_hor_za16_s16_vg4(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call { <vscale x 8 x i16>, <vscale x 8 x i16>, <vscale x 8 x i16>, <vscale x 8 x i16> } @llvm.aarch64.sme.read.hor.vg4.nxv8i16(i32 1, i32 [[BASE:%.*]])
+// CHECK-NEXT:    [[TMP1:%.*]] = extractvalue { <vscale x 8 x i16>, <vscale x 8 x i16>, <vscale x 8 x i16>, <vscale x 8 x i16> } [[TMP0]], 0
+// CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 32 x i16> @llvm.vector.insert.nxv32i16.nxv8i16(<vscale x 32 x i16> poison, <vscale x 8 x i16> [[TMP1]], i64 0)
+// CHECK-NEXT:    [[TMP3:%.*]] = extractvalue { <vscale x 8 x i16>, <vscale x 8 x i16>, <vscale x 8 x i16>, <vscale x 8 x i16> } [[TMP0]], 1
+// CHECK-NEXT:    [[TMP4:%.*]] = tail call <vscale x 32 x i16> @llvm.vector.insert.nxv32i16.nxv8i16(<vscale x 32 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP3]], i64 8)
+// CHECK-NEXT:    [[TMP5:%.*]] = extractvalue { <vscale x 8 x i16>, <vscale x 8 x i16>, <vscale x 8 x i16>, <vscale x 8 x i16> } [[TMP0]], 2
+// CHECK-NEXT:    [[TMP6:%.*]] = tail call <vscale x 32 x i16> @llvm.vector.insert.nxv32i16.nxv8i16(<vscale x 32 x i16> [[TMP4]], <vscale x 8 x i16> [[TMP5]], i64 16)
+// CHECK-NEXT:    [[TMP7:%.*]] = extractvalue { <vscale x 8 x i16>, <vscale x 8 x i16>, <vscale x 8 x i16>, <vscale x 8 x i16> } [[TMP0]], 3
+// CHECK-NEXT:    [[TMP8:%.*]] = tail call <vscale x 32 x i16> @llvm.vector.insert.nxv32i16.nxv8i16(<vscale x 32 x i16> [[TMP6]], <vscale x 8 x i16> [[TMP7]], i64 24)
+// CHECK-NEXT:    ret <vscale x 32 x i16> [[TMP8]]
+//
+// CPP-CHECK-LABEL: @_Z28test_svread_hor_za16_s16_vg4j(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call { <vscale x 8 x i16>, <vscale x 8 x i16>, <vscale x 8 x i16>, <vscale x 8 x i16> } @llvm.aarch64.sme.read.hor.vg4.nxv8i16(i32 1, i32 [[BASE:%.*]])
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = extractvalue { <vscale x 8 x i16>, <vscale x 8 x i16>, <vscale x 8 x i16>, <vscale x 8 x i16> } [[TMP0]], 0
+// CPP-CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 32 x i16> @llvm.vector.insert.nxv32i16.nxv8i16(<vscale x 32 x i16> poison, <vscale x 8 x i16> [[TMP1]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP3:%.*]] = extractvalue { <vscale x 8 x i16>, <vscale x 8 x i16>, <vscale x 8 x i16>, <vscale x 8 x i16> } [[TMP0]], 1
+// CPP-CHECK-NEXT:    [[TMP4:%.*]] = tail call <vscale x 32 x i16> @llvm.vector.insert.nxv32i16.nxv8i16(<vscale x 32 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP3]], i64 8)
+// CPP-CHECK-NEXT:    [[TMP5:%.*]] = extractvalue { <vscale x 8 x i16>, <vscale x 8 x i16>, <vscale x 8 x i16>, <vscale x 8 x i16> } [[TMP0]], 2
+// CPP-CHECK-NEXT:    [[TMP6:%.*]] = tail call <vscale x 32 x i16> @llvm.vector.insert.nxv32i16.nxv8i16(<vscale x 32 x i16> [[TMP4]], <vscale x 8 x i16> [[TMP5]], i64 16)
+// CPP-CHECK-NEXT:    [[TMP7:%.*]] = extractvalue { <vscale x 8 x i16>, <vscale x 8 x i16>, <vscale x 8 x i16>, <vscale x 8 x i16> } [[TMP0]], 3
+// CPP-CHECK-NEXT:    [[TMP8:%.*]] = tail call <vscale x 32 x i16> @llvm.vector.insert.nxv32i16.nxv8i16(<vscale x 32 x i16> [[TMP6]], <vscale x 8 x i16> [[TMP7]], i64 24)
+// CPP-CHECK-NEXT:    ret <vscale x 32 x i16> [[TMP8]]
+//
+svint16x4_t test_svread_hor_za16_s16_vg4(uint32_t base) {
+  return svread_hor_za16_s16_vg4(1, base);
+}
+
+// CHECK-LABEL: @test_svread_ver_za16_u16_vg4(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call { <vscale x 8 x i16>, <vscale x 8 x i16>, <vscale x 8 x i16>, <vscale x 8 x i16> } @llvm.aarch64.sme.read.ver.vg4.nxv8i16(i32 1, i32 [[BASE:%.*]])
+// CHECK-NEXT:    [[TMP1:%.*]] = extractvalue { <vscale x 8 x i16>, <vscale x 8 x i16>, <vscale x 8 x i16>, <vscale x 8 x i16> } [[TMP0]], 0
+// CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 32 x i16> @llvm.vector.insert.nxv32i16.nxv8i16(<vscale x 32 x i16> poison, <vscale x 8 x i16> [[TMP1]], i64 0)
+// CHECK-NEXT:    [[TMP3:%.*]] = extractvalue { <vscale x 8 x i16>, <vscale x 8 x i16>, <vscale x 8 x i16>, <vscale x 8 x i16> } [[TMP0]], 1
+// CHECK-NEXT:    [[TMP4:%.*]] = tail call <vscale x 32 x i16> @llvm.vector.insert.nxv32i16.nxv8i16(<vscale x 32 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP3]], i64 8)
+// CHECK-NEXT:    [[TMP5:%.*]] = extractvalue { <vscale x 8 x i16>, <vscale x 8 x i16>, <vscale x 8 x i16>, <vscale x 8 x i16> } [[TMP0]], 2
+// CHECK-NEXT:    [[TMP6:%.*]] = tail call <vscale x 32 x i16> @llvm.vector.insert.nxv32i16.nxv8i16(<vscale x 32 x i16> [[TMP4]], <vscale x 8 x i16> [[TMP5]], i64 16)
+// CHECK-NEXT:    [[TMP7:%.*]] = extractvalue { <vscale x 8 x i16>, <vscale x 8 x i16>, <vscale x 8 x i16>, <vscale x 8 x i16> } [[TMP0]], 3
+// CHECK-NEXT:    [[TMP8:%.*]] = tail call <vscale x 32 x i16> @llvm.vector.insert.nxv32i16.nxv8i16(<vscale x 32 x i16> [[TMP6]], <vscale x 8 x i16> [[TMP7]], i64 24)
+// CHECK-NEXT:    ret <vscale x 32 x i16> [[TMP8]]
+//
+// CPP-CHECK-LABEL: @_Z28test_svread_ver_za16_u16_vg4j(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call { <vscale x 8 x i16>, <vscale x 8 x i16>, <vscale x 8 x i16>, <vscale x 8 x i16> } @llvm.aarch64.sme.read.ver.vg4.nxv8i16(i32 1, i32 [[BASE:%.*]])
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = extractvalue { <vscale x 8 x i16>, <vscale x 8 x i16>, <vscale x 8 x i16>, <vscale x 8 x i16> } [[TMP0]], 0
+// CPP-CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 32 x i16> @llvm.vector.insert.nxv32i16.nxv8i16(<vscale x 32 x i16> poison, <vscale x 8 x i16> [[TMP1]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP3:%.*]] = extractvalue { <vscale x 8 x i16>, <vscale x 8 x i16>, <vscale x 8 x i16>, <vscale x 8 x i16> } [[TMP0]], 1
+// CPP-CHECK-NEXT:    [[TMP4:%.*]] = tail call <vscale x 32 x i16> @llvm.vector.insert.nxv32i16.nxv8i16(<vscale x 32 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP3]], i64 8)
+// CPP-CHECK-NEXT:    [[TMP5:%.*]] = extractvalue { <vscale x 8 x i16>, <vscale x 8 x i16>, <vscale x 8 x i16>, <vscale x 8 x i16> } [[TMP0]], 2
+// CPP-CHECK-NEXT:    [[TMP6:%.*]] = tail call <vscale x 32 x i16> @llvm.vector.insert.nxv32i16.nxv8i16(<vscale x 32 x i16> [[TMP4]], <vscale x 8 x i16> [[TMP5]], i64 16)
+// CPP-CHECK-NEXT:    [[TMP7:%.*]] = extractvalue { <vscale x 8 x i16>, <vscale x 8 x i16>, <vscale x 8 x i16>, <vscale x 8 x i16> } [[TMP0]], 3
+// CPP-CHECK-NEXT:    [[TMP8:%.*]] = tail call <vscale x 32 x i16> @llvm.vector.insert.nxv32i16.nxv8i16(<vscale x 32 x i16> [[TMP6]], <vscale x 8 x i16> [[TMP7]], i64 24)
+// CPP-CHECK-NEXT:    ret <vscale x 32 x i16> [[TMP8]]
+//
+svuint16x4_t test_svread_ver_za16_u16_vg4(uint32_t base) {
+  return svread_ver_za16_u16_vg4(1, base);
+}
+
+// CHECK-LABEL: @test_svread_ver_za16_bf16_vg4(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call { <vscale x 8 x bfloat>, <vscale x 8 x bfloat>, <vscale x 8 x bfloat>, <vscale x 8 x bfloat> } @llvm.aarch64.sme.read.ver.vg4.nxv8bf16(i32 1, i32 [[BASE:%.*]])
+// CHECK-NEXT:    [[TMP1:%.*]] = extractvalue { <vscale x 8 x bfloat>, <vscale x 8 x bfloat>, <vscale x 8 x bfloat>, <vscale x 8 x bfloat> } [[TMP0]], 0
+// CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 32 x bfloat> @llvm.vector.insert.nxv32bf16.nxv8bf16(<vscale x 32 x bfloat> poison, <vscale x 8 x bfloat> [[TMP1]], i64 0)
+// CHECK-NEXT:    [[TMP3:%.*]] = extractvalue { <vscale x 8 x bfloat>, <vscale x 8 x bfloat>, <vscale x 8 x bfloat>, <vscale x 8 x bfloat> } [[TMP0]], 1
+// CHECK-NEXT:    [[TMP4:%.*]] = tail call <vscale x 32 x bfloat> @llvm.vector.insert.nxv32bf16.nxv8bf16(<vscale x 32 x bfloat> [[TMP2]], <vscale x 8 x bfloat> [[TMP3]], i64 8)
+// CHECK-NEXT:    [[TMP5:%.*]] = extractvalue { <vscale x 8 x bfloat>, <vscale x 8 x bfloat>, <vscale x 8 x bfloat>, <vscale x 8 x bfloat> } [[TMP0]], 2
+// CHECK-NEXT:    [[TMP6:%.*]] = tail call <vscale x 32 x bfloat> @llvm.vector.insert.nxv32bf16.nxv8bf16(<vscale x 32 x bfloat> [[TMP4]], <vscale x 8 x bfloat> [[TMP5]], i64 16)
+// CHECK-NEXT:    [[TMP7:%.*]] = extractvalue { <vscale x 8 x bfloat>, <vscale x 8 x bfloat>, <vscale x 8 x bfloat>, <vscale x 8 x bfloat> } [[TMP0]], 3
+// CHECK-NEXT:    [[TMP8:%.*]] = tail call <vscale x 32 x bfloat> @llvm.vector.insert.nxv32bf16.nxv8bf16(<vscale x 32 x bfloat> [[TMP6]], <vscale x 8 x bfloat> [[TMP7]], i64 24)
+// CHECK-NEXT:    ret <vscale x 32 x bfloat> [[TMP8]]
+//
+// CPP-CHECK-LABEL: @_Z29test_svread_ver_za16_bf16_vg4j(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call { <vscale x 8 x bfloat>, <vscale x 8 x bfloat>, <vscale x 8 x bfloat>, <vscale x 8 x bfloat> } @llvm.aarch64.sme.read.ver.vg4.nxv8bf16(i32 1, i32 [[BASE:%.*]])
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = extractvalue { <vscale x 8 x bfloat>, <vscale x 8 x bfloat>, <vscale x 8 x bfloat>, <vscale x 8 x bfloat> } [[TMP0]], 0
+// CPP-CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 32 x bfloat> @llvm.vector.insert.nxv32bf16.nxv8bf16(<vscale x 32 x bfloat> poison, <vscale x 8 x bfloat> [[TMP1]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP3:%.*]] = extractvalue { <vscale x 8 x bfloat>, <vscale x 8 x bfloat>, <vscale x 8 x bfloat>, <vscale x 8 x bfloat> } [[TMP0]], 1
+// CPP-CHECK-NEXT:    [[TMP4:%.*]] = tail call <vscale x 32 x bfloat> @llvm.vector.insert.nxv32bf16.nxv8bf16(<vscale x 32 x bfloat> [[TMP2]], <vscale x 8 x bfloat> [[TMP3]], i64 8)
+// CPP-CHECK-NEXT:    [[TMP5:%.*]] = extractvalue { <vscale x 8 x bfloat>, <vscale x 8 x bfloat>, <vscale x 8 x bfloat>, <vscale x 8 x bfloat> } [[TMP0]], 2
+// CPP-CHECK-NEXT:    [[TMP6:%.*]] = tail call <vscale x 32 x bfloat> @llvm.vector.insert.nxv32bf16.nxv8bf16(<vscale x 32 x bfloat> [[TMP4]], <vscale x 8 x bfloat> [[TMP5]], i64 16)
+// CPP-CHECK-NEXT:    [[TMP7:%.*]] = extractvalue { <vscale x 8 x bfloat>, <vscale x 8 x bfloat>, <vscale x 8 x bfloat>, <vscale x 8 x bfloat> } [[TMP0]], 3
+// CPP-CHECK-NEXT:    [[TMP8:%.*]] = tail call <vscale x 32 x bfloat> @llvm.vector.insert.nxv32bf16.nxv8bf16(<vscale x 32 x bfloat> [[TMP6]], <vscale x 8 x bfloat> [[TMP7]], i64 24)
+// CPP-CHECK-NEXT:    ret <vscale x 32 x bfloat> [[TMP8]]
+//
+svbfloat16x4_t test_svread_ver_za16_bf16_vg4(uint32_t base) {
+  return svread_ver_za16_bf16_vg4(1, base);
+}
+
+// CHECK-LABEL: @test_svread_ver_za16_f16_vg4(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call { <vscale x 8 x half>, <vscale x 8 x half>, <vscale x 8 x half>, <vscale x 8 x half> } @llvm.aarch64.sme.read.ver.vg4.nxv8f16(i32 1, i32 [[BASE:%.*]])
+// CHECK-NEXT:    [[TMP1:%.*]] = extractvalue { <vscale x 8 x half>, <vscale x 8 x half>, <vscale x 8 x half>, <vscale x 8 x half> } [[TMP0]], 0
+// CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 32 x half> @llvm.vector.insert.nxv32f16.nxv8f16(<vscale x 32 x half> poison, <vscale x 8 x half> [[TMP1]], i64 0)
+// CHECK-NEXT:    [[TMP3:%.*]] = extractvalue { <vscale x 8 x half>, <vscale x 8 x half>, <vscale x 8 x half>, <vscale x 8 x half> } [[TMP0]], 1
+// CHECK-NEXT:    [[TMP4:%.*]] = tail call <vscale x 32 x half> @llvm.vector.insert.nxv32f16.nxv8f16(<vscale x 32 x half> [[TMP2]], <vscale x 8 x half> [[TMP3]], i64 8)
+// CHECK-NEXT:    [[TMP5:%.*]] = extractvalue { <vscale x 8 x half>, <vscale x 8 x half>, <vscale x 8 x half>, <vscale x 8 x half> } [[TMP0]], 2
+// CHECK-NEXT:    [[TMP6:%.*]] = tail call <vscale x 32 x half> @llvm.vector.insert.nxv32f16.nxv8f16(<vscale x 32 x half> [[TMP4]], <vscale x 8 x half> [[TMP5]], i64 16)
+// CHECK-NEXT:    [[TMP7:%.*]] = extractvalue { <vscale x 8 x half>, <vscale x 8 x half>, <vscale x 8 x half>, <vscale x 8 x half> } [[TMP0]], 3
+// CHECK-NEXT:    [[TMP8:%.*]] = tail call <vscale x 32 x half> @llvm.vector.insert.nxv32f16.nxv8f16(<vscale x 32 x half> [[TMP6]], <vscale x 8 x half> [[TMP7]], i64 24)
+// CHECK-NEXT:    ret <vscale x 32 x half> [[TMP8]]
+//
+// CPP-CHECK-LABEL: @_Z28test_svread_ver_za16_f16_vg4j(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call { <vscale x 8 x half>, <vscale x 8 x half>, <vscale x 8 x half>, <vscale x 8 x half> } @llvm.aarch64.sme.read.ver.vg4.nxv8f16(i32 1, i32 [[BASE:%.*]])
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = extractvalue { <vscale x 8 x half>, <vscale x 8 x half>, <vscale x 8 x half>, <vscale x 8 x half> } [[TMP0]], 0
+// CPP-CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 32 x half> @llvm.vector.insert.nxv32f16.nxv8f16(<vscale x 32 x half> poison, <vscale x 8 x half> [[TMP1]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP3:%.*]] = extractvalue { <vscale x 8 x half>, <vscale x 8 x half>, <vscale x 8 x half>, <vscale x 8 x half> } [[TMP0]], 1
+// CPP-CHECK-NEXT:    [[TMP4:%.*]] = tail call <vscale x 32 x half> @llvm.vector.insert.nxv32f16.nxv8f16(<vscale x 32 x half> [[TMP2]], <vscale x 8 x half> [[TMP3]], i64 8)
+// CPP-CHECK-NEXT:    [[TMP5:%.*]] = extractvalue { <vscale x 8 x half>, <vscale x 8 x half>, <vscale x 8 x half>, <vscale x 8 x half> } [[TMP0]], 2
+// CPP-CHECK-NEXT:    [[TMP6:%.*]] = tail call <vscale x 32 x half> @llvm.vector.insert.nxv32f16.nxv8f16(<vscale x 32 x half> [[TMP4]], <vscale x 8 x half> [[TMP5]], i64 16)
+// CPP-CHECK-NEXT:    [[TMP7:%.*]] = extractvalue { <vscale x 8 x half>, <vscale x 8 x half>, <vscale x 8 x half>, <vscale x 8 x half> } [[TMP0]], 3
+// CPP-CHECK-NEXT:    [[TMP8:%.*]] = tail call <vscale x 32 x half> @llvm.vector.insert.nxv32f16.nxv8f16(<vscale x 32 x half> [[TMP6]], <vscale x 8 x half> [[TMP7]], i64 24)
+// CPP-CHECK-NEXT:    ret <vscale x 32 x half> [[TMP8]]
+//
+svfloat16x4_t test_svread_ver_za16_f16_vg4(uint32_t base) {
+  return svread_ver_za16_f16_vg4(1, base);
+}
+
+// CHECK-LABEL: @test_svread_ver_za16_s16_vg4(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call { <vscale x 8 x i16>, <vscale x 8 x i16>, <vscale x 8 x i16>, <vscale x 8 x i16> } @llvm.aarch64.sme.read.ver.vg4.nxv8i16(i32 1, i32 [[BASE:%.*]])
+// CHECK-NEXT:    [[TMP1:%.*]] = extractvalue { <vscale x 8 x i16>, <vscale x 8 x i16>, <vscale x 8 x i16>, <vscale x 8 x i16> } [[TMP0]], 0
+// CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 32 x i16> @llvm.vector.insert.nxv32i16.nxv8i16(<vscale x 32 x i16> poison, <vscale x 8 x i16> [[TMP1]], i64 0)
+// CHECK-NEXT:    [[TMP3:%.*]] = extractvalue { <vscale x 8 x i16>, <vscale x 8 x i16>, <vscale x 8 x i16>, <vscale x 8 x i16> } [[TMP0]], 1
+// CHECK-NEXT:    [[TMP4:%.*]] = tail call <vscale x 32 x i16> @llvm.vector.insert.nxv32i16.nxv8i16(<vscale x 32 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP3]], i64 8)
+// CHECK-NEXT:    [[TMP5:%.*]] = extractvalue { <vscale x 8 x i16>, <vscale x 8 x i16>, <vscale x 8 x i16>, <vscale x 8 x i16> } [[TMP0]], 2
+// CHECK-NEXT:    [[TMP6:%.*]] = tail call <vscale x 32 x i16> @llvm.vector.insert.nxv32i16.nxv8i16(<vscale x 32 x i16> [[TMP4]], <vscale x 8 x i16> [[TMP5]], i64 16)
+// CHECK-NEXT:    [[TMP7:%.*]] = extractvalue { <vscale x 8 x i16>, <vscale x 8 x i16>, <vscale x 8 x i16>, <vscale x 8 x i16> } [[TMP0]], 3
+// CHECK-NEXT:    [[TMP8:%.*]] = tail call <vscale x 32 x i16> @llvm.vector.insert.nxv32i16.nxv8i16(<vscale x 32 x i16> [[TMP6]], <vscale x 8 x i16> [[TMP7]], i64 24)
+// CHECK-NEXT:    ret <vscale x 32 x i16> [[TMP8]]
+//
+// CPP-CHECK-LABEL: @_Z28test_svread_ver_za16_s16_vg4j(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call { <vscale x 8 x i16>, <vscale x 8 x i16>, <vscale x 8 x i16>, <vscale x 8 x i16> } @llvm.aarch64.sme.read.ver.vg4.nxv8i16(i32 1, i32 [[BASE:%.*]])
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = extractvalue { <vscale x 8 x i16>, <vscale x 8 x i16>, <vscale x 8 x i16>, <vscale x 8 x i16> } [[TMP0]], 0
+// CPP-CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 32 x i16> @llvm.vector.insert.nxv32i16.nxv8i16(<vscale x 32 x i16> poison, <vscale x 8 x i16> [[TMP1]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP3:%.*]] = extractvalue { <vscale x 8 x i16>, <vscale x 8 x i16>, <vscale x 8 x i16>, <vscale x 8 x i16> } [[TMP0]], 1
+// CPP-CHECK-NEXT:    [[TMP4:%.*]] = tail call <vscale x 32 x i16> @llvm.vector.insert.nxv32i16.nxv8i16(<vscale x 32 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP3]], i64 8)
+// CPP-CHECK-NEXT:    [[TMP5:%.*]] = extractvalue { <vscale x 8 x i16>, <vscale x 8 x i16>, <vscale x 8 x i16>, <vscale x 8 x i16> } [[TMP0]], 2
+// CPP-CHECK-NEXT:    [[TMP6:%.*]] = tail call <vscale x 32 x i16> @llvm.vector.insert.nxv32i16.nxv8i16(<vscale x 32 x i16> [[TMP4]], <vscale x 8 x i16> [[TMP5]], i64 16)
+// CPP-CHECK-NEXT:    [[TMP7:%.*]] = extractvalue { <vscale x 8 x i16>, <vscale x 8 x i16>, <vscale x 8 x i16>, <vscale x 8 x i16> } [[TMP0]], 3
+// CPP-CHECK-NEXT:    [[TMP8:%.*]] = tail call <vscale x 32 x i16> @llvm.vector.insert.nxv32i16.nxv8i16(<vscale x 32 x i16> [[TMP6]], <vscale x 8 x i16> [[TMP7]], i64 24)
+// CPP-CHECK-NEXT:    ret <vscale x 32 x i16> [[TMP8]]
+//
+svint16x4_t test_svread_ver_za16_s16_vg4(uint32_t base) {
+  return svread_ver_za16_s16_vg4(1, base);
+}
+
+// CHECK-LABEL: @test_svread_hor_za32_u32_vg2(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call { <vscale x 4 x i32>, <vscale x 4 x i32> } @llvm.aarch64.sme.read.hor.vg2.nxv4i32(i32 3, i32 [[BASE:%.*]])
+// CHECK-NEXT:    [[TMP1:%.*]] = extractvalue { <vscale x 4 x i32>, <vscale x 4 x i32> } [[TMP0]], 0
+// CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 8 x i32> @llvm.vector.insert.nxv8i32.nxv4i32(<vscale x 8 x i32> poison, <vscale x 4 x i32> [[TMP1]], i64 0)
+// CHECK-NEXT:    [[TMP3:%.*]] = extractvalue { <vscale x 4 x i32>, <vscale x 4 x i32> } [[TMP0]], 1
+// CHECK-NEXT:    [[TMP4:%.*]] = tail call <vscale x 8 x i32> @llvm.vector.insert.nxv8i32.nxv4i32(<vscale x 8 x i32> [[TMP2]], <vscale x 4 x i32> [[TMP3]], i64 4)
+// CHECK-NEXT:    ret <vscale x 8 x i32> [[TMP4]]
+//
+// CPP-CHECK-LABEL: @_Z28test_svread_hor_za32_u32_vg2j(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call { <vscale x 4 x i32>, <vscale x 4 x i32> } @llvm.aarch64.sme.read.hor.vg2.nxv4i32(i32 3, i32 [[BASE:%.*]])
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = extractvalue { <vscale x 4 x i32>, <vscale x 4 x i32> } [[TMP0]], 0
+// CPP-CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 8 x i32> @llvm.vector.insert.nxv8i32.nxv4i32(<vscale x 8 x i32> poison, <vscale x 4 x i32> [[TMP1]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP3:%.*]] = extractvalue { <vscale x 4 x i32>, <vscale x 4 x i32> } [[TMP0]], 1
+// CPP-CHECK-NEXT:    [[TMP4:%.*]] = tail call <vscale x 8 x i32> @llvm.vector.insert.nxv8i32.nxv4i32(<vscale x 8 x i32> [[TMP2]], <vscale x 4 x i32> [[TMP3]], i64 4)
+// CPP-CHECK-NEXT:    ret <vscale x 8 x i32> [[TMP4]]
+//
+svuint32x2_t test_svread_hor_za32_u32_vg2(uint32_t base) {
+  return svread_hor_za32_u32_vg2(3, base);
+}
+
+// CHECK-LABEL: @test_svread_hor_za32_f32_vg2(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call { <vscale x 4 x float>, <vscale x 4 x float> } @llvm.aarch64.sme.read.hor.vg2.nxv4f32(i32 3, i32 [[BASE:%.*]])
+// CHECK-NEXT:    [[TMP1:%.*]] = extractvalue { <vscale x 4 x float>, <vscale x 4 x float> } [[TMP0]], 0
+// CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 8 x float> @llvm.vector.insert.nxv8f32.nxv4f32(<vscale x 8 x float> poison, <vscale x 4 x float> [[TMP1]], i64 0)
+// CHECK-NEXT:    [[TMP3:%.*]] = extractvalue { <vscale x 4 x float>, <vscale x 4 x float> } [[TMP0]], 1
+// CHECK-NEXT:    [[TMP4:%.*]] = tail call <vscale x 8 x float> @llvm.vector.insert.nxv8f32.nxv4f32(<vscale x 8 x float> [[TMP2]], <vscale x 4 x float> [[TMP3]], i64 4)
+// CHECK-NEXT:    ret <vscale x 8 x float> [[TMP4]]
+//
+// CPP-CHECK-LABEL: @_Z28test_svread_hor_za32_f32_vg2j(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call { <vscale x 4 x float>, <vscale x 4 x float> } @llvm.aarch64.sme.read.hor.vg2.nxv4f32(i32 3, i32 [[BASE:%.*]])
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = extractvalue { <vscale x 4 x float>, <vscale x 4 x float> } [[TMP0]], 0
+// CPP-CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 8 x float> @llvm.vector.insert.nxv8f32.nxv4f32(<vscale x 8 x float> poison, <vscale x 4 x float> [[TMP1]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP3:%.*]] = extractvalue { <vscale x 4 x float>, <vscale x 4 x float> } [[TMP0]], 1
+// CPP-CHECK-NEXT:    [[TMP4:%.*]] = tail call <vscale x 8 x float> @llvm.vector.insert.nxv8f32.nxv4f32(<vscale x 8 x float> [[TMP2]], <vscale x 4 x float> [[TMP3]], i64 4)
+// CPP-CHECK-NEXT:    ret <vscale x 8 x float> [[TMP4]]
+//
+svfloat32x2_t test_svread_hor_za32_f32_vg2(uint32_t base) {
+  return svread_hor_za32_f32_vg2(3, base);
+}
+
+// CHECK-LABEL: @test_svread_hor_za32_s32_vg2(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call { <vscale x 4 x i32>, <vscale x 4 x i32> } @llvm.aarch64.sme.read.hor.vg2.nxv4i32(i32 3, i32 [[BASE:%.*]])
+// CHECK-NEXT:    [[TMP1:%.*]] = extractvalue { <vscale x 4 x i32>, <vscale x 4 x i32> } [[TMP0]], 0
+// CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 8 x i32> @llvm.vector.insert.nxv8i32.nxv4i32(<vscale x 8 x i32> poison, <vscale x 4 x i32> [[TMP1]], i64 0)
+// CHECK-NEXT:    [[TMP3:%.*]] = extractvalue { <vscale x 4 x i32>, <vscale x 4 x i32> } [[TMP0]], 1
+// CHECK-NEXT:    [[TMP4:%.*]] = tail call <vscale x 8 x i32> @llvm.vector.insert.nxv8i32.nxv4i32(<vscale x 8 x i32> [[TMP2]], <vscale x 4 x i32> [[TMP3]], i64 4)
+// CHECK-NEXT:    ret <vscale x 8 x i32> [[TMP4]]
+//
+// CPP-CHECK-LABEL: @_Z28test_svread_hor_za32_s32_vg2j(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call { <vscale x 4 x i32>, <vscale x 4 x i32> } @llvm.aarch64.sme.read.hor.vg2.nxv4i32(i32 3, i32 [[BASE:%.*]])
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = extractvalue { <vscale x 4 x i32>, <vscale x 4 x i32> } [[TMP0]], 0
+// CPP-CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 8 x i32> @llvm.vector.insert.nxv8i32.nxv4i32(<vscale x 8 x i32> poison, <vscale x 4 x i32> [[TMP1]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP3:%.*]] = extractvalue { <vscale x 4 x i32>, <vscale x 4 x i32> } [[TMP0]], 1
+// CPP-CHECK-NEXT:    [[TMP4:%.*]] = tail call <vscale x 8 x i32> @llvm.vector.insert.nxv8i32.nxv4i32(<vscale x 8 x i32> [[TMP2]], <vscale x 4 x i32> [[TMP3]], i64 4)
+// CPP-CHECK-NEXT:    ret <vscale x 8 x i32> [[TMP4]]
+//
+svint32x2_t test_svread_hor_za32_s32_vg2(uint32_t base) {
+  return svread_hor_za32_s32_vg2(3, base);
+}
+
+// CHECK-LABEL: @test_svread_ver_za32_u32_vg2(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call { <vscale x 4 x i32>, <vscale x 4 x i32> } @llvm.aarch64.sme.read.ver.vg2.nxv4i32(i32 3, i32 [[BASE:%.*]])
+// CHECK-NEXT:    [[TMP1:%.*]] = extractvalue { <vscale x 4 x i32>, <vscale x 4 x i32> } [[TMP0]], 0
+// CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 8 x i32> @llvm.vector.insert.nxv8i32.nxv4i32(<vscale x 8 x i32> poison, <vscale x 4 x i32> [[TMP1]], i64 0)
+// CHECK-NEXT:    [[TMP3:%.*]] = extractvalue { <vscale x 4 x i32>, <vscale x 4 x i32> } [[TMP0]], 1
+// CHECK-NEXT:    [[TMP4:%.*]] = tail call <vscale x 8 x i32> @llvm.vector.insert.nxv8i32.nxv4i32(<vscale x 8 x i32> [[TMP2]], <vscale x 4 x i32> [[TMP3]], i64 4)
+// CHECK-NEXT:    ret <vscale x 8 x i32> [[TMP4]]
+//
+// CPP-CHECK-LABEL: @_Z28test_svread_ver_za32_u32_vg2j(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call { <vscale x 4 x i32>, <vscale x 4 x i32> } @llvm.aarch64.sme.read.ver.vg2.nxv4i32(i32 3, i32 [[BASE:%.*]])
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = extractvalue { <vscale x 4 x i32>, <vscale x 4 x i32> } [[TMP0]], 0
+// CPP-CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 8 x i32> @llvm.vector.insert.nxv8i32.nxv4i32(<vscale x 8 x i32> poison, <vscale x 4 x i32> [[TMP1]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP3:%.*]] = extractvalue { <vscale x 4 x i32>, <vscale x 4 x i32> } [[TMP0]], 1
+// CPP-CHECK-NEXT:    [[TMP4:%.*]] = tail call <vscale x 8 x i32> @llvm.vector.insert.nxv8i32.nxv4i32(<vscale x 8 x i32> [[TMP2]], <vscale x 4 x i32> [[TMP3]], i64 4)
+// CPP-CHECK-NEXT:    ret <vscale x 8 x i32> [[TMP4]]
+//
+svuint32x2_t test_svread_ver_za32_u32_vg2(uint32_t base) {
+  return svread_ver_za32_u32_vg2(3, base);
+}
+
+// CHECK-LABEL: @test_svread_ver_za32_f32_vg2(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call { <vscale x 4 x float>, <vscale x 4 x float> } @llvm.aarch64.sme.read.ver.vg2.nxv4f32(i32 3, i32 [[BASE:%.*]])
+// CHECK-NEXT:    [[TMP1:%.*]] = extractvalue { <vscale x 4 x float>, <vscale x 4 x float> } [[TMP0]], 0
+// CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 8 x float> @llvm.vector.insert.nxv8f32.nxv4f32(<vscale x 8 x float> poison, <vscale x 4 x float> [[TMP1]], i64 0)
+// CHECK-NEXT:    [[TMP3:%.*]] = extractvalue { <vscale x 4 x float>, <vscale x 4 x float> } [[TMP0]], 1
+// CHECK-NEXT:    [[TMP4:%.*]] = tail call <vscale x 8 x float> @llvm.vector.insert.nxv8f32.nxv4f32(<vscale x 8 x float> [[TMP2]], <vscale x 4 x float> [[TMP3]], i64 4)
+// CHECK-NEXT:    ret <vscale x 8 x float> [[TMP4]]
+//
+// CPP-CHECK-LABEL: @_Z28test_svread_ver_za32_f32_vg2j(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call { <vscale x 4 x float>, <vscale x 4 x float> } @llvm.aarch64.sme.read.ver.vg2.nxv4f32(i32 3, i32 [[BASE:%.*]])
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = extractvalue { <vscale x 4 x float>, <vscale x 4 x float> } [[TMP0]], 0
+// CPP-CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 8 x float> @llvm.vector.insert.nxv8f32.nxv4f32(<vscale x 8 x float> poison, <vscale x 4 x float> [[TMP1]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP3:%.*]] = extractvalue { <vscale x 4 x float>, <vscale x 4 x float> } [[TMP0]], 1
+// CPP-CHECK-NEXT:    [[TMP4:%.*]] = tail call <vscale x 8 x float> @llvm.vector.insert.nxv8f32.nxv4f32(<vscale x 8 x float> [[TMP2]], <vscale x 4 x float> [[TMP3]], i64 4)
+// CPP-CHECK-NEXT:    ret <vscale x 8 x float> [[TMP4]]
+//
+svfloat32x2_t test_svread_ver_za32_f32_vg2(uint32_t base) {
+  return svread_ver_za32_f32_vg2(3, base);
+}
+
+// CHECK-LABEL: @test_svread_ver_za32_s32_vg2(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call { <vscale x 4 x i32>, <vscale x 4 x i32> } @llvm.aarch64.sme.read.ver.vg2.nxv4i32(i32 3, i32 [[BASE:%.*]])
+// CHECK-NEXT:    [[TMP1:%.*]] = extractvalue { <vscale x 4 x i32>, <vscale x 4 x i32> } [[TMP0]], 0
+// CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 8 x i32> @llvm.vector.insert.nxv8i32.nxv4i32(<vscale x 8 x i32> poison, <vscale x 4 x i32> [[TMP1]], i64 0)
+// CHECK-NEXT:    [[TMP3:%.*]] = extractvalue { <vscale x 4 x i32>, <vscale x 4 x i32> } [[TMP0]], 1
+// CHECK-NEXT:    [[TMP4:%.*]] = tail call <vscale x 8 x i32> @llvm.vector.insert.nxv8i32.nxv4i32(<vscale x 8 x i32> [[TMP2]], <vscale x 4 x i32> [[TMP3]], i64 4)
+// CHECK-NEXT:    ret <vscale x 8 x i32> [[TMP4]]
+//
+// CPP-CHECK-LABEL: @_Z28test_svread_ver_za32_s32_vg2j(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call { <vscale x 4 x i32>, <vscale x 4 x i32> } @llvm.aarch64.sme.read.ver.vg2.nxv4i32(i32 3, i32 [[BASE:%.*]])
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = extractvalue { <vscale x 4 x i32>, <vscale x 4 x i32> } [[TMP0]], 0
+// CPP-CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 8 x i32> @llvm.vector.insert.nxv8i32.nxv4i32(<vscale x 8 x i32> poison, <vscale x 4 x i32> [[TMP1]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP3:%.*]] = extractvalue { <vscale x 4 x i32>, <vscale x 4 x i32> } [[TMP0]], 1
+// CPP-CHECK-NEXT:    [[TMP4:%.*]] = tail call <vscale x 8 x i32> @llvm.vector.insert.nxv8i32.nxv4i32(<vscale x 8 x i32> [[TMP2]], <vscale x 4 x i32> [[TMP3]], i64 4)
+// CPP-CHECK-NEXT:    ret <vscale x 8 x i32> [[TMP4]]
+//
+svint32x2_t test_svread_ver_za32_s32_vg2(uint32_t base) {
+  return svread_ver_za32_s32_vg2(3, base);
+}
+
+// CHECK-LABEL: @test_svread_hor_za32_u32_vg4(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call { <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32> } @llvm.aarch64.sme.read.hor.vg4.nxv4i32(i32 3, i32 [[BASE:%.*]])
+// CHECK-NEXT:    [[TMP1:%.*]] = extractvalue { <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32> } [[TMP0]], 0
+// CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 16 x i32> @llvm.vector.insert.nxv16i32.nxv4i32(<vscale x 16 x i32> poison, <vscale x 4 x i32> [[TMP1]], i64 0)
+// CHECK-NEXT:    [[TMP3:%.*]] = extractvalue { <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32> } [[TMP0]], 1
+// CHECK-NEXT:    [[TMP4:%.*]] = tail call <vscale x 16 x i32> @llvm.vector.insert.nxv16i32.nxv4i32(<vscale x 16 x i32> [[TMP2]], <vscale x 4 x i32> [[TMP3]], i64 4)
+// CHECK-NEXT:    [[TMP5:%.*]] = extractvalue { <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32> } [[TMP0]], 2
+// CHECK-NEXT:    [[TMP6:%.*]] = tail call <vscale x 16 x i32> @llvm.vector.insert.nxv16i32.nxv4i32(<vscale x 16 x i32> [[TMP4]], <vscale x 4 x i32> [[TMP5]], i64 8)
+// CHECK-NEXT:    [[TMP7:%.*]] = extractvalue { <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32> } [[TMP0]], 3
+// CHECK-NEXT:    [[TMP8:%.*]] = tail call <vscale x 16 x i32> @llvm.vector.insert.nxv16i32.nxv4i32(<vscale x 16 x i32> [[TMP6]], <vscale x 4 x i32> [[TMP7]], i64 12)
+// CHECK-NEXT:    ret <vscale x 16 x i32> [[TMP8]]
+//
+// CPP-CHECK-LABEL: @_Z28test_svread_hor_za32_u32_vg4j(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call { <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32> } @llvm.aarch64.sme.read.hor.vg4.nxv4i32(i32 3, i32 [[BASE:%.*]])
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = extractvalue { <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32> } [[TMP0]], 0
+// CPP-CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 16 x i32> @llvm.vector.insert.nxv16i32.nxv4i32(<vscale x 16 x i32> poison, <vscale x 4 x i32> [[TMP1]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP3:%.*]] = extractvalue { <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32> } [[TMP0]], 1
+// CPP-CHECK-NEXT:    [[TMP4:%.*]] = tail call <vscale x 16 x i32> @llvm.vector.insert.nxv16i32.nxv4i32(<vscale x 16 x i32> [[TMP2]], <vscale x 4 x i32> [[TMP3]], i64 4)
+// CPP-CHECK-NEXT:    [[TMP5:%.*]] = extractvalue { <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32> } [[TMP0]], 2
+// CPP-CHECK-NEXT:    [[TMP6:%.*]] = tail call <vscale x 16 x i32> @llvm.vector.insert.nxv16i32.nxv4i32(<vscale x 16 x i32> [[TMP4]], <vscale x 4 x i32> [[TMP5]], i64 8)
+// CPP-CHECK-NEXT:    [[TMP7:%.*]] = extractvalue { <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32> } [[TMP0]], 3
+// CPP-CHECK-NEXT:    [[TMP8:%.*]] = tail call <vscale x 16 x i32> @llvm.vector.insert.nxv16i32.nxv4i32(<vscale x 16 x i32> [[TMP6]], <vscale x 4 x i32> [[TMP7]], i64 12)
+// CPP-CHECK-NEXT:    ret <vscale x 16 x i32> [[TMP8]]
+//
+svuint32x4_t test_svread_hor_za32_u32_vg4(uint32_t base) {
+  return svread_hor_za32_u32_vg4(3, base);
+}
+
+// CHECK-LABEL: @test_svread_hor_za32_f32_vg4(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call { <vscale x 4 x float>, <vscale x 4 x float>, <vscale x 4 x float>, <vscale x 4 x float> } @llvm.aarch64.sme.read.hor.vg4.nxv4f32(i32 3, i32 [[BASE:%.*]])
+// CHECK-NEXT:    [[TMP1:%.*]] = extractvalue { <vscale x 4 x float>, <vscale x 4 x float>, <vscale x 4 x float>, <vscale x 4 x float> } [[TMP0]], 0
+// CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 16 x float> @llvm.vector.insert.nxv16f32.nxv4f32(<vscale x 16 x float> poison, <vscale x 4 x float> [[TMP1]], i64 0)
+// CHECK-NEXT:    [[TMP3:%.*]] = extractvalue { <vscale x 4 x float>, <vscale x 4 x float>, <vscale x 4 x float>, <vscale x 4 x float> } [[TMP0]], 1
+// CHECK-NEXT:    [[TMP4:%.*]] = tail call <vscale x 16 x float> @llvm.vector.insert.nxv16f32.nxv4f32(<vscale x 16 x float> [[TMP2]], <vscale x 4 x float> [[TMP3]], i64 4)
+// CHECK-NEXT:    [[TMP5:%.*]] = extractvalue { <vscale x 4 x float>, <vscale x 4 x float>, <vscale x 4 x float>, <vscale x 4 x float> } [[TMP0]], 2
+// CHECK-NEXT:    [[TMP6:%.*]] = tail call <vscale x 16 x float> @llvm.vector.insert.nxv16f32.nxv4f32(<vscale x 16 x float> [[TMP4]], <vscale x 4 x float> [[TMP5]], i64 8)
+// CHECK-NEXT:    [[TMP7:%.*]] = extractvalue { <vscale x 4 x float>, <vscale x 4 x float>, <vscale x 4 x float>, <vscale x 4 x float> } [[TMP0]], 3
+// CHECK-NEXT:    [[TMP8:%.*]] = tail call <vscale x 16 x float> @llvm.vector.insert.nxv16f32.nxv4f32(<vscale x 16 x float> [[TMP6]], <vscale x 4 x float> [[TMP7]], i64 12)
+// CHECK-NEXT:    ret <vscale x 16 x float> [[TMP8]]
+//
+// CPP-CHECK-LABEL: @_Z28test_svread_hor_za32_f32_vg4j(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call { <vscale x 4 x float>, <vscale x 4 x float>, <vscale x 4 x float>, <vscale x 4 x float> } @llvm.aarch64.sme.read.hor.vg4.nxv4f32(i32 3, i32 [[BASE:%.*]])
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = extractvalue { <vscale x 4 x float>, <vscale x 4 x float>, <vscale x 4 x float>, <vscale x 4 x float> } [[TMP0]], 0
+// CPP-CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 16 x float> @llvm.vector.insert.nxv16f32.nxv4f32(<vscale x 16 x float> poison, <vscale x 4 x float> [[TMP1]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP3:%.*]] = extractvalue { <vscale x 4 x float>, <vscale x 4 x float>, <vscale x 4 x float>, <vscale x 4 x float> } [[TMP0]], 1
+// CPP-CHECK-NEXT:    [[TMP4:%.*]] = tail call <vscale x 16 x float> @llvm.vector.insert.nxv16f32.nxv4f32(<vscale x 16 x float> [[TMP2]], <vscale x 4 x float> [[TMP3]], i64 4)
+// CPP-CHECK-NEXT:    [[TMP5:%.*]] = extractvalue { <vscale x 4 x float>, <vscale x 4 x float>, <vscale x 4 x float>, <vscale x 4 x float> } [[TMP0]], 2
+// CPP-CHECK-NEXT:    [[TMP6:%.*]] = tail call <vscale x 16 x float> @llvm.vector.insert.nxv16f32.nxv4f32(<vscale x 16 x float> [[TMP4]], <vscale x 4 x float> [[TMP5]], i64 8)
+// CPP-CHECK-NEXT:    [[TMP7:%.*]] = extractvalue { <vscale x 4 x float>, <vscale x 4 x float>, <vscale x 4 x float>, <vscale x 4 x float> } [[TMP0]], 3
+// CPP-CHECK-NEXT:    [[TMP8:%.*]] = tail call <vscale x 16 x float> @llvm.vector.insert.nxv16f32.nxv4f32(<vscale x 16 x float> [[TMP6]], <vscale x 4 x float> [[TMP7]], i64 12)
+// CPP-CHECK-NEXT:    ret <vscale x 16 x float> [[TMP8]]
+//
+svfloat32x4_t test_svread_hor_za32_f32_vg4(uint32_t base) {
+  return svread_hor_za32_f32_vg4(3, base);
+}
+
+// CHECK-LABEL: @test_svread_hor_za32_s32_vg4(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call { <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32> } @llvm.aarch64.sme.read.hor.vg4.nxv4i32(i32 3, i32 [[BASE:%.*]])
+// CHECK-NEXT:    [[TMP1:%.*]] = extractvalue { <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32> } [[TMP0]], 0
+// CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 16 x i32> @llvm.vector.insert.nxv16i32.nxv4i32(<vscale x 16 x i32> poison, <vscale x 4 x i32> [[TMP1]], i64 0)
+// CHECK-NEXT:    [[TMP3:%.*]] = extractvalue { <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32> } [[TMP0]], 1
+// CHECK-NEXT:    [[TMP4:%.*]] = tail call <vscale x 16 x i32> @llvm.vector.insert.nxv16i32.nxv4i32(<vscale x 16 x i32> [[TMP2]], <vscale x 4 x i32> [[TMP3]], i64 4)
+// CHECK-NEXT:    [[TMP5:%.*]] = extractvalue { <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32> } [[TMP0]], 2
+// CHECK-NEXT:    [[TMP6:%.*]] = tail call <vscale x 16 x i32> @llvm.vector.insert.nxv16i32.nxv4i32(<vscale x 16 x i32> [[TMP4]], <vscale x 4 x i32> [[TMP5]], i64 8)
+// CHECK-NEXT:    [[TMP7:%.*]] = extractvalue { <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32> } [[TMP0]], 3
+// CHECK-NEXT:    [[TMP8:%.*]] = tail call <vscale x 16 x i32> @llvm.vector.insert.nxv16i32.nxv4i32(<vscale x 16 x i32> [[TMP6]], <vscale x 4 x i32> [[TMP7]], i64 12)
+// CHECK-NEXT:    ret <vscale x 16 x i32> [[TMP8]]
+//
+// CPP-CHECK-LABEL: @_Z28test_svread_hor_za32_s32_vg4j(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call { <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32> } @llvm.aarch64.sme.read.hor.vg4.nxv4i32(i32 3, i32 [[BASE:%.*]])
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = extractvalue { <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32> } [[TMP0]], 0
+// CPP-CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 16 x i32> @llvm.vector.insert.nxv16i32.nxv4i32(<vscale x 16 x i32> poison, <vscale x 4 x i32> [[TMP1]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP3:%.*]] = extractvalue { <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32> } [[TMP0]], 1
+// CPP-CHECK-NEXT:    [[TMP4:%.*]] = tail call <vscale x 16 x i32> @llvm.vector.insert.nxv16i32.nxv4i32(<vscale x 16 x i32> [[TMP2]], <vscale x 4 x i32> [[TMP3]], i64 4)
+// CPP-CHECK-NEXT:    [[TMP5:%.*]] = extractvalue { <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32> } [[TMP0]], 2
+// CPP-CHECK-NEXT:    [[TMP6:%.*]] = tail call <vscale x 16 x i32> @llvm.vector.insert.nxv16i32.nxv4i32(<vscale x 16 x i32> [[TMP4]], <vscale x 4 x i32> [[TMP5]], i64 8)
+// CPP-CHECK-NEXT:    [[TMP7:%.*]] = extractvalue { <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32> } [[TMP0]], 3
+// CPP-CHECK-NEXT:    [[TMP8:%.*]] = tail call <vscale x 16 x i32> @llvm.vector.insert.nxv16i32.nxv4i32(<vscale x 16 x i32> [[TMP6]], <vscale x 4 x i32> [[TMP7]], i64 12)
+// CPP-CHECK-NEXT:    ret <vscale x 16 x i32> [[TMP8]]
+//
+svint32x4_t test_svread_hor_za32_s32_vg4(uint32_t base) {
+  return svread_hor_za32_s32_vg4(3, base);
+}
+
+// CHECK-LABEL: @test_svread_ver_za32_u32_vg4(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call { <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32> } @llvm.aarch64.sme.read.ver.vg4.nxv4i32(i32 3, i32 [[BASE:%.*]])
+// CHECK-NEXT:    [[TMP1:%.*]] = extractvalue { <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32> } [[TMP0]], 0
+// CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 16 x i32> @llvm.vector.insert.nxv16i32.nxv4i32(<vscale x 16 x i32> poison, <vscale x 4 x i32> [[TMP1]], i64 0)
+// CHECK-NEXT:    [[TMP3:%.*]] = extractvalue { <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32> } [[TMP0]], 1
+// CHECK-NEXT:    [[TMP4:%.*]] = tail call <vscale x 16 x i32> @llvm.vector.insert.nxv16i32.nxv4i32(<vscale x 16 x i32> [[TMP2]], <vscale x 4 x i32> [[TMP3]], i64 4)
+// CHECK-NEXT:    [[TMP5:%.*]] = extractvalue { <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32> } [[TMP0]], 2
+// CHECK-NEXT:    [[TMP6:%.*]] = tail call <vscale x 16 x i32> @llvm.vector.insert.nxv16i32.nxv4i32(<vscale x 16 x i32> [[TMP4]], <vscale x 4 x i32> [[TMP5]], i64 8)
+// CHECK-NEXT:    [[TMP7:%.*]] = extractvalue { <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32> } [[TMP0]], 3
+// CHECK-NEXT:    [[TMP8:%.*]] = tail call <vscale x 16 x i32> @llvm.vector.insert.nxv16i32.nxv4i32(<vscale x 16 x i32> [[TMP6]], <vscale x 4 x i32> [[TMP7]], i64 12)
+// CHECK-NEXT:    ret <vscale x 16 x i32> [[TMP8]]
+//
+// CPP-CHECK-LABEL: @_Z28test_svread_ver_za32_u32_vg4j(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call { <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32> } @llvm.aarch64.sme.read.ver.vg4.nxv4i32(i32 3, i32 [[BASE:%.*]])
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = extractvalue { <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32> } [[TMP0]], 0
+// CPP-CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 16 x i32> @llvm.vector.insert.nxv16i32.nxv4i32(<vscale x 16 x i32> poison, <vscale x 4 x i32> [[TMP1]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP3:%.*]] = extractvalue { <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32> } [[TMP0]], 1
+// CPP-CHECK-NEXT:    [[TMP4:%.*]] = tail call <vscale x 16 x i32> @llvm.vector.insert.nxv16i32.nxv4i32(<vscale x 16 x i32> [[TMP2]], <vscale x 4 x i32> [[TMP3]], i64 4)
+// CPP-CHECK-NEXT:    [[TMP5:%.*]] = extractvalue { <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32> } [[TMP0]], 2
+// CPP-CHECK-NEXT:    [[TMP6:%.*]] = tail call <vscale x 16 x i32> @llvm.vector.insert.nxv16i32.nxv4i32(<vscale x 16 x i32> [[TMP4]], <vscale x 4 x i32> [[TMP5]], i64 8)
+// CPP-CHECK-NEXT:    [[TMP7:%.*]] = extractvalue { <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32> } [[TMP0]], 3
+// CPP-CHECK-NEXT:    [[TMP8:%.*]] = tail call <vscale x 16 x i32> @llvm.vector.insert.nxv16i32.nxv4i32(<vscale x 16 x i32> [[TMP6]], <vscale x 4 x i32> [[TMP7]], i64 12)
+// CPP-CHECK-NEXT:    ret <vscale x 16 x i32> [[TMP8]]
+//
+svuint32x4_t test_svread_ver_za32_u32_vg4(uint32_t base) {
+  return svread_ver_za32_u32_vg4(3, base);
+}
+
+// CHECK-LABEL: @test_svread_ver_za32_f32_vg4(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call { <vscale x 4 x float>, <vscale x 4 x float>, <vscale x 4 x float>, <vscale x 4 x float> } @llvm.aarch64.sme.read.ver.vg4.nxv4f32(i32 3, i32 [[BASE:%.*]])
+// CHECK-NEXT:    [[TMP1:%.*]] = extractvalue { <vscale x 4 x float>, <vscale x 4 x float>, <vscale x 4 x float>, <vscale x 4 x float> } [[TMP0]], 0
+// CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 16 x float> @llvm.vector.insert.nxv16f32.nxv4f32(<vscale x 16 x float> poison, <vscale x 4 x float> [[TMP1]], i64 0)
+// CHECK-NEXT:    [[TMP3:%.*]] = extractvalue { <vscale x 4 x float>, <vscale x 4 x float>, <vscale x 4 x float>, <vscale x 4 x float> } [[TMP0]], 1
+// CHECK-NEXT:    [[TMP4:%.*]] = tail call <vscale x 16 x float> @llvm.vector.insert.nxv16f32.nxv4f32(<vscale x 16 x float> [[TMP2]], <vscale x 4 x float> [[TMP3]], i64 4)
+// CHECK-NEXT:    [[TMP5:%.*]] = extractvalue { <vscale x 4 x float>, <vscale x 4 x float>, <vscale x 4 x float>, <vscale x 4 x float> } [[TMP0]], 2
+// CHECK-NEXT:    [[TMP6:%.*]] = tail call <vscale x 16 x float> @llvm.vector.insert.nxv16f32.nxv4f32(<vscale x 16 x float> [[TMP4]], <vscale x 4 x float> [[TMP5]], i64 8)
+// CHECK-NEXT:    [[TMP7:%.*]] = extractvalue { <vscale x 4 x float>, <vscale x 4 x float>, <vscale x 4 x float>, <vscale x 4 x float> } [[TMP0]], 3
+// CHECK-NEXT:    [[TMP8:%.*]] = tail call <vscale x 16 x float> @llvm.vector.insert.nxv16f32.nxv4f32(<vscale x 16 x float> [[TMP6]], <vscale x 4 x float> [[TMP7]], i64 12)
+// CHECK-NEXT:    ret <vscale x 16 x float> [[TMP8]]
+//
+// CPP-CHECK-LABEL: @_Z28test_svread_ver_za32_f32_vg4j(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call { <vscale x 4 x float>, <vscale x 4 x float>, <vscale x 4 x float>, <vscale x 4 x float> } @llvm.aarch64.sme.read.ver.vg4.nxv4f32(i32 3, i32 [[BASE:%.*]])
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = extractvalue { <vscale x 4 x float>, <vscale x 4 x float>, <vscale x 4 x float>, <vscale x 4 x float> } [[TMP0]], 0
+// CPP-CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 16 x float> @llvm.vector.insert.nxv16f32.nxv4f32(<vscale x 16 x float> poison, <vscale x 4 x float> [[TMP1]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP3:%.*]] = extractvalue { <vscale x 4 x float>, <vscale x 4 x float>, <vscale x 4 x float>, <vscale x 4 x float> } [[TMP0]], 1
+// CPP-CHECK-NEXT:    [[TMP4:%.*]] = tail call <vscale x 16 x float> @llvm.vector.insert.nxv16f32.nxv4f32(<vscale x 16 x float> [[TMP2]], <vscale x 4 x float> [[TMP3]], i64 4)
+// CPP-CHECK-NEXT:    [[TMP5:%.*]] = extractvalue { <vscale x 4 x float>, <vscale x 4 x float>, <vscale x 4 x float>, <vscale x 4 x float> } [[TMP0]], 2
+// CPP-CHECK-NEXT:    [[TMP6:%.*]] = tail call <vscale x 16 x float> @llvm.vector.insert.nxv16f32.nxv4f32(<vscale x 16 x float> [[TMP4]], <vscale x 4 x float> [[TMP5]], i64 8)
+// CPP-CHECK-NEXT:    [[TMP7:%.*]] = extractvalue { <vscale x 4 x float>, <vscale x 4 x float>, <vscale x 4 x float>, <vscale x 4 x float> } [[TMP0]], 3
+// CPP-CHECK-NEXT:    [[TMP8:%.*]] = tail call <vscale x 16 x float> @llvm.vector.insert.nxv16f32.nxv4f32(<vscale x 16 x float> [[TMP6]], <vscale x 4 x float> [[TMP7]], i64 12)
+// CPP-CHECK-NEXT:    ret <vscale x 16 x float> [[TMP8]]
+//
+svfloat32x4_t test_svread_ver_za32_f32_vg4(uint32_t base) {
+  return svread_ver_za32_f32_vg4(3, base);
+}
+
+// CHECK-LABEL: @test_svread_ver_za32_s32_vg4(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call { <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32> } @llvm.aarch64.sme.read.ver.vg4.nxv4i32(i32 3, i32 [[BASE:%.*]])
+// CHECK-NEXT:    [[TMP1:%.*]] = extractvalue { <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32> } [[TMP0]], 0
+// CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 16 x i32> @llvm.vector.insert.nxv16i32.nxv4i32(<vscale x 16 x i32> poison, <vscale x 4 x i32> [[TMP1]], i64 0)
+// CHECK-NEXT:    [[TMP3:%.*]] = extractvalue { <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32> } [[TMP0]], 1
+// CHECK-NEXT:    [[TMP4:%.*]] = tail call <vscale x 16 x i32> @llvm.vector.insert.nxv16i32.nxv4i32(<vscale x 16 x i32> [[TMP2]], <vscale x 4 x i32> [[TMP3]], i64 4)
+// CHECK-NEXT:    [[TMP5:%.*]] = extractvalue { <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32> } [[TMP0]], 2
+// CHECK-NEXT:    [[TMP6:%.*]] = tail call <vscale x 16 x i32> @llvm.vector.insert.nxv16i32.nxv4i32(<vscale x 16 x i32> [[TMP4]], <vscale x 4 x i32> [[TMP5]], i64 8)
+// CHECK-NEXT:    [[TMP7:%.*]] = extractvalue { <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32> } [[TMP0]], 3
+// CHECK-NEXT:    [[TMP8:%.*]] = tail call <vscale x 16 x i32> @llvm.vector.insert.nxv16i32.nxv4i32(<vscale x 16 x i32> [[TMP6]], <vscale x 4 x i32> [[TMP7]], i64 12)
+// CHECK-NEXT:    ret <vscale x 16 x i32> [[TMP8]]
+//
+// CPP-CHECK-LABEL: @_Z28test_svread_ver_za32_s32_vg4j(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call { <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32> } @llvm.aarch64.sme.read.ver.vg4.nxv4i32(i32 3, i32 [[BASE:%.*]])
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = extractvalue { <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32> } [[TMP0]], 0
+// CPP-CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 16 x i32> @llvm.vector.insert.nxv16i32.nxv4i32(<vscale x 16 x i32> poison, <vscale x 4 x i32> [[TMP1]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP3:%.*]] = extractvalue { <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32> } [[TMP0]], 1
+// CPP-CHECK-NEXT:    [[TMP4:%.*]] = tail call <vscale x 16 x i32> @llvm.vector.insert.nxv16i32.nxv4i32(<vscale x 16 x i32> [[TMP2]], <vscale x 4 x i32> [[TMP3]], i64 4)
+// CPP-CHECK-NEXT:    [[TMP5:%.*]] = extractvalue { <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32> } [[TMP0]], 2
+// CPP-CHECK-NEXT:    [[TMP6:%.*]] = tail call <vscale x 16 x i32> @llvm.vector.insert.nxv16i32.nxv4i32(<vscale x 16 x i32> [[TMP4]], <vscale x 4 x i32> [[TMP5]], i64 8)
+// CPP-CHECK-NEXT:    [[TMP7:%.*]] = extractvalue { <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32> } [[TMP0]], 3
+// CPP-CHECK-NEXT:    [[TMP8:%.*]] = tail call <vscale x 16 x i32> @llvm.vector.insert.nxv16i32.nxv4i32(<vscale x 16 x i32> [[TMP6]], <vscale x 4 x i32> [[TMP7]], i64 12)
+// CPP-CHECK-NEXT:    ret <vscale x 16 x i32> [[TMP8]]
+//
+svint32x4_t test_svread_ver_za32_s32_vg4(uint32_t base) {
+  return svread_ver_za32_s32_vg4(3, base);
+}
+
+// CHECK-LABEL: @test_svread_hor_za64_u64_vg2(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call { <vscale x 2 x i64>, <vscale x 2 x i64> } @llvm.aarch64.sme.read.hor.vg2.nxv2i64(i32 7, i32 [[BASE:%.*]])
+// CHECK-NEXT:    [[TMP1:%.*]] = extractvalue { <vscale x 2 x i64>, <vscale x 2 x i64> } [[TMP0]], 0
+// CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 4 x i64> @llvm.vector.insert.nxv4i64.nxv2i64(<vscale x 4 x i64> poison, <vscale x 2 x i64> [[TMP1]], i64 0)
+// CHECK-NEXT:    [[TMP3:%.*]] = extractvalue { <vscale x 2 x i64>, <vscale x 2 x i64> } [[TMP0]], 1
+// CHECK-NEXT:    [[TMP4:%.*]] = tail call <vscale x 4 x i64> @llvm.vector.insert.nxv4i64.nxv2i64(<vscale x 4 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP3]], i64 2)
+// CHECK-NEXT:    ret <vscale x 4 x i64> [[TMP4]]
+//
+// CPP-CHECK-LABEL: @_Z28test_svread_hor_za64_u64_vg2j(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call { <vscale x 2 x i64>, <vscale x 2 x i64> } @llvm.aarch64.sme.read.hor.vg2.nxv2i64(i32 7, i32 [[BASE:%.*]])
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = extractvalue { <vscale x 2 x i64>, <vscale x 2 x i64> } [[TMP0]], 0
+// CPP-CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 4 x i64> @llvm.vector.insert.nxv4i64.nxv2i64(<vscale x 4 x i64> poison, <vscale x 2 x i64> [[TMP1]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP3:%.*]] = extractvalue { <vscale x 2 x i64>, <vscale x 2 x i64> } [[TMP0]], 1
+// CPP-CHECK-NEXT:    [[TMP4:%.*]] = tail call <vscale x 4 x i64> @llvm.vector.insert.nxv4i64.nxv2i64(<vscale x 4 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP3]], i64 2)
+// CPP-CHECK-NEXT:    ret <vscale x 4 x i64> [[TMP4]]
+//
+svuint64x2_t test_svread_hor_za64_u64_vg2(uint32_t base) {
+  return svread_hor_za64_u64_vg2(7, base);
+}
+
+// CHECK-LABEL: @test_svread_hor_za64_f64_vg2(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call { <vscale x 2 x double>, <vscale x 2 x double> } @llvm.aarch64.sme.read.hor.vg2.nxv2f64(i32 7, i32 [[BASE:%.*]])
+// CHECK-NEXT:    [[TMP1:%.*]] = extractvalue { <vscale x 2 x double>, <vscale x 2 x double> } [[TMP0]], 0
+// CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 4 x double> @llvm.vector.insert.nxv4f64.nxv2f64(<vscale x 4 x double> poison, <vscale x 2 x double> [[TMP1]], i64 0)
+// CHECK-NEXT:    [[TMP3:%.*]] = extractvalue { <vscale x 2 x double>, <vscale x 2 x double> } [[TMP0]], 1
+// CHECK-NEXT:    [[TMP4:%.*]] = tail call <vscale x 4 x double> @llvm.vector.insert.nxv4f64.nxv2f64(<vscale x 4 x double> [[TMP2]], <vscale x 2 x double> [[TMP3]], i64 2)
+// CHECK-NEXT:    ret <vscale x 4 x double> [[TMP4]]
+//
+// CPP-CHECK-LABEL: @_Z28test_svread_hor_za64_f64_vg2j(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call { <vscale x 2 x double>, <vscale x 2 x double> } @llvm.aarch64.sme.read.hor.vg2.nxv2f64(i32 7, i32 [[BASE:%.*]])
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = extractvalue { <vscale x 2 x double>, <vscale x 2 x double> } [[TMP0]], 0
+// CPP-CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 4 x double> @llvm.vector.insert.nxv4f64.nxv2f64(<vscale x 4 x double> poison, <vscale x 2 x double> [[TMP1]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP3:%.*]] = extractvalue { <vscale x 2 x double>, <vscale x 2 x double> } [[TMP0]], 1
+// CPP-CHECK-NEXT:    [[TMP4:%.*]] = tail call <vscale x 4 x double> @llvm.vector.insert.nxv4f64.nxv2f64(<vscale x 4 x double> [[TMP2]], <vscale x 2 x double> [[TMP3]], i64 2)
+// CPP-CHECK-NEXT:    ret <vscale x 4 x double> [[TMP4]]
+//
+svfloat64x2_t test_svread_hor_za64_f64_vg2(uint32_t base) {
+  return svread_hor_za64_f64_vg2(7, base);
+}
+
+// CHECK-LABEL: @test_svread_hor_za64_s64_vg2(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call { <vscale x 2 x i64>, <vscale x 2 x i64> } @llvm.aarch64.sme.read.hor.vg2.nxv2i64(i32 7, i32 [[BASE:%.*]])
+// CHECK-NEXT:    [[TMP1:%.*]] = extractvalue { <vscale x 2 x i64>, <vscale x 2 x i64> } [[TMP0]], 0
+// CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 4 x i64> @llvm.vector.insert.nxv4i64.nxv2i64(<vscale x 4 x i64> poison, <vscale x 2 x i64> [[TMP1]], i64 0)
+// CHECK-NEXT:    [[TMP3:%.*]] = extractvalue { <vscale x 2 x i64>, <vscale x 2 x i64> } [[TMP0]], 1
+// CHECK-NEXT:    [[TMP4:%.*]] = tail call <vscale x 4 x i64> @llvm.vector.insert.nxv4i64.nxv2i64(<vscale x 4 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP3]], i64 2)
+// CHECK-NEXT:    ret <vscale x 4 x i64> [[TMP4]]
+//
+// CPP-CHECK-LABEL: @_Z28test_svread_hor_za64_s64_vg2j(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call { <vscale x 2 x i64>, <vscale x 2 x i64> } @llvm.aarch64.sme.read.hor.vg2.nxv2i64(i32 7, i32 [[BASE:%.*]])
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = extractvalue { <vscale x 2 x i64>, <vscale x 2 x i64> } [[TMP0]], 0
+// CPP-CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 4 x i64> @llvm.vector.insert.nxv4i64.nxv2i64(<vscale x 4 x i64> poison, <vscale x 2 x i64> [[TMP1]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP3:%.*]] = extractvalue { <vscale x 2 x i64>, <vscale x 2 x i64> } [[TMP0]], 1
+// CPP-CHECK-NEXT:    [[TMP4:%.*]] = tail call <vscale x 4 x i64> @llvm.vector.insert.nxv4i64.nxv2i64(<vscale x 4 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP3]], i64 2)
+// CPP-CHECK-NEXT:    ret <vscale x 4 x i64> [[TMP4]]
+//
+svint64x2_t test_svread_hor_za64_s64_vg2(uint32_t base) {
+  return svread_hor_za64_s64_vg2(7, base);
+}
+
+// CHECK-LABEL: @test_svread_ver_za64_u64_vg2(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call { <vscale x 2 x i64>, <vscale x 2 x i64> } @llvm.aarch64.sme.read.ver.vg2.nxv2i64(i32 7, i32 [[BASE:%.*]])
+// CHECK-NEXT:    [[TMP1:%.*]] = extractvalue { <vscale x 2 x i64>, <vscale x 2 x i64> } [[TMP0]], 0
+// CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 4 x i64> @llvm.vector.insert.nxv4i64.nxv2i64(<vscale x 4 x i64> poison, <vscale x 2 x i64> [[TMP1]], i64 0)
+// CHECK-NEXT:    [[TMP3:%.*]] = extractvalue { <vscale x 2 x i64>, <vscale x 2 x i64> } [[TMP0]], 1
+// CHECK-NEXT:    [[TMP4:%.*]] = tail call <vscale x 4 x i64> @llvm.vector.insert.nxv4i64.nxv2i64(<vscale x 4 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP3]], i64 2)
+// CHECK-NEXT:    ret <vscale x 4 x i64> [[TMP4]]
+//
+// CPP-CHECK-LABEL: @_Z28test_svread_ver_za64_u64_vg2j(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call { <vscale x 2 x i64>, <vscale x 2 x i64> } @llvm.aarch64.sme.read.ver.vg2.nxv2i64(i32 7, i32 [[BASE:%.*]])
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = extractvalue { <vscale x 2 x i64>, <vscale x 2 x i64> } [[TMP0]], 0
+// CPP-CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 4 x i64> @llvm.vector.insert.nxv4i64.nxv2i64(<vscale x 4 x i64> poison, <vscale x 2 x i64> [[TMP1]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP3:%.*]] = extractvalue { <vscale x 2 x i64>, <vscale x 2 x i64> } [[TMP0]], 1
+// CPP-CHECK-NEXT:    [[TMP4:%.*]] = tail call <vscale x 4 x i64> @llvm.vector.insert.nxv4i64.nxv2i64(<vscale x 4 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP3]], i64 2)
+// CPP-CHECK-NEXT:    ret <vscale x 4 x i64> [[TMP4]]
+//
+svuint64x2_t test_svread_ver_za64_u64_vg2(uint32_t base) {
+  return svread_ver_za64_u64_vg2(7, base);
+}
+
+// CHECK-LABEL: @test_svread_ver_za64_f64_vg2(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call { <vscale x 2 x double>, <vscale x 2 x double> } @llvm.aarch64.sme.read.ver.vg2.nxv2f64(i32 7, i32 [[BASE:%.*]])
+// CHECK-NEXT:    [[TMP1:%.*]] = extractvalue { <vscale x 2 x double>, <vscale x 2 x double> } [[TMP0]], 0
+// CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 4 x double> @llvm.vector.insert.nxv4f64.nxv2f64(<vscale x 4 x double> poison, <vscale x 2 x double> [[TMP1]], i64 0)
+// CHECK-NEXT:    [[TMP3:%.*]] = extractvalue { <vscale x 2 x double>, <vscale x 2 x double> } [[TMP0]], 1
+// CHECK-NEXT:    [[TMP4:%.*]] = tail call <vscale x 4 x double> @llvm.vector.insert.nxv4f64.nxv2f64(<vscale x 4 x double> [[TMP2]], <vscale x 2 x double> [[TMP3]], i64 2)
+// CHECK-NEXT:    ret <vscale x 4 x double> [[TMP4]]
+//
+// CPP-CHECK-LABEL: @_Z28test_svread_ver_za64_f64_vg2j(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call { <vscale x 2 x double>, <vscale x 2 x double> } @llvm.aarch64.sme.read.ver.vg2.nxv2f64(i32 7, i32 [[BASE:%.*]])
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = extractvalue { <vscale x 2 x double>, <vscale x 2 x double> } [[TMP0]], 0
+// CPP-CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 4 x double> @llvm.vector.insert.nxv4f64.nxv2f64(<vscale x 4 x double> poison, <vscale x 2 x double> [[TMP1]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP3:%.*]] = extractvalue { <vscale x 2 x double>, <vscale x 2 x double> } [[TMP0]], 1
+// CPP-CHECK-NEXT:    [[TMP4:%.*]] = tail call <vscale x 4 x double> @llvm.vector.insert.nxv4f64.nxv2f64(<vscale x 4 x double> [[TMP2]], <vscale x 2 x double> [[TMP3]], i64 2)
+// CPP-CHECK-NEXT:    ret <vscale x 4 x double> [[TMP4]]
+//
+svfloat64x2_t test_svread_ver_za64_f64_vg2(uint32_t base) {
+  return svread_ver_za64_f64_vg2(7, base);
+}
+
+// CHECK-LABEL: @test_svread_ver_za64_s64_vg2(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call { <vscale x 2 x i64>, <vscale x 2 x i64> } @llvm.aarch64.sme.read.ver.vg2.nxv2i64(i32 7, i32 [[BASE:%.*]])
+// CHECK-NEXT:    [[TMP1:%.*]] = extractvalue { <vscale x 2 x i64>, <vscale x 2 x i64> } [[TMP0]], 0
+// CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 4 x i64> @llvm.vector.insert.nxv4i64.nxv2i64(<vscale x 4 x i64> poison, <vscale x 2 x i64> [[TMP1]], i64 0)
+// CHECK-NEXT:    [[TMP3:%.*]] = extractvalue { <vscale x 2 x i64>, <vscale x 2 x i64> } [[TMP0]], 1
+// CHECK-NEXT:    [[TMP4:%.*]] = tail call <vscale x 4 x i64> @llvm.vector.insert.nxv4i64.nxv2i64(<vscale x 4 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP3]], i64 2)
+// CHECK-NEXT:    ret <vscale x 4 x i64> [[TMP4]]
+//
+// CPP-CHECK-LABEL: @_Z28test_svread_ver_za64_s64_vg2j(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call { <vscale x 2 x i64>, <vscale x 2 x i64> } @llvm.aarch64.sme.read.ver.vg2.nxv2i64(i32 7, i32 [[BASE:%.*]])
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = extractvalue { <vscale x 2 x i64>, <vscale x 2 x i64> } [[TMP0]], 0
+// CPP-CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 4 x i64> @llvm.vector.insert.nxv4i64.nxv2i64(<vscale x 4 x i64> poison, <vscale x 2 x i64> [[TMP1]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP3:%.*]] = extractvalue { <vscale x 2 x i64>, <vscale x 2 x i64> } [[TMP0]], 1
+// CPP-CHECK-NEXT:    [[TMP4:%.*]] = tail call <vscale x 4 x i64> @llvm.vector.insert.nxv4i64.nxv2i64(<vscale x 4 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP3]], i64 2)
+// CPP-CHECK-NEXT:    ret <vscale x 4 x i64> [[TMP4]]
+//
+svint64x2_t test_svread_ver_za64_s64_vg2(uint32_t base) {
+  return svread_ver_za64_s64_vg2(7, base);
+}
+
+// CHECK-LABEL: @test_svread_hor_za64_u64_vg4(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call { <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64> } @llvm.aarch64.sme.read.hor.vg4.nxv2i64(i32 7, i32 [[BASE:%.*]])
+// CHECK-NEXT:    [[TMP1:%.*]] = extractvalue { <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64> } [[TMP0]], 0
+// CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 8 x i64> @llvm.vector.insert.nxv8i64.nxv2i64(<vscale x 8 x i64> poison, <vscale x 2 x i64> [[TMP1]], i64 0)
+// CHECK-NEXT:    [[TMP3:%.*]] = extractvalue { <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64> } [[TMP0]], 1
+// CHECK-NEXT:    [[TMP4:%.*]] = tail call <vscale x 8 x i64> @llvm.vector.insert.nxv8i64.nxv2i64(<vscale x 8 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP3]], i64 2)
+// CHECK-NEXT:    [[TMP5:%.*]] = extractvalue { <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64> } [[TMP0]], 2
+// CHECK-NEXT:    [[TMP6:%.*]] = tail call <vscale x 8 x i64> @llvm.vector.insert.nxv8i64.nxv2i64(<vscale x 8 x i64> [[TMP4]], <vscale x 2 x i64> [[TMP5]], i64 4)
+// CHECK-NEXT:    [[TMP7:%.*]] = extractvalue { <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64> } [[TMP0]], 3
+// CHECK-NEXT:    [[TMP8:%.*]] = tail call <vscale x 8 x i64> @llvm.vector.insert.nxv8i64.nxv2i64(<vscale x 8 x i64> [[TMP6]], <vscale x 2 x i64> [[TMP7]], i64 6)
+// CHECK-NEXT:    ret <vscale x 8 x i64> [[TMP8]]
+//
+// CPP-CHECK-LABEL: @_Z28test_svread_hor_za64_u64_vg4j(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call { <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64> } @llvm.aarch64.sme.read.hor.vg4.nxv2i64(i32 7, i32 [[BASE:%.*]])
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = extractvalue { <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64> } [[TMP0]], 0
+// CPP-CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 8 x i64> @llvm.vector.insert.nxv8i64.nxv2i64(<vscale x 8 x i64> poison, <vscale x 2 x i64> [[TMP1]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP3:%.*]] = extractvalue { <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64> } [[TMP0]], 1
+// CPP-CHECK-NEXT:    [[TMP4:%.*]] = tail call <vscale x 8 x i64> @llvm.vector.insert.nxv8i64.nxv2i64(<vscale x 8 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP3]], i64 2)
+// CPP-CHECK-NEXT:    [[TMP5:%.*]] = extractvalue { <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64> } [[TMP0]], 2
+// CPP-CHECK-NEXT:    [[TMP6:%.*]] = tail call <vscale x 8 x i64> @llvm.vector.insert.nxv8i64.nxv2i64(<vscale x 8 x i64> [[TMP4]], <vscale x 2 x i64> [[TMP5]], i64 4)
+// CPP-CHECK-NEXT:    [[TMP7:%.*]] = extractvalue { <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64> } [[TMP0]], 3
+// CPP-CHECK-NEXT:    [[TMP8:%.*]] = tail call <vscale x 8 x i64> @llvm.vector.insert.nxv8i64.nxv2i64(<vscale x 8 x i64> [[TMP6]], <vscale x 2 x i64> [[TMP7]], i64 6)
+// CPP-CHECK-NEXT:    ret <vscale x 8 x i64> [[TMP8]]
+//
+svuint64x4_t test_svread_hor_za64_u64_vg4(uint32_t base) {
+  return svread_hor_za64_u64_vg4(7, base);
+}
+
+// CHECK-LABEL: @test_svread_hor_za64_f64_vg4(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call { <vscale x 2 x double>, <vscale x 2 x double>, <vscale x 2 x double>, <vscale x 2 x double> } @llvm.aarch64.sme.read.hor.vg4.nxv2f64(i32 7, i32 [[BASE:%.*]])
+// CHECK-NEXT:    [[TMP1:%.*]] = extractvalue { <vscale x 2 x double>, <vscale x 2 x double>, <vscale x 2 x double>, <vscale x 2 x double> } [[TMP0]], 0
+// CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 8 x double> @llvm.vector.insert.nxv8f64.nxv2f64(<vscale x 8 x double> poison, <vscale x 2 x double> [[TMP1]], i64 0)
+// CHECK-NEXT:    [[TMP3:%.*]] = extractvalue { <vscale x 2 x double>, <vscale x 2 x double>, <vscale x 2 x double>, <vscale x 2 x double> } [[TMP0]], 1
+// CHECK-NEXT:    [[TMP4:%.*]] = tail call <vscale x 8 x double> @llvm.vector.insert.nxv8f64.nxv2f64(<vscale x 8 x double> [[TMP2]], <vscale x 2 x double> [[TMP3]], i64 2)
+// CHECK-NEXT:    [[TMP5:%.*]] = extractvalue { <vscale x 2 x double>, <vscale x 2 x double>, <vscale x 2 x double>, <vscale x 2 x double> } [[TMP0]], 2
+// CHECK-NEXT:    [[TMP6:%.*]] = tail call <vscale x 8 x double> @llvm.vector.insert.nxv8f64.nxv2f64(<vscale x 8 x double> [[TMP4]], <vscale x 2 x double> [[TMP5]], i64 4)
+// CHECK-NEXT:    [[TMP7:%.*]] = extractvalue { <vscale x 2 x double>, <vscale x 2 x double>, <vscale x 2 x double>, <vscale x 2 x double> } [[TMP0]], 3
+// CHECK-NEXT:    [[TMP8:%.*]] = tail call <vscale x 8 x double> @llvm.vector.insert.nxv8f64.nxv2f64(<vscale x 8 x double> [[TMP6]], <vscale x 2 x double> [[TMP7]], i64 6)
+// CHECK-NEXT:    ret <vscale x 8 x double> [[TMP8]]
+//
+// CPP-CHECK-LABEL: @_Z28test_svread_hor_za64_f64_vg4j(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call { <vscale x 2 x double>, <vscale x 2 x double>, <vscale x 2 x double>, <vscale x 2 x double> } @llvm.aarch64.sme.read.hor.vg4.nxv2f64(i32 7, i32 [[BASE:%.*]])
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = extractvalue { <vscale x 2 x double>, <vscale x 2 x double>, <vscale x 2 x double>, <vscale x 2 x double> } [[TMP0]], 0
+// CPP-CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 8 x double> @llvm.vector.insert.nxv8f64.nxv2f64(<vscale x 8 x double> poison, <vscale x 2 x double> [[TMP1]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP3:%.*]] = extractvalue { <vscale x 2 x double>, <vscale x 2 x double>, <vscale x 2 x double>, <vscale x 2 x double> } [[TMP0]], 1
+// CPP-CHECK-NEXT:    [[TMP4:%.*]] = tail call <vscale x 8 x double> @llvm.vector.insert.nxv8f64.nxv2f64(<vscale x 8 x double> [[TMP2]], <vscale x 2 x double> [[TMP3]], i64 2)
+// CPP-CHECK-NEXT:    [[TMP5:%.*]] = extractvalue { <vscale x 2 x double>, <vscale x 2 x double>, <vscale x 2 x double>, <vscale x 2 x double> } [[TMP0]], 2
+// CPP-CHECK-NEXT:    [[TMP6:%.*]] = tail call <vscale x 8 x double> @llvm.vector.insert.nxv8f64.nxv2f64(<vscale x 8 x double> [[TMP4]], <vscale x 2 x double> [[TMP5]], i64 4)
+// CPP-CHECK-NEXT:    [[TMP7:%.*]] = extractvalue { <vscale x 2 x double>, <vscale x 2 x double>, <vscale x 2 x double>, <vscale x 2 x double> } [[TMP0]], 3
+// CPP-CHECK-NEXT:    [[TMP8:%.*]] = tail call <vscale x 8 x double> @llvm.vector.insert.nxv8f64.nxv2f64(<vscale x 8 x double> [[TMP6]], <vscale x 2 x double> [[TMP7]], i64 6)
+// CPP-CHECK-NEXT:    ret <vscale x 8 x double> [[TMP8]]
+//
+svfloat64x4_t test_svread_hor_za64_f64_vg4(uint32_t base) {
+  return svread_hor_za64_f64_vg4(7, base);
+}
+
+// CHECK-LABEL: @test_svread_hor_za64_s64_vg4(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call { <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64> } @llvm.aarch64.sme.read.hor.vg4.nxv2i64(i32 7, i32 [[BASE:%.*]])
+// CHECK-NEXT:    [[TMP1:%.*]] = extractvalue { <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64> } [[TMP0]], 0
+// CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 8 x i64> @llvm.vector.insert.nxv8i64.nxv2i64(<vscale x 8 x i64> poison, <vscale x 2 x i64> [[TMP1]], i64 0)
+// CHECK-NEXT:    [[TMP3:%.*]] = extractvalue { <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64> } [[TMP0]], 1
+// CHECK-NEXT:    [[TMP4:%.*]] = tail call <vscale x 8 x i64> @llvm.vector.insert.nxv8i64.nxv2i64(<vscale x 8 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP3]], i64 2)
+// CHECK-NEXT:    [[TMP5:%.*]] = extractvalue { <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64> } [[TMP0]], 2
+// CHECK-NEXT:    [[TMP6:%.*]] = tail call <vscale x 8 x i64> @llvm.vector.insert.nxv8i64.nxv2i64(<vscale x 8 x i64> [[TMP4]], <vscale x 2 x i64> [[TMP5]], i64 4)
+// CHECK-NEXT:    [[TMP7:%.*]] = extractvalue { <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64> } [[TMP0]], 3
+// CHECK-NEXT:    [[TMP8:%.*]] = tail call <vscale x 8 x i64> @llvm.vector.insert.nxv8i64.nxv2i64(<vscale x 8 x i64> [[TMP6]], <vscale x 2 x i64> [[TMP7]], i64 6)
+// CHECK-NEXT:    ret <vscale x 8 x i64> [[TMP8]]
+//
+// CPP-CHECK-LABEL: @_Z28test_svread_hor_za64_s64_vg4j(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call { <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64> } @llvm.aarch64.sme.read.hor.vg4.nxv2i64(i32 7, i32 [[BASE:%.*]])
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = extractvalue { <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64> } [[TMP0]], 0
+// CPP-CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 8 x i64> @llvm.vector.insert.nxv8i64.nxv2i64(<vscale x 8 x i64> poison, <vscale x 2 x i64> [[TMP1]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP3:%.*]] = extractvalue { <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64> } [[TMP0]], 1
+// CPP-CHECK-NEXT:    [[TMP4:%.*]] = tail call <vscale x 8 x i64> @llvm.vector.insert.nxv8i64.nxv2i64(<vscale x 8 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP3]], i64 2)
+// CPP-CHECK-NEXT:    [[TMP5:%.*]] = extractvalue { <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64> } [[TMP0]], 2
+// CPP-CHECK-NEXT:    [[TMP6:%.*]] = tail call <vscale x 8 x i64> @llvm.vector.insert.nxv8i64.nxv2i64(<vscale x 8 x i64> [[TMP4]], <vscale x 2 x i64> [[TMP5]], i64 4)
+// CPP-CHECK-NEXT:    [[TMP7:%.*]] = extractvalue { <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64> } [[TMP0]], 3
+// CPP-CHECK-NEXT:    [[TMP8:%.*]] = tail call <vscale x 8 x i64> @llvm.vector.insert.nxv8i64.nxv2i64(<vscale x 8 x i64> [[TMP6]], <vscale x 2 x i64> [[TMP7]], i64 6)
+// CPP-CHECK-NEXT:    ret <vscale x 8 x i64> [[TMP8]]
+//
+svint64x4_t test_svread_hor_za64_s64_vg4(uint32_t base) {
+  return svread_hor_za64_s64_vg4(7, base);
+}
+
+// CHECK-LABEL: @test_svread_ver_za64_u64_vg4(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call { <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64> } @llvm.aarch64.sme.read.ver.vg4.nxv2i64(i32 7, i32 [[BASE:%.*]])
+// CHECK-NEXT:    [[TMP1:%.*]] = extractvalue { <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64> } [[TMP0]], 0
+// CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 8 x i64> @llvm.vector.insert.nxv8i64.nxv2i64(<vscale x 8 x i64> poison, <vscale x 2 x i64> [[TMP1]], i64 0)
+// CHECK-NEXT:    [[TMP3:%.*]] = extractvalue { <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64> } [[TMP0]], 1
+// CHECK-NEXT:    [[TMP4:%.*]] = tail call <vscale x 8 x i64> @llvm.vector.insert.nxv8i64.nxv2i64(<vscale x 8 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP3]], i64 2)
+// CHECK-NEXT:    [[TMP5:%.*]] = extractvalue { <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64> } [[TMP0]], 2
+// CHECK-NEXT:    [[TMP6:%.*]] = tail call <vscale x 8 x i64> @llvm.vector.insert.nxv8i64.nxv2i64(<vscale x 8 x i64> [[TMP4]], <vscale x 2 x i64> [[TMP5]], i64 4)
+// CHECK-NEXT:    [[TMP7:%.*]] = extractvalue { <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64> } [[TMP0]], 3
+// CHECK-NEXT:    [[TMP8:%.*]] = tail call <vscale x 8 x i64> @llvm.vector.insert.nxv8i64.nxv2i64(<vscale x 8 x i64> [[TMP6]], <vscale x 2 x i64> [[TMP7]], i64 6)
+// CHECK-NEXT:    ret <vscale x 8 x i64> [[TMP8]]
+//
+// CPP-CHECK-LABEL: @_Z28test_svread_ver_za64_u64_vg4j(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call { <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64> } @llvm.aarch64.sme.read.ver.vg4.nxv2i64(i32 7, i32 [[BASE:%.*]])
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = extractvalue { <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64> } [[TMP0]], 0
+// CPP-CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 8 x i64> @llvm.vector.insert.nxv8i64.nxv2i64(<vscale x 8 x i64> poison, <vscale x 2 x i64> [[TMP1]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP3:%.*]] = extractvalue { <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64> } [[TMP0]], 1
+// CPP-CHECK-NEXT:    [[TMP4:%.*]] = tail call <vscale x 8 x i64> @llvm.vector.insert.nxv8i64.nxv2i64(<vscale x 8 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP3]], i64 2)
+// CPP-CHECK-NEXT:    [[TMP5:%.*]] = extractvalue { <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64> } [[TMP0]], 2
+// CPP-CHECK-NEXT:    [[TMP6:%.*]] = tail call <vscale x 8 x i64> @llvm.vector.insert.nxv8i64.nxv2i64(<vscale x 8 x i64> [[TMP4]], <vscale x 2 x i64> [[TMP5]], i64 4)
+// CPP-CHECK-NEXT:    [[TMP7:%.*]] = extractvalue { <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64> } [[TMP0]], 3
+// CPP-CHECK-NEXT:    [[TMP8:%.*]] = tail call <vscale x 8 x i64> @llvm.vector.insert.nxv8i64.nxv2i64(<vscale x 8 x i64> [[TMP6]], <vscale x 2 x i64> [[TMP7]], i64 6)
+// CPP-CHECK-NEXT:    ret <vscale x 8 x i64> [[TMP8]]
+//
+svuint64x4_t test_svread_ver_za64_u64_vg4(uint32_t base) {
+  return svread_ver_za64_u64_vg4(7, base);
+}
+
+// CHECK-LABEL: @test_svread_ver_za64_f64_vg4(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call { <vscale x 2 x double>, <vscale x 2 x double>, <vscale x 2 x double>, <vscale x 2 x double> } @llvm.aarch64.sme.read.ver.vg4.nxv2f64(i32 7, i32 [[BASE:%.*]])
+// CHECK-NEXT:    [[TMP1:%.*]] = extractvalue { <vscale x 2 x double>, <vscale x 2 x double>, <vscale x 2 x double>, <vscale x 2 x double> } [[TMP0]], 0
+// CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 8 x double> @llvm.vector.insert.nxv8f64.nxv2f64(<vscale x 8 x double> poison, <vscale x 2 x double> [[TMP1]], i64 0)
+// CHECK-NEXT:    [[TMP3:%.*]] = extractvalue { <vscale x 2 x double>, <vscale x 2 x double>, <vscale x 2 x double>, <vscale x 2 x double> } [[TMP0]], 1
+// CHECK-NEXT:    [[TMP4:%.*]] = tail call <vscale x 8 x double> @llvm.vector.insert.nxv8f64.nxv2f64(<vscale x 8 x double> [[TMP2]], <vscale x 2 x double> [[TMP3]], i64 2)
+// CHECK-NEXT:    [[TMP5:%.*]] = extractvalue { <vscale x 2 x double>, <vscale x 2 x double>, <vscale x 2 x double>, <vscale x 2 x double> } [[TMP0]], 2
+// CHECK-NEXT:    [[TMP6:%.*]] = tail call <vscale x 8 x double> @llvm.vector.insert.nxv8f64.nxv2f64(<vscale x 8 x double> [[TMP4]], <vscale x 2 x double> [[TMP5]], i64 4)
+// CHECK-NEXT:    [[TMP7:%.*]] = extractvalue { <vscale x 2 x double>, <vscale x 2 x double>, <vscale x 2 x double>, <vscale x 2 x double> } [[TMP0]], 3
+// CHECK-NEXT:    [[TMP8:%.*]] = tail call <vscale x 8 x double> @llvm.vector.insert.nxv8f64.nxv2f64(<vscale x 8 x double> [[TMP6]], <vscale x 2 x double> [[TMP7]], i64 6)
+// CHECK-NEXT:    ret <vscale x 8 x double> [[TMP8]]
+//
+// CPP-CHECK-LABEL: @_Z28test_svread_ver_za64_f64_vg4j(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call { <vscale x 2 x double>, <vscale x 2 x double>, <vscale x 2 x double>, <vscale x 2 x double> } @llvm.aarch64.sme.read.ver.vg4.nxv2f64(i32 7, i32 [[BASE:%.*]])
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = extractvalue { <vscale x 2 x double>, <vscale x 2 x double>, <vscale x 2 x double>, <vscale x 2 x double> } [[TMP0]], 0
+// CPP-CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 8 x double> @llvm.vector.insert.nxv8f64.nxv2f64(<vscale x 8 x double> poison, <vscale x 2 x double> [[TMP1]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP3:%.*]] = extractvalue { <vscale x 2 x double>, <vscale x 2 x double>, <vscale x 2 x double>, <vscale x 2 x double> } [[TMP0]], 1
+// CPP-CHECK-NEXT:    [[TMP4:%.*]] = tail call <vscale x 8 x double> @llvm.vector.insert.nxv8f64.nxv2f64(<vscale x 8 x double> [[TMP2]], <vscale x 2 x double> [[TMP3]], i64 2)
+// CPP-CHECK-NEXT:    [[TMP5:%.*]] = extractvalue { <vscale x 2 x double>, <vscale x 2 x double>, <vscale x 2 x double>, <vscale x 2 x double> } [[TMP0]], 2
+// CPP-CHECK-NEXT:    [[TMP6:%.*]] = tail call <vscale x 8 x double> @llvm.vector.insert.nxv8f64.nxv2f64(<vscale x 8 x double> [[TMP4]], <vscale x 2 x double> [[TMP5]], i64 4)
+// CPP-CHECK-NEXT:    [[TMP7:%.*]] = extractvalue { <vscale x 2 x double>, <vscale x 2 x double>, <vscale x 2 x double>, <vscale x 2 x double> } [[TMP0]], 3
+// CPP-CHECK-NEXT:    [[TMP8:%.*]] = tail call <vscale x 8 x double> @llvm.vector.insert.nxv8f64.nxv2f64(<vscale x 8 x double> [[TMP6]], <vscale x 2 x double> [[TMP7]], i64 6)
+// CPP-CHECK-NEXT:    ret <vscale x 8 x double> [[TMP8]]
+//
+svfloat64x4_t test_svread_ver_za64_f64_vg4(uint32_t base) {
+  return svread_ver_za64_f64_vg4(7, base);
+}
+
+// CHECK-LABEL: @test_svread_ver_za64_s64_vg4(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call { <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64> } @llvm.aarch64.sme.read.ver.vg4.nxv2i64(i32 7, i32 [[BASE:%.*]])
+// CHECK-NEXT:    [[TMP1:%.*]] = extractvalue { <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64> } [[TMP0]], 0
+// CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 8 x i64> @llvm.vector.insert.nxv8i64.nxv2i64(<vscale x 8 x i64> poison, <vscale x 2 x i64> [[TMP1]], i64 0)
+// CHECK-NEXT:    [[TMP3:%.*]] = extractvalue { <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64> } [[TMP0]], 1
+// CHECK-NEXT:    [[TMP4:%.*]] = tail call <vscale x 8 x i64> @llvm.vector.insert.nxv8i64.nxv2i64(<vscale x 8 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP3]], i64 2)
+// CHECK-NEXT:    [[TMP5:%.*]] = extractvalue { <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64> } [[TMP0]], 2
+// CHECK-NEXT:    [[TMP6:%.*]] = tail call <vscale x 8 x i64> @llvm.vector.insert.nxv8i64.nxv2i64(<vscale x 8 x i64> [[TMP4]], <vscale x 2 x i64> [[TMP5]], i64 4)
+// CHECK-NEXT:    [[TMP7:%.*]] = extractvalue { <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64> } [[TMP0]], 3
+// CHECK-NEXT:    [[TMP8:%.*]] = tail call <vscale x 8 x i64> @llvm.vector.insert.nxv8i64.nxv2i64(<vscale x 8 x i64> [[TMP6]], <vscale x 2 x i64> [[TMP7]], i64 6)
+// CHECK-NEXT:    ret <vscale x 8 x i64> [[TMP8]]
+//
+// CPP-CHECK-LABEL: @_Z28test_svread_ver_za64_s64_vg4j(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call { <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64> } @llvm.aarch64.sme.read.ver.vg4.nxv2i64(i32 7, i32 [[BASE:%.*]])
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = extractvalue { <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64> } [[TMP0]], 0
+// CPP-CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 8 x i64> @llvm.vector.insert.nxv8i64.nxv2i64(<vscale x 8 x i64> poison, <vscale x 2 x i64> [[TMP1]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP3:%.*]] = extractvalue { <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64> } [[TMP0]], 1
+// CPP-CHECK-NEXT:    [[TMP4:%.*]] = tail call <vscale x 8 x i64> @llvm.vector.insert.nxv8i64.nxv2i64(<vscale x 8 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP3]], i64 2)
+// CPP-CHECK-NEXT:    [[TMP5:%.*]] = extractvalue { <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64> } [[TMP0]], 2
+// CPP-CHECK-NEXT:    [[TMP6:%.*]] = tail call <vscale x 8 x i64> @llvm.vector.insert.nxv8i64.nxv2i64(<vscale x 8 x i64> [[TMP4]], <vscale x 2 x i64> [[TMP5]], i64 4)
+// CPP-CHECK-NEXT:    [[TMP7:%.*]] = extractvalue { <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64> } [[TMP0]], 3
+// CPP-CHECK-NEXT:    [[TMP8:%.*]] = tail call <vscale x 8 x i64> @llvm.vector.insert.nxv8i64.nxv2i64(<vscale x 8 x i64> [[TMP6]], <vscale x 2 x i64> [[TMP7]], i64 6)
+// CPP-CHECK-NEXT:    ret <vscale x 8 x i64> [[TMP8]]
+//
+svint64x4_t test_svread_ver_za64_s64_vg4(uint32_t base) {
+  return svread_ver_za64_s64_vg4(7, base);
+}
+
+// CHECK-LABEL: @test_svread_za64_u64_vg1x2(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call { <vscale x 2 x i64>, <vscale x 2 x i64> } @llvm.aarch64.sme.read.vg1x2.nxv2i64(i32 [[BASE:%.*]])
+// CHECK-NEXT:    [[TMP1:%.*]] = extractvalue { <vscale x 2 x i64>, <vscale x 2 x i64> } [[TMP0]], 0
+// CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 4 x i64> @llvm.vector.insert.nxv4i64.nxv2i64(<vscale x 4 x i64> poison, <vscale x 2 x i64> [[TMP1]], i64 0)
+// CHECK-NEXT:    [[TMP3:%.*]] = extractvalue { <vscale x 2 x i64>, <vscale x 2 x i64> } [[TMP0]], 1
+// CHECK-NEXT:    [[TMP4:%.*]] = tail call <vscale x 4 x i64> @llvm.vector.insert.nxv4i64.nxv2i64(<vscale x 4 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP3]], i64 2)
+// CHECK-NEXT:    ret <vscale x 4 x i64> [[TMP4]]
+//
+// CPP-CHECK-LABEL: @_Z26test_svread_za64_u64_vg1x2j(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call { <vscale x 2 x i64>, <vscale x 2 x i64> } @llvm.aarch64.sme.read.vg1x2.nxv2i64(i32 [[BASE:%.*]])
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = extractvalue { <vscale x 2 x i64>, <vscale x 2 x i64> } [[TMP0]], 0
+// CPP-CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 4 x i64> @llvm.vector.insert.nxv4i64.nxv2i64(<vscale x 4 x i64> poison, <vscale x 2 x i64> [[TMP1]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP3:%.*]] = extractvalue { <vscale x 2 x i64>, <vscale x 2 x i64> } [[TMP0]], 1
+// CPP-CHECK-NEXT:    [[TMP4:%.*]] = tail call <vscale x 4 x i64> @llvm.vector.insert.nxv4i64.nxv2i64(<vscale x 4 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP3]], i64 2)
+// CPP-CHECK-NEXT:    ret <vscale x 4 x i64> [[TMP4]]
+//
+svuint64x2_t test_svread_za64_u64_vg1x2(uint32_t base) {
+  return svread_za64_u64_vg1x2(base);
+}
+
+// CHECK-LABEL: @test_svread_za64_f64_vg1x2(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call { <vscale x 2 x double>, <vscale x 2 x double> } @llvm.aarch64.sme.read.vg1x2.nxv2f64(i32 [[BASE:%.*]])
+// CHECK-NEXT:    [[TMP1:%.*]] = extractvalue { <vscale x 2 x double>, <vscale x 2 x double> } [[TMP0]], 0
+// CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 4 x double> @llvm.vector.insert.nxv4f64.nxv2f64(<vscale x 4 x double> poison, <vscale x 2 x double> [[TMP1]], i64 0)
+// CHECK-NEXT:    [[TMP3:%.*]] = extractvalue { <vscale x 2 x double>, <vscale x 2 x double> } [[TMP0]], 1
+// CHECK-NEXT:    [[TMP4:%.*]] = tail call <vscale x 4 x double> @llvm.vector.insert.nxv4f64.nxv2f64(<vscale x 4 x double> [[TMP2]], <vscale x 2 x double> [[TMP3]], i64 2)
+// CHECK-NEXT:    ret <vscale x 4 x double> [[TMP4]]
+//
+// CPP-CHECK-LABEL: @_Z26test_svread_za64_f64_vg1x2j(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call { <vscale x 2 x double>, <vscale x 2 x double> } @llvm.aarch64.sme.read.vg1x2.nxv2f64(i32 [[BASE:%.*]])
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = extractvalue { <vscale x 2 x double>, <vscale x 2 x double> } [[TMP0]], 0
+// CPP-CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 4 x double> @llvm.vector.insert.nxv4f64.nxv2f64(<vscale x 4 x double> poison, <vscale x 2 x double> [[TMP1]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP3:%.*]] = extractvalue { <vscale x 2 x double>, <vscale x 2 x double> } [[TMP0]], 1
+// CPP-CHECK-NEXT:    [[TMP4:%.*]] = tail call <vscale x 4 x double> @llvm.vector.insert.nxv4f64.nxv2f64(<vscale x 4 x double> [[TMP2]], <vscale x 2 x double> [[TMP3]], i64 2)
+// CPP-CHECK-NEXT:    ret <vscale x 4 x double> [[TMP4]]
+//
+svfloat64x2_t test_svread_za64_f64_vg1x2(uint32_t base) {
+  return svread_za64_f64_vg1x2(base);
+}
+
+// CHECK-LABEL: @test_svread_za64_s64_vg1x2(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call { <vscale x 2 x i64>, <vscale x 2 x i64> } @llvm.aarch64.sme.read.vg1x2.nxv2i64(i32 [[BASE:%.*]])
+// CHECK-NEXT:    [[TMP1:%.*]] = extractvalue { <vscale x 2 x i64>, <vscale x 2 x i64> } [[TMP0]], 0
+// CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 4 x i64> @llvm.vector.insert.nxv4i64.nxv2i64(<vscale x 4 x i64> poison, <vscale x 2 x i64> [[TMP1]], i64 0)
+// CHECK-NEXT:    [[TMP3:%.*]] = extractvalue { <vscale x 2 x i64>, <vscale x 2 x i64> } [[TMP0]], 1
+// CHECK-NEXT:    [[TMP4:%.*]] = tail call <vscale x 4 x i64> @llvm.vector.insert.nxv4i64.nxv2i64(<vscale x 4 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP3]], i64 2)
+// CHECK-NEXT:    ret <vscale x 4 x i64> [[TMP4]]
+//
+// CPP-CHECK-LABEL: @_Z26test_svread_za64_s64_vg1x2j(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call { <vscale x 2 x i64>, <vscale x 2 x i64> } @llvm.aarch64.sme.read.vg1x2.nxv2i64(i32 [[BASE:%.*]])
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = extractvalue { <vscale x 2 x i64>, <vscale x 2 x i64> } [[TMP0]], 0
+// CPP-CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 4 x i64> @llvm.vector.insert.nxv4i64.nxv2i64(<vscale x 4 x i64> poison, <vscale x 2 x i64> [[TMP1]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP3:%.*]] = extractvalue { <vscale x 2 x i64>, <vscale x 2 x i64> } [[TMP0]], 1
+// CPP-CHECK-NEXT:    [[TMP4:%.*]] = tail call <vscale x 4 x i64> @llvm.vector.insert.nxv4i64.nxv2i64(<vscale x 4 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP3]], i64 2)
+// CPP-CHECK-NEXT:    ret <vscale x 4 x i64> [[TMP4]]
+//
+svint64x2_t test_svread_za64_s64_vg1x2(uint32_t base) {
+  return svread_za64_s64_vg1x2(base);
+}
+
+// CHECK-LABEL: @test_svread_za64_u64_vg1x4(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call { <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64> } @llvm.aarch64.sme.read.vg1x4.nxv2i64(i32 [[BASE:%.*]])
+// CHECK-NEXT:    [[TMP1:%.*]] = extractvalue { <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64> } [[TMP0]], 0
+// CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 8 x i64> @llvm.vector.insert.nxv8i64.nxv2i64(<vscale x 8 x i64> poison, <vscale x 2 x i64> [[TMP1]], i64 0)
+// CHECK-NEXT:    [[TMP3:%.*]] = extractvalue { <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64> } [[TMP0]], 1
+// CHECK-NEXT:    [[TMP4:%.*]] = tail call <vscale x 8 x i64> @llvm.vector.insert.nxv8i64.nxv2i64(<vscale x 8 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP3]], i64 2)
+// CHECK-NEXT:    [[TMP5:%.*]] = extractvalue { <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64> } [[TMP0]], 2
+// CHECK-NEXT:    [[TMP6:%.*]] = tail call <vscale x 8 x i64> @llvm.vector.insert.nxv8i64.nxv2i64(<vscale x 8 x i64> [[TMP4]], <vscale x 2 x i64> [[TMP5]], i64 4)
+// CHECK-NEXT:    [[TMP7:%.*]] = extractvalue { <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64> } [[TMP0]], 3
+// CHECK-NEXT:    [[TMP8:%.*]] = tail call <vscale x 8 x i64> @llvm.vector.insert.nxv8i64.nxv2i64(<vscale x 8 x i64> [[TMP6]], <vscale x 2 x i64> [[TMP7]], i64 6)
+// CHECK-NEXT:    ret <vscale x 8 x i64> [[TMP8]]
+//
+// CPP-CHECK-LABEL: @_Z26test_svread_za64_u64_vg1x4j(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call { <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64> } @llvm.aarch64.sme.read.vg1x4.nxv2i64(i32 [[BASE:%.*]])
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = extractvalue { <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64> } [[TMP0]], 0
+// CPP-CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 8 x i64> @llvm.vector.insert.nxv8i64.nxv2i64(<vscale x 8 x i64> poison, <vscale x 2 x i64> [[TMP1]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP3:%.*]] = extractvalue { <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64> } [[TMP0]], 1
+// CPP-CHECK-NEXT:    [[TMP4:%.*]] = tail call <vscale x 8 x i64> @llvm.vector.insert.nxv8i64.nxv2i64(<vscale x 8 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP3]], i64 2)
+// CPP-CHECK-NEXT:    [[TMP5:%.*]] = extractvalue { <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64> } [[TMP0]], 2
+// CPP-CHECK-NEXT:    [[TMP6:%.*]] = tail call <vscale x 8 x i64> @llvm.vector.insert.nxv8i64.nxv2i64(<vscale x 8 x i64> [[TMP4]], <vscale x 2 x i64> [[TMP5]], i64 4)
+// CPP-CHECK-NEXT:    [[TMP7:%.*]] = extractvalue { <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64> } [[TMP0]], 3
+// CPP-CHECK-NEXT:    [[TMP8:%.*]] = tail call <vscale x 8 x i64> @llvm.vector.insert.nxv8i64.nxv2i64(<vscale x 8 x i64> [[TMP6]], <vscale x 2 x i64> [[TMP7]], i64 6)
+// CPP-CHECK-NEXT:    ret <vscale x 8 x i64> [[TMP8]]
+//
+svuint64x4_t test_svread_za64_u64_vg1x4(uint32_t base) {
+  return svread_za64_u64_vg1x4(base);
+}
+
+// CHECK-LABEL: @test_svread_za64_f64_vg1x4(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call { <vscale x 2 x double>, <vscale x 2 x double>, <vscale x 2 x double>, <vscale x 2 x double> } @llvm.aarch64.sme.read.vg1x4.nxv2f64(i32 [[BASE:%.*]])
+// CHECK-NEXT:    [[TMP1:%.*]] = extractvalue { <vscale x 2 x double>, <vscale x 2 x double>, <vscale x 2 x double>, <vscale x 2 x double> } [[TMP0]], 0
+// CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 8 x double> @llvm.vector.insert.nxv8f64.nxv2f64(<vscale x 8 x double> poison, <vscale x 2 x double> [[TMP1]], i64 0)
+// CHECK-NEXT:    [[TMP3:%.*]] = extractvalue { <vscale x 2 x double>, <vscale x 2 x double>, <vscale x 2 x double>, <vscale x 2 x double> } [[TMP0]], 1
+// CHECK-NEXT:    [[TMP4:%.*]] = tail call <vscale x 8 x double> @llvm.vector.insert.nxv8f64.nxv2f64(<vscale x 8 x double> [[TMP2]], <vscale x 2 x double> [[TMP3]], i64 2)
+// CHECK-NEXT:    [[TMP5:%.*]] = extractvalue { <vscale x 2 x double>, <vscale x 2 x double>, <vscale x 2 x double>, <vscale x 2 x double> } [[TMP0]], 2
+// CHECK-NEXT:    [[TMP6:%.*]] = tail call <vscale x 8 x double> @llvm.vector.insert.nxv8f64.nxv2f64(<vscale x 8 x double> [[TMP4]], <vscale x 2 x double> [[TMP5]], i64 4)
+// CHECK-NEXT:    [[TMP7:%.*]] = extractvalue { <vscale x 2 x double>, <vscale x 2 x double>, <vscale x 2 x double>, <vscale x 2 x double> } [[TMP0]], 3
+// CHECK-NEXT:    [[TMP8:%.*]] = tail call <vscale x 8 x double> @llvm.vector.insert.nxv8f64.nxv2f64(<vscale x 8 x double> [[TMP6]], <vscale x 2 x double> [[TMP7]], i64 6)
+// CHECK-NEXT:    ret <vscale x 8 x double> [[TMP8]]
+//
+// CPP-CHECK-LABEL: @_Z26test_svread_za64_f64_vg1x4j(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call { <vscale x 2 x double>, <vscale x 2 x double>, <vscale x 2 x double>, <vscale x 2 x double> } @llvm.aarch64.sme.read.vg1x4.nxv2f64(i32 [[BASE:%.*]])
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = extractvalue { <vscale x 2 x double>, <vscale x 2 x double>, <vscale x 2 x double>, <vscale x 2 x double> } [[TMP0]], 0
+// CPP-CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 8 x double> @llvm.vector.insert.nxv8f64.nxv2f64(<vscale x 8 x double> poison, <vscale x 2 x double> [[TMP1]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP3:%.*]] = extractvalue { <vscale x 2 x double>, <vscale x 2 x double>, <vscale x 2 x double>, <vscale x 2 x double> } [[TMP0]], 1
+// CPP-CHECK-NEXT:    [[TMP4:%.*]] = tail call <vscale x 8 x double> @llvm.vector.insert.nxv8f64.nxv2f64(<vscale x 8 x double> [[TMP2]], <vscale x 2 x double> [[TMP3]], i64 2)
+// CPP-CHECK-NEXT:    [[TMP5:%.*]] = extractvalue { <vscale x 2 x double>, <vscale x 2 x double>, <vscale x 2 x double>, <vscale x 2 x double> } [[TMP0]], 2
+// CPP-CHECK-NEXT:    [[TMP6:%.*]] = tail call <vscale x 8 x double> @llvm.vector.insert.nxv8f64.nxv2f64(<vscale x 8 x double> [[TMP4]], <vscale x 2 x double> [[TMP5]], i64 4)
+// CPP-CHECK-NEXT:    [[TMP7:%.*]] = extractvalue { <vscale x 2 x double>, <vscale x 2 x double>, <vscale x 2 x double>, <vscale x 2 x double> } [[TMP0]], 3
+// CPP-CHECK-NEXT:    [[TMP8:%.*]] = tail call <vscale x 8 x double> @llvm.vector.insert.nxv8f64.nxv2f64(<vscale x 8 x double> [[TMP6]], <vscale x 2 x double> [[TMP7]], i64 6)
+// CPP-CHECK-NEXT:    ret <vscale x 8 x double> [[TMP8]]
+//
+svfloat64x4_t test_svread_za64_f64_vg1x4(uint32_t base) {
+  return svread_za64_f64_vg1x4(base);
+}
+
+// CHECK-LABEL: @test_svread_za64_s64_vg1x4(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call { <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64> } @llvm.aarch64.sme.read.vg1x4.nxv2i64(i32 [[BASE:%.*]])
+// CHECK-NEXT:    [[TMP1:%.*]] = extractvalue { <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64> } [[TMP0]], 0
+// CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 8 x i64> @llvm.vector.insert.nxv8i64.nxv2i64(<vscale x 8 x i64> poison, <vscale x 2 x i64> [[TMP1]], i64 0)
+// CHECK-NEXT:    [[TMP3:%.*]] = extractvalue { <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64> } [[TMP0]], 1
+// CHECK-NEXT:    [[TMP4:%.*]] = tail call <vscale x 8 x i64> @llvm.vector.insert.nxv8i64.nxv2i64(<vscale x 8 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP3]], i64 2)
+// CHECK-NEXT:    [[TMP5:%.*]] = extractvalue { <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64> } [[TMP0]], 2
+// CHECK-NEXT:    [[TMP6:%.*]] = tail call <vscale x 8 x i64> @llvm.vector.insert.nxv8i64.nxv2i64(<vscale x 8 x i64> [[TMP4]], <vscale x 2 x i64> [[TMP5]], i64 4)
+// CHECK-NEXT:    [[TMP7:%.*]] = extractvalue { <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64> } [[TMP0]], 3
+// CHECK-NEXT:    [[TMP8:%.*]] = tail call <vscale x 8 x i64> @llvm.vector.insert.nxv8i64.nxv2i64(<vscale x 8 x i64> [[TMP6]], <vscale x 2 x i64> [[TMP7]], i64 6)
+// CHECK-NEXT:    ret <vscale x 8 x i64> [[TMP8]]
+//
+// CPP-CHECK-LABEL: @_Z26test_svread_za64_s64_vg1x4j(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call { <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64> } @llvm.aarch64.sme.read.vg1x4.nxv2i64(i32 [[BASE:%.*]])
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = extractvalue { <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64> } [[TMP0]], 0
+// CPP-CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 8 x i64> @llvm.vector.insert.nxv8i64.nxv2i64(<vscale x 8 x i64> poison, <vscale x 2 x i64> [[TMP1]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP3:%.*]] = extractvalue { <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64> } [[TMP0]], 1
+// CPP-CHECK-NEXT:    [[TMP4:%.*]] = tail call <vscale x 8 x i64> @llvm.vector.insert.nxv8i64.nxv2i64(<vscale x 8 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP3]], i64 2)
+// CPP-CHECK-NEXT:    [[TMP5:%.*]] = extractvalue { <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64> } [[TMP0]], 2
+// CPP-CHECK-NEXT:    [[TMP6:%.*]] = tail call <vscale x 8 x i64> @llvm.vector.insert.nxv8i64.nxv2i64(<vscale x 8 x i64> [[TMP4]], <vscale x 2 x i64> [[TMP5]], i64 4)
+// CPP-CHECK-NEXT:    [[TMP7:%.*]] = extractvalue { <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64> } [[TMP0]], 3
+// CPP-CHECK-NEXT:    [[TMP8:%.*]] = tail call <vscale x 8 x i64> @llvm.vector.insert.nxv8i64.nxv2i64(<vscale x 8 x i64> [[TMP6]], <vscale x 2 x i64> [[TMP7]], i64 6)
+// CPP-CHECK-NEXT:    ret <vscale x 8 x i64> [[TMP8]]
+//
+svint64x4_t test_svread_za64_s64_vg1x4(uint32_t base) {
+  return svread_za64_s64_vg1x4(base);
+}
+
diff --git a/clang/test/CodeGen/aarch64-sme2-intrinsics/acle_sme2_write.c b/clang/test/CodeGen/aarch64-sme2-intrinsics/acle_sme2_write.c
new file mode 100644
index 0000000000000..745eb7f3edb56
--- /dev/null
+++ b/clang/test/CodeGen/aarch64-sme2-intrinsics/acle_sme2_write.c
@@ -0,0 +1,1097 @@
+// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py
+// REQUIRES: aarch64-registered-target
+
+// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sme2 -target-feature +sve -S -disable-O0-optnone -Werror -Wall -emit-llvm -o - %s | opt -S -p mem2reg,instcombine,tailcallelim | FileCheck %s
+// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sme2 -target-feature +sve -S -disable-O0-optnone -Werror -Wall -emit-llvm -o - -x c++ %s | opt -S -p mem2reg,instcombine,tailcallelim | FileCheck %s -check-prefix=CPP-CHECK
+// RUN: %clang_cc1 -DSVE_OVERLOADED_FORMS -triple aarch64-none-linux-gnu -target-feature +sme2 -target-feature +sve -S -disable-O0-optnone -Werror -Wall -emit-llvm -o - %s | opt -S -p mem2reg,instcombine,tailcallelim | FileCheck %s
+// RUN: %clang_cc1 -DSVE_OVERLOADED_FORMS -triple aarch64-none-linux-gnu -target-feature +sme2 -S -target-feature +sve -disable-O0-optnone -Werror -Wall -emit-llvm -o - -x c++ %s | opt -S -p mem2reg,instcombine,tailcallelim | FileCheck %s -check-prefix=CPP-CHECK
+// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sme2 -S -target-feature +sve -disable-O0-optnone -Werror -Wall -o /dev/null %s
+
+#include <arm_sme_draft_spec_subject_to_change.h>
+
+#ifdef SVE_OVERLOADED_FORMS
+// A simple used,unused... macro, long enough to represent any SVE builtin.
+#define SVE_ACLE_FUNC(A1,A2_UNUSED,A3,A4_UNUSED) A1##A3
+#else
+#define SVE_ACLE_FUNC(A1,A2,A3,A4) A1##A2##A3##A4
+#endif
+
+// CHECK-LABEL: @test_svwrite_ver_za8_u8_vg2(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv32i8(<vscale x 32 x i8> [[VAL:%.*]], i64 0)
+// CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv32i8(<vscale x 32 x i8> [[VAL]], i64 16)
+// CHECK-NEXT:    tail call void @llvm.aarch64.sme.write.ver.vg2.nxv16i8(i32 0, i32 [[BASE:%.*]], <vscale x 16 x i8> [[TMP0]], <vscale x 16 x i8> [[TMP1]])
+// CHECK-NEXT:    ret void
+//
+// CPP-CHECK-LABEL: @_Z27test_svwrite_ver_za8_u8_vg2j11svuint8x2_t(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv32i8(<vscale x 32 x i8> [[VAL:%.*]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv32i8(<vscale x 32 x i8> [[VAL]], i64 16)
+// CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.write.ver.vg2.nxv16i8(i32 0, i32 [[BASE:%.*]], <vscale x 16 x i8> [[TMP0]], <vscale x 16 x i8> [[TMP1]])
+// CPP-CHECK-NEXT:    ret void
+//
+void test_svwrite_ver_za8_u8_vg2(uint32_t base, svuint8x2_t val) {
+  SVE_ACLE_FUNC(svwrite_ver_za8,_u8,_vg2,)(0, base, val);
+}
+
+// CHECK-LABEL: @test_svwrite_ver_za8_s8_vg2(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv32i8(<vscale x 32 x i8> [[VAL:%.*]], i64 0)
+// CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv32i8(<vscale x 32 x i8> [[VAL]], i64 16)
+// CHECK-NEXT:    tail call void @llvm.aarch64.sme.write.ver.vg2.nxv16i8(i32 0, i32 [[BASE:%.*]], <vscale x 16 x i8> [[TMP0]], <vscale x 16 x i8> [[TMP1]])
+// CHECK-NEXT:    ret void
+//
+// CPP-CHECK-LABEL: @_Z27test_svwrite_ver_za8_s8_vg2j10svint8x2_t(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv32i8(<vscale x 32 x i8> [[VAL:%.*]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv32i8(<vscale x 32 x i8> [[VAL]], i64 16)
+// CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.write.ver.vg2.nxv16i8(i32 0, i32 [[BASE:%.*]], <vscale x 16 x i8> [[TMP0]], <vscale x 16 x i8> [[TMP1]])
+// CPP-CHECK-NEXT:    ret void
+//
+void test_svwrite_ver_za8_s8_vg2(uint32_t base, svint8x2_t val) {
+  SVE_ACLE_FUNC(svwrite_ver_za8,_s8,_vg2,)(0, base, val);
+}
+
+// CHECK-LABEL: @test_svwrite_hor_za8_u8_vg2(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv32i8(<vscale x 32 x i8> [[VAL:%.*]], i64 0)
+// CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv32i8(<vscale x 32 x i8> [[VAL]], i64 16)
+// CHECK-NEXT:    tail call void @llvm.aarch64.sme.write.hor.vg2.nxv16i8(i32 0, i32 [[BASE:%.*]], <vscale x 16 x i8> [[TMP0]], <vscale x 16 x i8> [[TMP1]])
+// CHECK-NEXT:    ret void
+//
+// CPP-CHECK-LABEL: @_Z27test_svwrite_hor_za8_u8_vg2j11svuint8x2_t(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv32i8(<vscale x 32 x i8> [[VAL:%.*]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv32i8(<vscale x 32 x i8> [[VAL]], i64 16)
+// CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.write.hor.vg2.nxv16i8(i32 0, i32 [[BASE:%.*]], <vscale x 16 x i8> [[TMP0]], <vscale x 16 x i8> [[TMP1]])
+// CPP-CHECK-NEXT:    ret void
+//
+void test_svwrite_hor_za8_u8_vg2(uint32_t base, svuint8x2_t val) {
+  SVE_ACLE_FUNC(svwrite_hor_za8,_u8,_vg2,)(0, base, val);
+}
+
+// CHECK-LABEL: @test_svwrite_hor_za8_s8_vg2(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv32i8(<vscale x 32 x i8> [[VAL:%.*]], i64 0)
+// CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv32i8(<vscale x 32 x i8> [[VAL]], i64 16)
+// CHECK-NEXT:    tail call void @llvm.aarch64.sme.write.hor.vg2.nxv16i8(i32 0, i32 [[BASE:%.*]], <vscale x 16 x i8> [[TMP0]], <vscale x 16 x i8> [[TMP1]])
+// CHECK-NEXT:    ret void
+//
+// CPP-CHECK-LABEL: @_Z27test_svwrite_hor_za8_s8_vg2j10svint8x2_t(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv32i8(<vscale x 32 x i8> [[VAL:%.*]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv32i8(<vscale x 32 x i8> [[VAL]], i64 16)
+// CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.write.hor.vg2.nxv16i8(i32 0, i32 [[BASE:%.*]], <vscale x 16 x i8> [[TMP0]], <vscale x 16 x i8> [[TMP1]])
+// CPP-CHECK-NEXT:    ret void
+//
+void test_svwrite_hor_za8_s8_vg2(uint32_t base, svint8x2_t val) {
+  SVE_ACLE_FUNC(svwrite_hor_za8,_s8,_vg2,)(0, base, val);
+}
+
+// CHECK-LABEL: @test_svwrite_hor_za8_u8_vg4(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[VAL:%.*]], i64 0)
+// CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[VAL]], i64 16)
+// CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[VAL]], i64 32)
+// CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[VAL]], i64 48)
+// CHECK-NEXT:    tail call void @llvm.aarch64.sme.write.hor.vg4.nxv16i8(i32 0, i32 [[BASE:%.*]], <vscale x 16 x i8> [[TMP0]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP2]], <vscale x 16 x i8> [[TMP3]])
+// CHECK-NEXT:    ret void
+//
+// CPP-CHECK-LABEL: @_Z27test_svwrite_hor_za8_u8_vg4j11svuint8x4_t(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[VAL:%.*]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[VAL]], i64 16)
+// CPP-CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[VAL]], i64 32)
+// CPP-CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[VAL]], i64 48)
+// CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.write.hor.vg4.nxv16i8(i32 0, i32 [[BASE:%.*]], <vscale x 16 x i8> [[TMP0]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP2]], <vscale x 16 x i8> [[TMP3]])
+// CPP-CHECK-NEXT:    ret void
+//
+void test_svwrite_hor_za8_u8_vg4(uint32_t base, svuint8x4_t val) {
+  SVE_ACLE_FUNC(svwrite_hor_za8,_u8,_vg4,)(0, base, val);
+}
+
+// CHECK-LABEL: @test_svwrite_hor_za8_s8_vg4(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[VAL:%.*]], i64 0)
+// CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[VAL]], i64 16)
+// CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[VAL]], i64 32)
+// CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[VAL]], i64 48)
+// CHECK-NEXT:    tail call void @llvm.aarch64.sme.write.hor.vg4.nxv16i8(i32 0, i32 [[BASE:%.*]], <vscale x 16 x i8> [[TMP0]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP2]], <vscale x 16 x i8> [[TMP3]])
+// CHECK-NEXT:    ret void
+//
+// CPP-CHECK-LABEL: @_Z27test_svwrite_hor_za8_s8_vg4j10svint8x4_t(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[VAL:%.*]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[VAL]], i64 16)
+// CPP-CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[VAL]], i64 32)
+// CPP-CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[VAL]], i64 48)
+// CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.write.hor.vg4.nxv16i8(i32 0, i32 [[BASE:%.*]], <vscale x 16 x i8> [[TMP0]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP2]], <vscale x 16 x i8> [[TMP3]])
+// CPP-CHECK-NEXT:    ret void
+//
+void test_svwrite_hor_za8_s8_vg4(uint32_t base, svint8x4_t val) {
+  SVE_ACLE_FUNC(svwrite_hor_za8,_s8,_vg4,)(0, base, val);
+}
+
+// CHECK-LABEL: @test_svwrite_ver_za8_u8_vg4(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[VAL:%.*]], i64 0)
+// CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[VAL]], i64 16)
+// CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[VAL]], i64 32)
+// CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[VAL]], i64 48)
+// CHECK-NEXT:    tail call void @llvm.aarch64.sme.write.ver.vg4.nxv16i8(i32 0, i32 [[BASE:%.*]], <vscale x 16 x i8> [[TMP0]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP2]], <vscale x 16 x i8> [[TMP3]])
+// CHECK-NEXT:    ret void
+//
+// CPP-CHECK-LABEL: @_Z27test_svwrite_ver_za8_u8_vg4j11svuint8x4_t(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[VAL:%.*]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[VAL]], i64 16)
+// CPP-CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[VAL]], i64 32)
+// CPP-CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[VAL]], i64 48)
+// CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.write.ver.vg4.nxv16i8(i32 0, i32 [[BASE:%.*]], <vscale x 16 x i8> [[TMP0]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP2]], <vscale x 16 x i8> [[TMP3]])
+// CPP-CHECK-NEXT:    ret void
+//
+void test_svwrite_ver_za8_u8_vg4(uint32_t base, svuint8x4_t val) {
+  SVE_ACLE_FUNC(svwrite_ver_za8,_u8,_vg4,)(0, base, val);
+}
+
+// CHECK-LABEL: @test_svwrite_ver_za8_s8_vg4(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[VAL:%.*]], i64 0)
+// CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[VAL]], i64 16)
+// CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[VAL]], i64 32)
+// CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[VAL]], i64 48)
+// CHECK-NEXT:    tail call void @llvm.aarch64.sme.write.ver.vg4.nxv16i8(i32 0, i32 [[BASE:%.*]], <vscale x 16 x i8> [[TMP0]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP2]], <vscale x 16 x i8> [[TMP3]])
+// CHECK-NEXT:    ret void
+//
+// CPP-CHECK-LABEL: @_Z27test_svwrite_ver_za8_s8_vg4j10svint8x4_t(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[VAL:%.*]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[VAL]], i64 16)
+// CPP-CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[VAL]], i64 32)
+// CPP-CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[VAL]], i64 48)
+// CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.write.ver.vg4.nxv16i8(i32 0, i32 [[BASE:%.*]], <vscale x 16 x i8> [[TMP0]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP2]], <vscale x 16 x i8> [[TMP3]])
+// CPP-CHECK-NEXT:    ret void
+//
+void test_svwrite_ver_za8_s8_vg4(uint32_t base, svint8x4_t val) {
+  SVE_ACLE_FUNC(svwrite_ver_za8,_s8,_vg4,)(0, base, val);
+}
+
+// CHECK-LABEL: @test_svwrite_hor_za16_u16_vg2(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv16i16(<vscale x 16 x i16> [[VAL:%.*]], i64 0)
+// CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv16i16(<vscale x 16 x i16> [[VAL]], i64 8)
+// CHECK-NEXT:    tail call void @llvm.aarch64.sme.write.hor.vg2.nxv8i16(i32 1, i32 [[BASE:%.*]], <vscale x 8 x i16> [[TMP0]], <vscale x 8 x i16> [[TMP1]])
+// CHECK-NEXT:    ret void
+//
+// CPP-CHECK-LABEL: @_Z29test_svwrite_hor_za16_u16_vg2j12svuint16x2_t(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv16i16(<vscale x 16 x i16> [[VAL:%.*]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv16i16(<vscale x 16 x i16> [[VAL]], i64 8)
+// CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.write.hor.vg2.nxv8i16(i32 1, i32 [[BASE:%.*]], <vscale x 8 x i16> [[TMP0]], <vscale x 8 x i16> [[TMP1]])
+// CPP-CHECK-NEXT:    ret void
+//
+void test_svwrite_hor_za16_u16_vg2(uint32_t base, svuint16x2_t val) {
+  SVE_ACLE_FUNC(svwrite_hor_za16,_u16,_vg2,)(1, base, val);
+}
+
+// CHECK-LABEL: @test_svwrite_hor_za16_bf16_vg2(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 8 x bfloat> @llvm.vector.extract.nxv8bf16.nxv16bf16(<vscale x 16 x bfloat> [[VAL:%.*]], i64 0)
+// CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 8 x bfloat> @llvm.vector.extract.nxv8bf16.nxv16bf16(<vscale x 16 x bfloat> [[VAL]], i64 8)
+// CHECK-NEXT:    tail call void @llvm.aarch64.sme.write.hor.vg2.nxv8bf16(i32 1, i32 [[BASE:%.*]], <vscale x 8 x bfloat> [[TMP0]], <vscale x 8 x bfloat> [[TMP1]])
+// CHECK-NEXT:    ret void
+//
+// CPP-CHECK-LABEL: @_Z30test_svwrite_hor_za16_bf16_vg2j14svbfloat16x2_t(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 8 x bfloat> @llvm.vector.extract.nxv8bf16.nxv16bf16(<vscale x 16 x bfloat> [[VAL:%.*]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 8 x bfloat> @llvm.vector.extract.nxv8bf16.nxv16bf16(<vscale x 16 x bfloat> [[VAL]], i64 8)
+// CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.write.hor.vg2.nxv8bf16(i32 1, i32 [[BASE:%.*]], <vscale x 8 x bfloat> [[TMP0]], <vscale x 8 x bfloat> [[TMP1]])
+// CPP-CHECK-NEXT:    ret void
+//
+void test_svwrite_hor_za16_bf16_vg2(uint32_t base, svbfloat16x2_t val) {
+  SVE_ACLE_FUNC(svwrite_hor_za16,_bf16,_vg2,)(1, base, val);
+}
+
+// CHECK-LABEL: @test_svwrite_hor_za16_f16_vg2(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 8 x half> @llvm.vector.extract.nxv8f16.nxv16f16(<vscale x 16 x half> [[VAL:%.*]], i64 0)
+// CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 8 x half> @llvm.vector.extract.nxv8f16.nxv16f16(<vscale x 16 x half> [[VAL]], i64 8)
+// CHECK-NEXT:    tail call void @llvm.aarch64.sme.write.hor.vg2.nxv8f16(i32 1, i32 [[BASE:%.*]], <vscale x 8 x half> [[TMP0]], <vscale x 8 x half> [[TMP1]])
+// CHECK-NEXT:    ret void
+//
+// CPP-CHECK-LABEL: @_Z29test_svwrite_hor_za16_f16_vg2j13svfloat16x2_t(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 8 x half> @llvm.vector.extract.nxv8f16.nxv16f16(<vscale x 16 x half> [[VAL:%.*]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 8 x half> @llvm.vector.extract.nxv8f16.nxv16f16(<vscale x 16 x half> [[VAL]], i64 8)
+// CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.write.hor.vg2.nxv8f16(i32 1, i32 [[BASE:%.*]], <vscale x 8 x half> [[TMP0]], <vscale x 8 x half> [[TMP1]])
+// CPP-CHECK-NEXT:    ret void
+//
+void test_svwrite_hor_za16_f16_vg2(uint32_t base, svfloat16x2_t val) {
+  SVE_ACLE_FUNC(svwrite_hor_za16,_f16,_vg2,)(1, base, val);
+}
+
+// CHECK-LABEL: @test_svwrite_hor_za16_s16_vg2(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv16i16(<vscale x 16 x i16> [[VAL:%.*]], i64 0)
+// CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv16i16(<vscale x 16 x i16> [[VAL]], i64 8)
+// CHECK-NEXT:    tail call void @llvm.aarch64.sme.write.hor.vg2.nxv8i16(i32 1, i32 [[BASE:%.*]], <vscale x 8 x i16> [[TMP0]], <vscale x 8 x i16> [[TMP1]])
+// CHECK-NEXT:    ret void
+//
+// CPP-CHECK-LABEL: @_Z29test_svwrite_hor_za16_s16_vg2j11svint16x2_t(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv16i16(<vscale x 16 x i16> [[VAL:%.*]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv16i16(<vscale x 16 x i16> [[VAL]], i64 8)
+// CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.write.hor.vg2.nxv8i16(i32 1, i32 [[BASE:%.*]], <vscale x 8 x i16> [[TMP0]], <vscale x 8 x i16> [[TMP1]])
+// CPP-CHECK-NEXT:    ret void
+//
+void test_svwrite_hor_za16_s16_vg2(uint32_t base, svint16x2_t val) {
+  SVE_ACLE_FUNC(svwrite_hor_za16,_s16,_vg2,)(1, base, val);
+}
+
+// CHECK-LABEL: @test_svwrite_ver_za16_u16_vg2(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv16i16(<vscale x 16 x i16> [[VAL:%.*]], i64 0)
+// CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv16i16(<vscale x 16 x i16> [[VAL]], i64 8)
+// CHECK-NEXT:    tail call void @llvm.aarch64.sme.write.ver.vg2.nxv8i16(i32 1, i32 [[BASE:%.*]], <vscale x 8 x i16> [[TMP0]], <vscale x 8 x i16> [[TMP1]])
+// CHECK-NEXT:    ret void
+//
+// CPP-CHECK-LABEL: @_Z29test_svwrite_ver_za16_u16_vg2j12svuint16x2_t(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv16i16(<vscale x 16 x i16> [[VAL:%.*]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv16i16(<vscale x 16 x i16> [[VAL]], i64 8)
+// CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.write.ver.vg2.nxv8i16(i32 1, i32 [[BASE:%.*]], <vscale x 8 x i16> [[TMP0]], <vscale x 8 x i16> [[TMP1]])
+// CPP-CHECK-NEXT:    ret void
+//
+void test_svwrite_ver_za16_u16_vg2(uint32_t base, svuint16x2_t val) {
+  SVE_ACLE_FUNC(svwrite_ver_za16,_u16,_vg2,)(1, base, val);
+}
+
+// CHECK-LABEL: @test_svwrite_ver_za16_bf16_vg2(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 8 x bfloat> @llvm.vector.extract.nxv8bf16.nxv16bf16(<vscale x 16 x bfloat> [[VAL:%.*]], i64 0)
+// CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 8 x bfloat> @llvm.vector.extract.nxv8bf16.nxv16bf16(<vscale x 16 x bfloat> [[VAL]], i64 8)
+// CHECK-NEXT:    tail call void @llvm.aarch64.sme.write.ver.vg2.nxv8bf16(i32 1, i32 [[BASE:%.*]], <vscale x 8 x bfloat> [[TMP0]], <vscale x 8 x bfloat> [[TMP1]])
+// CHECK-NEXT:    ret void
+//
+// CPP-CHECK-LABEL: @_Z30test_svwrite_ver_za16_bf16_vg2j14svbfloat16x2_t(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 8 x bfloat> @llvm.vector.extract.nxv8bf16.nxv16bf16(<vscale x 16 x bfloat> [[VAL:%.*]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 8 x bfloat> @llvm.vector.extract.nxv8bf16.nxv16bf16(<vscale x 16 x bfloat> [[VAL]], i64 8)
+// CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.write.ver.vg2.nxv8bf16(i32 1, i32 [[BASE:%.*]], <vscale x 8 x bfloat> [[TMP0]], <vscale x 8 x bfloat> [[TMP1]])
+// CPP-CHECK-NEXT:    ret void
+//
+void test_svwrite_ver_za16_bf16_vg2(uint32_t base, svbfloat16x2_t val) {
+  SVE_ACLE_FUNC(svwrite_ver_za16,_bf16,_vg2,)(1, base, val);
+}
+
+// CHECK-LABEL: @test_svwrite_ver_za16_f16_vg2(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 8 x half> @llvm.vector.extract.nxv8f16.nxv16f16(<vscale x 16 x half> [[VAL:%.*]], i64 0)
+// CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 8 x half> @llvm.vector.extract.nxv8f16.nxv16f16(<vscale x 16 x half> [[VAL]], i64 8)
+// CHECK-NEXT:    tail call void @llvm.aarch64.sme.write.ver.vg2.nxv8f16(i32 1, i32 [[BASE:%.*]], <vscale x 8 x half> [[TMP0]], <vscale x 8 x half> [[TMP1]])
+// CHECK-NEXT:    ret void
+//
+// CPP-CHECK-LABEL: @_Z29test_svwrite_ver_za16_f16_vg2j13svfloat16x2_t(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 8 x half> @llvm.vector.extract.nxv8f16.nxv16f16(<vscale x 16 x half> [[VAL:%.*]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 8 x half> @llvm.vector.extract.nxv8f16.nxv16f16(<vscale x 16 x half> [[VAL]], i64 8)
+// CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.write.ver.vg2.nxv8f16(i32 1, i32 [[BASE:%.*]], <vscale x 8 x half> [[TMP0]], <vscale x 8 x half> [[TMP1]])
+// CPP-CHECK-NEXT:    ret void
+//
+void test_svwrite_ver_za16_f16_vg2(uint32_t base, svfloat16x2_t val) {
+  SVE_ACLE_FUNC(svwrite_ver_za16,_f16,_vg2,)(1, base, val);
+}
+
+// CHECK-LABEL: @test_svwrite_ver_za16_s16_vg2(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv16i16(<vscale x 16 x i16> [[VAL:%.*]], i64 0)
+// CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv16i16(<vscale x 16 x i16> [[VAL]], i64 8)
+// CHECK-NEXT:    tail call void @llvm.aarch64.sme.write.ver.vg2.nxv8i16(i32 1, i32 [[BASE:%.*]], <vscale x 8 x i16> [[TMP0]], <vscale x 8 x i16> [[TMP1]])
+// CHECK-NEXT:    ret void
+//
+// CPP-CHECK-LABEL: @_Z29test_svwrite_ver_za16_s16_vg2j11svint16x2_t(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv16i16(<vscale x 16 x i16> [[VAL:%.*]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv16i16(<vscale x 16 x i16> [[VAL]], i64 8)
+// CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.write.ver.vg2.nxv8i16(i32 1, i32 [[BASE:%.*]], <vscale x 8 x i16> [[TMP0]], <vscale x 8 x i16> [[TMP1]])
+// CPP-CHECK-NEXT:    ret void
+//
+void test_svwrite_ver_za16_s16_vg2(uint32_t base, svint16x2_t val) {
+  SVE_ACLE_FUNC(svwrite_ver_za16,_s16,_vg2,)(1, base, val);
+}
+
+// CHECK-LABEL: @test_svwrite_hor_za16_u16_vg4(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[VAL:%.*]], i64 0)
+// CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[VAL]], i64 8)
+// CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[VAL]], i64 16)
+// CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[VAL]], i64 24)
+// CHECK-NEXT:    tail call void @llvm.aarch64.sme.write.hor.vg4.nxv8i16(i32 1, i32 [[BASE:%.*]], <vscale x 8 x i16> [[TMP0]], <vscale x 8 x i16> [[TMP1]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP3]])
+// CHECK-NEXT:    ret void
+//
+// CPP-CHECK-LABEL: @_Z29test_svwrite_hor_za16_u16_vg4j12svuint16x4_t(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[VAL:%.*]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[VAL]], i64 8)
+// CPP-CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[VAL]], i64 16)
+// CPP-CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[VAL]], i64 24)
+// CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.write.hor.vg4.nxv8i16(i32 1, i32 [[BASE:%.*]], <vscale x 8 x i16> [[TMP0]], <vscale x 8 x i16> [[TMP1]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP3]])
+// CPP-CHECK-NEXT:    ret void
+//
+void test_svwrite_hor_za16_u16_vg4(uint32_t base, svuint16x4_t val) {
+  SVE_ACLE_FUNC(svwrite_hor_za16,_u16,_vg4,)(1, base, val);
+}
+
+// CHECK-LABEL: @test_svwrite_hor_za16_bf16_vg4(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 8 x bfloat> @llvm.vector.extract.nxv8bf16.nxv32bf16(<vscale x 32 x bfloat> [[VAL:%.*]], i64 0)
+// CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 8 x bfloat> @llvm.vector.extract.nxv8bf16.nxv32bf16(<vscale x 32 x bfloat> [[VAL]], i64 8)
+// CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 8 x bfloat> @llvm.vector.extract.nxv8bf16.nxv32bf16(<vscale x 32 x bfloat> [[VAL]], i64 16)
+// CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 8 x bfloat> @llvm.vector.extract.nxv8bf16.nxv32bf16(<vscale x 32 x bfloat> [[VAL]], i64 24)
+// CHECK-NEXT:    tail call void @llvm.aarch64.sme.write.hor.vg4.nxv8bf16(i32 1, i32 [[BASE:%.*]], <vscale x 8 x bfloat> [[TMP0]], <vscale x 8 x bfloat> [[TMP1]], <vscale x 8 x bfloat> [[TMP2]], <vscale x 8 x bfloat> [[TMP3]])
+// CHECK-NEXT:    ret void
+//
+// CPP-CHECK-LABEL: @_Z30test_svwrite_hor_za16_bf16_vg4j14svbfloat16x4_t(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 8 x bfloat> @llvm.vector.extract.nxv8bf16.nxv32bf16(<vscale x 32 x bfloat> [[VAL:%.*]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 8 x bfloat> @llvm.vector.extract.nxv8bf16.nxv32bf16(<vscale x 32 x bfloat> [[VAL]], i64 8)
+// CPP-CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 8 x bfloat> @llvm.vector.extract.nxv8bf16.nxv32bf16(<vscale x 32 x bfloat> [[VAL]], i64 16)
+// CPP-CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 8 x bfloat> @llvm.vector.extract.nxv8bf16.nxv32bf16(<vscale x 32 x bfloat> [[VAL]], i64 24)
+// CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.write.hor.vg4.nxv8bf16(i32 1, i32 [[BASE:%.*]], <vscale x 8 x bfloat> [[TMP0]], <vscale x 8 x bfloat> [[TMP1]], <vscale x 8 x bfloat> [[TMP2]], <vscale x 8 x bfloat> [[TMP3]])
+// CPP-CHECK-NEXT:    ret void
+//
+void test_svwrite_hor_za16_bf16_vg4(uint32_t base, svbfloat16x4_t val) {
+  SVE_ACLE_FUNC(svwrite_hor_za16,_bf16,_vg4,)(1, base, val);
+}
+
+// CHECK-LABEL: @test_svwrite_hor_za16_f16_vg4(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 8 x half> @llvm.vector.extract.nxv8f16.nxv32f16(<vscale x 32 x half> [[VAL:%.*]], i64 0)
+// CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 8 x half> @llvm.vector.extract.nxv8f16.nxv32f16(<vscale x 32 x half> [[VAL]], i64 8)
+// CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 8 x half> @llvm.vector.extract.nxv8f16.nxv32f16(<vscale x 32 x half> [[VAL]], i64 16)
+// CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 8 x half> @llvm.vector.extract.nxv8f16.nxv32f16(<vscale x 32 x half> [[VAL]], i64 24)
+// CHECK-NEXT:    tail call void @llvm.aarch64.sme.write.hor.vg4.nxv8f16(i32 1, i32 [[BASE:%.*]], <vscale x 8 x half> [[TMP0]], <vscale x 8 x half> [[TMP1]], <vscale x 8 x half> [[TMP2]], <vscale x 8 x half> [[TMP3]])
+// CHECK-NEXT:    ret void
+//
+// CPP-CHECK-LABEL: @_Z29test_svwrite_hor_za16_f16_vg4j13svfloat16x4_t(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 8 x half> @llvm.vector.extract.nxv8f16.nxv32f16(<vscale x 32 x half> [[VAL:%.*]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 8 x half> @llvm.vector.extract.nxv8f16.nxv32f16(<vscale x 32 x half> [[VAL]], i64 8)
+// CPP-CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 8 x half> @llvm.vector.extract.nxv8f16.nxv32f16(<vscale x 32 x half> [[VAL]], i64 16)
+// CPP-CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 8 x half> @llvm.vector.extract.nxv8f16.nxv32f16(<vscale x 32 x half> [[VAL]], i64 24)
+// CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.write.hor.vg4.nxv8f16(i32 1, i32 [[BASE:%.*]], <vscale x 8 x half> [[TMP0]], <vscale x 8 x half> [[TMP1]], <vscale x 8 x half> [[TMP2]], <vscale x 8 x half> [[TMP3]])
+// CPP-CHECK-NEXT:    ret void
+//
+void test_svwrite_hor_za16_f16_vg4(uint32_t base, svfloat16x4_t val) {
+  SVE_ACLE_FUNC(svwrite_hor_za16,_f16,_vg4,)(1, base, val);
+}
+
+// CHECK-LABEL: @test_svwrite_hor_za16_s16_vg4(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[VAL:%.*]], i64 0)
+// CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[VAL]], i64 8)
+// CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[VAL]], i64 16)
+// CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[VAL]], i64 24)
+// CHECK-NEXT:    tail call void @llvm.aarch64.sme.write.hor.vg4.nxv8i16(i32 1, i32 [[BASE:%.*]], <vscale x 8 x i16> [[TMP0]], <vscale x 8 x i16> [[TMP1]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP3]])
+// CHECK-NEXT:    ret void
+//
+// CPP-CHECK-LABEL: @_Z29test_svwrite_hor_za16_s16_vg4j11svint16x4_t(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[VAL:%.*]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[VAL]], i64 8)
+// CPP-CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[VAL]], i64 16)
+// CPP-CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[VAL]], i64 24)
+// CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.write.hor.vg4.nxv8i16(i32 1, i32 [[BASE:%.*]], <vscale x 8 x i16> [[TMP0]], <vscale x 8 x i16> [[TMP1]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP3]])
+// CPP-CHECK-NEXT:    ret void
+//
+void test_svwrite_hor_za16_s16_vg4(uint32_t base, svint16x4_t val) {
+  SVE_ACLE_FUNC(svwrite_hor_za16,_s16,_vg4,)(1, base, val);
+}
+
+// CHECK-LABEL: @test_svwrite_ver_za16_u16_vg4(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[VAL:%.*]], i64 0)
+// CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[VAL]], i64 8)
+// CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[VAL]], i64 16)
+// CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[VAL]], i64 24)
+// CHECK-NEXT:    tail call void @llvm.aarch64.sme.write.ver.vg4.nxv8i16(i32 1, i32 [[BASE:%.*]], <vscale x 8 x i16> [[TMP0]], <vscale x 8 x i16> [[TMP1]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP3]])
+// CHECK-NEXT:    ret void
+//
+// CPP-CHECK-LABEL: @_Z29test_svwrite_ver_za16_u16_vg4j12svuint16x4_t(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[VAL:%.*]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[VAL]], i64 8)
+// CPP-CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[VAL]], i64 16)
+// CPP-CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[VAL]], i64 24)
+// CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.write.ver.vg4.nxv8i16(i32 1, i32 [[BASE:%.*]], <vscale x 8 x i16> [[TMP0]], <vscale x 8 x i16> [[TMP1]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP3]])
+// CPP-CHECK-NEXT:    ret void
+//
+void test_svwrite_ver_za16_u16_vg4(uint32_t base, svuint16x4_t val) {
+  SVE_ACLE_FUNC(svwrite_ver_za16,_u16,_vg4,)(1, base, val);
+}
+
+// CHECK-LABEL: @test_svwrite_ver_za16_bf16_vg4(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 8 x bfloat> @llvm.vector.extract.nxv8bf16.nxv32bf16(<vscale x 32 x bfloat> [[VAL:%.*]], i64 0)
+// CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 8 x bfloat> @llvm.vector.extract.nxv8bf16.nxv32bf16(<vscale x 32 x bfloat> [[VAL]], i64 8)
+// CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 8 x bfloat> @llvm.vector.extract.nxv8bf16.nxv32bf16(<vscale x 32 x bfloat> [[VAL]], i64 16)
+// CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 8 x bfloat> @llvm.vector.extract.nxv8bf16.nxv32bf16(<vscale x 32 x bfloat> [[VAL]], i64 24)
+// CHECK-NEXT:    tail call void @llvm.aarch64.sme.write.ver.vg4.nxv8bf16(i32 1, i32 [[BASE:%.*]], <vscale x 8 x bfloat> [[TMP0]], <vscale x 8 x bfloat> [[TMP1]], <vscale x 8 x bfloat> [[TMP2]], <vscale x 8 x bfloat> [[TMP3]])
+// CHECK-NEXT:    ret void
+//
+// CPP-CHECK-LABEL: @_Z30test_svwrite_ver_za16_bf16_vg4j14svbfloat16x4_t(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 8 x bfloat> @llvm.vector.extract.nxv8bf16.nxv32bf16(<vscale x 32 x bfloat> [[VAL:%.*]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 8 x bfloat> @llvm.vector.extract.nxv8bf16.nxv32bf16(<vscale x 32 x bfloat> [[VAL]], i64 8)
+// CPP-CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 8 x bfloat> @llvm.vector.extract.nxv8bf16.nxv32bf16(<vscale x 32 x bfloat> [[VAL]], i64 16)
+// CPP-CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 8 x bfloat> @llvm.vector.extract.nxv8bf16.nxv32bf16(<vscale x 32 x bfloat> [[VAL]], i64 24)
+// CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.write.ver.vg4.nxv8bf16(i32 1, i32 [[BASE:%.*]], <vscale x 8 x bfloat> [[TMP0]], <vscale x 8 x bfloat> [[TMP1]], <vscale x 8 x bfloat> [[TMP2]], <vscale x 8 x bfloat> [[TMP3]])
+// CPP-CHECK-NEXT:    ret void
+//
+void test_svwrite_ver_za16_bf16_vg4(uint32_t base, svbfloat16x4_t val) {
+  SVE_ACLE_FUNC(svwrite_ver_za16,_bf16,_vg4,)(1, base, val);
+}
+
+// CHECK-LABEL: @test_svwrite_ver_za16_f16_vg4(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 8 x half> @llvm.vector.extract.nxv8f16.nxv32f16(<vscale x 32 x half> [[VAL:%.*]], i64 0)
+// CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 8 x half> @llvm.vector.extract.nxv8f16.nxv32f16(<vscale x 32 x half> [[VAL]], i64 8)
+// CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 8 x half> @llvm.vector.extract.nxv8f16.nxv32f16(<vscale x 32 x half> [[VAL]], i64 16)
+// CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 8 x half> @llvm.vector.extract.nxv8f16.nxv32f16(<vscale x 32 x half> [[VAL]], i64 24)
+// CHECK-NEXT:    tail call void @llvm.aarch64.sme.write.ver.vg4.nxv8f16(i32 1, i32 [[BASE:%.*]], <vscale x 8 x half> [[TMP0]], <vscale x 8 x half> [[TMP1]], <vscale x 8 x half> [[TMP2]], <vscale x 8 x half> [[TMP3]])
+// CHECK-NEXT:    ret void
+//
+// CPP-CHECK-LABEL: @_Z29test_svwrite_ver_za16_f16_vg4j13svfloat16x4_t(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 8 x half> @llvm.vector.extract.nxv8f16.nxv32f16(<vscale x 32 x half> [[VAL:%.*]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 8 x half> @llvm.vector.extract.nxv8f16.nxv32f16(<vscale x 32 x half> [[VAL]], i64 8)
+// CPP-CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 8 x half> @llvm.vector.extract.nxv8f16.nxv32f16(<vscale x 32 x half> [[VAL]], i64 16)
+// CPP-CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 8 x half> @llvm.vector.extract.nxv8f16.nxv32f16(<vscale x 32 x half> [[VAL]], i64 24)
+// CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.write.ver.vg4.nxv8f16(i32 1, i32 [[BASE:%.*]], <vscale x 8 x half> [[TMP0]], <vscale x 8 x half> [[TMP1]], <vscale x 8 x half> [[TMP2]], <vscale x 8 x half> [[TMP3]])
+// CPP-CHECK-NEXT:    ret void
+//
+void test_svwrite_ver_za16_f16_vg4(uint32_t base, svfloat16x4_t val) {
+  SVE_ACLE_FUNC(svwrite_ver_za16,_f16,_vg4,)(1, base, val);
+}
+
+// CHECK-LABEL: @test_svwrite_ver_za16_s16_vg4(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[VAL:%.*]], i64 0)
+// CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[VAL]], i64 8)
+// CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[VAL]], i64 16)
+// CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[VAL]], i64 24)
+// CHECK-NEXT:    tail call void @llvm.aarch64.sme.write.ver.vg4.nxv8i16(i32 1, i32 [[BASE:%.*]], <vscale x 8 x i16> [[TMP0]], <vscale x 8 x i16> [[TMP1]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP3]])
+// CHECK-NEXT:    ret void
+//
+// CPP-CHECK-LABEL: @_Z29test_svwrite_ver_za16_s16_vg4j11svint16x4_t(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[VAL:%.*]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[VAL]], i64 8)
+// CPP-CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[VAL]], i64 16)
+// CPP-CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[VAL]], i64 24)
+// CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.write.ver.vg4.nxv8i16(i32 1, i32 [[BASE:%.*]], <vscale x 8 x i16> [[TMP0]], <vscale x 8 x i16> [[TMP1]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP3]])
+// CPP-CHECK-NEXT:    ret void
+//
+void test_svwrite_ver_za16_s16_vg4(uint32_t base, svint16x4_t val) {
+  SVE_ACLE_FUNC(svwrite_ver_za16,_s16,_vg4,)(1, base, val);
+}
+
+// CHECK-LABEL: @test_svwrite_hor_za32_u32_vg2(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 4 x i32> @llvm.vector.extract.nxv4i32.nxv8i32(<vscale x 8 x i32> [[VAL:%.*]], i64 0)
+// CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 4 x i32> @llvm.vector.extract.nxv4i32.nxv8i32(<vscale x 8 x i32> [[VAL]], i64 4)
+// CHECK-NEXT:    tail call void @llvm.aarch64.sme.write.hor.vg2.nxv4i32(i32 3, i32 [[BASE:%.*]], <vscale x 4 x i32> [[TMP0]], <vscale x 4 x i32> [[TMP1]])
+// CHECK-NEXT:    ret void
+//
+// CPP-CHECK-LABEL: @_Z29test_svwrite_hor_za32_u32_vg2j12svuint32x2_t(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 4 x i32> @llvm.vector.extract.nxv4i32.nxv8i32(<vscale x 8 x i32> [[VAL:%.*]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 4 x i32> @llvm.vector.extract.nxv4i32.nxv8i32(<vscale x 8 x i32> [[VAL]], i64 4)
+// CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.write.hor.vg2.nxv4i32(i32 3, i32 [[BASE:%.*]], <vscale x 4 x i32> [[TMP0]], <vscale x 4 x i32> [[TMP1]])
+// CPP-CHECK-NEXT:    ret void
+//
+void test_svwrite_hor_za32_u32_vg2(uint32_t base, svuint32x2_t val) {
+  SVE_ACLE_FUNC(svwrite_hor_za32,_u32,_vg2,)(3, base, val);
+}
+
+// CHECK-LABEL: @test_svwrite_hor_za32_f32_vg2(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 4 x float> @llvm.vector.extract.nxv4f32.nxv8f32(<vscale x 8 x float> [[VAL:%.*]], i64 0)
+// CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 4 x float> @llvm.vector.extract.nxv4f32.nxv8f32(<vscale x 8 x float> [[VAL]], i64 4)
+// CHECK-NEXT:    tail call void @llvm.aarch64.sme.write.hor.vg2.nxv4f32(i32 3, i32 [[BASE:%.*]], <vscale x 4 x float> [[TMP0]], <vscale x 4 x float> [[TMP1]])
+// CHECK-NEXT:    ret void
+//
+// CPP-CHECK-LABEL: @_Z29test_svwrite_hor_za32_f32_vg2j13svfloat32x2_t(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 4 x float> @llvm.vector.extract.nxv4f32.nxv8f32(<vscale x 8 x float> [[VAL:%.*]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 4 x float> @llvm.vector.extract.nxv4f32.nxv8f32(<vscale x 8 x float> [[VAL]], i64 4)
+// CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.write.hor.vg2.nxv4f32(i32 3, i32 [[BASE:%.*]], <vscale x 4 x float> [[TMP0]], <vscale x 4 x float> [[TMP1]])
+// CPP-CHECK-NEXT:    ret void
+//
+void test_svwrite_hor_za32_f32_vg2(uint32_t base, svfloat32x2_t val) {
+  SVE_ACLE_FUNC(svwrite_hor_za32,_f32,_vg2,)(3, base, val);
+}
+
+// CHECK-LABEL: @test_svwrite_hor_za32_s32_vg2(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 4 x i32> @llvm.vector.extract.nxv4i32.nxv8i32(<vscale x 8 x i32> [[VAL:%.*]], i64 0)
+// CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 4 x i32> @llvm.vector.extract.nxv4i32.nxv8i32(<vscale x 8 x i32> [[VAL]], i64 4)
+// CHECK-NEXT:    tail call void @llvm.aarch64.sme.write.hor.vg2.nxv4i32(i32 3, i32 [[BASE:%.*]], <vscale x 4 x i32> [[TMP0]], <vscale x 4 x i32> [[TMP1]])
+// CHECK-NEXT:    ret void
+//
+// CPP-CHECK-LABEL: @_Z29test_svwrite_hor_za32_s32_vg2j11svint32x2_t(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 4 x i32> @llvm.vector.extract.nxv4i32.nxv8i32(<vscale x 8 x i32> [[VAL:%.*]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 4 x i32> @llvm.vector.extract.nxv4i32.nxv8i32(<vscale x 8 x i32> [[VAL]], i64 4)
+// CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.write.hor.vg2.nxv4i32(i32 3, i32 [[BASE:%.*]], <vscale x 4 x i32> [[TMP0]], <vscale x 4 x i32> [[TMP1]])
+// CPP-CHECK-NEXT:    ret void
+//
+void test_svwrite_hor_za32_s32_vg2(uint32_t base, svint32x2_t val) {
+  SVE_ACLE_FUNC(svwrite_hor_za32,_s32,_vg2,)(3, base, val);
+}
+
+// CHECK-LABEL: @test_svwrite_ver_za32_u32_vg2(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 4 x i32> @llvm.vector.extract.nxv4i32.nxv8i32(<vscale x 8 x i32> [[VAL:%.*]], i64 0)
+// CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 4 x i32> @llvm.vector.extract.nxv4i32.nxv8i32(<vscale x 8 x i32> [[VAL]], i64 4)
+// CHECK-NEXT:    tail call void @llvm.aarch64.sme.write.ver.vg2.nxv4i32(i32 3, i32 [[BASE:%.*]], <vscale x 4 x i32> [[TMP0]], <vscale x 4 x i32> [[TMP1]])
+// CHECK-NEXT:    ret void
+//
+// CPP-CHECK-LABEL: @_Z29test_svwrite_ver_za32_u32_vg2j12svuint32x2_t(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 4 x i32> @llvm.vector.extract.nxv4i32.nxv8i32(<vscale x 8 x i32> [[VAL:%.*]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 4 x i32> @llvm.vector.extract.nxv4i32.nxv8i32(<vscale x 8 x i32> [[VAL]], i64 4)
+// CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.write.ver.vg2.nxv4i32(i32 3, i32 [[BASE:%.*]], <vscale x 4 x i32> [[TMP0]], <vscale x 4 x i32> [[TMP1]])
+// CPP-CHECK-NEXT:    ret void
+//
+void test_svwrite_ver_za32_u32_vg2(uint32_t base, svuint32x2_t val) {
+  SVE_ACLE_FUNC(svwrite_ver_za32,_u32,_vg2,)(3, base, val);
+}
+
+// CHECK-LABEL: @test_svwrite_ver_za32_f32_vg2(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 4 x float> @llvm.vector.extract.nxv4f32.nxv8f32(<vscale x 8 x float> [[VAL:%.*]], i64 0)
+// CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 4 x float> @llvm.vector.extract.nxv4f32.nxv8f32(<vscale x 8 x float> [[VAL]], i64 4)
+// CHECK-NEXT:    tail call void @llvm.aarch64.sme.write.ver.vg2.nxv4f32(i32 3, i32 [[BASE:%.*]], <vscale x 4 x float> [[TMP0]], <vscale x 4 x float> [[TMP1]])
+// CHECK-NEXT:    ret void
+//
+// CPP-CHECK-LABEL: @_Z29test_svwrite_ver_za32_f32_vg2j13svfloat32x2_t(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 4 x float> @llvm.vector.extract.nxv4f32.nxv8f32(<vscale x 8 x float> [[VAL:%.*]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 4 x float> @llvm.vector.extract.nxv4f32.nxv8f32(<vscale x 8 x float> [[VAL]], i64 4)
+// CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.write.ver.vg2.nxv4f32(i32 3, i32 [[BASE:%.*]], <vscale x 4 x float> [[TMP0]], <vscale x 4 x float> [[TMP1]])
+// CPP-CHECK-NEXT:    ret void
+//
+void test_svwrite_ver_za32_f32_vg2(uint32_t base, svfloat32x2_t val) {
+  SVE_ACLE_FUNC(svwrite_ver_za32,_f32,_vg2,)(3, base, val);
+}
+
+// CHECK-LABEL: @test_svwrite_ver_za32_s32_vg2(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 4 x i32> @llvm.vector.extract.nxv4i32.nxv8i32(<vscale x 8 x i32> [[VAL:%.*]], i64 0)
+// CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 4 x i32> @llvm.vector.extract.nxv4i32.nxv8i32(<vscale x 8 x i32> [[VAL]], i64 4)
+// CHECK-NEXT:    tail call void @llvm.aarch64.sme.write.ver.vg2.nxv4i32(i32 3, i32 [[BASE:%.*]], <vscale x 4 x i32> [[TMP0]], <vscale x 4 x i32> [[TMP1]])
+// CHECK-NEXT:    ret void
+//
+// CPP-CHECK-LABEL: @_Z29test_svwrite_ver_za32_s32_vg2j11svint32x2_t(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 4 x i32> @llvm.vector.extract.nxv4i32.nxv8i32(<vscale x 8 x i32> [[VAL:%.*]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 4 x i32> @llvm.vector.extract.nxv4i32.nxv8i32(<vscale x 8 x i32> [[VAL]], i64 4)
+// CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.write.ver.vg2.nxv4i32(i32 3, i32 [[BASE:%.*]], <vscale x 4 x i32> [[TMP0]], <vscale x 4 x i32> [[TMP1]])
+// CPP-CHECK-NEXT:    ret void
+//
+void test_svwrite_ver_za32_s32_vg2(uint32_t base, svint32x2_t val) {
+  SVE_ACLE_FUNC(svwrite_ver_za32,_s32,_vg2,)(3, base, val);
+}
+
+// CHECK-LABEL: @test_svwrite_hor_za32_u32_vg4(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 4 x i32> @llvm.vector.extract.nxv4i32.nxv16i32(<vscale x 16 x i32> [[VAL:%.*]], i64 0)
+// CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 4 x i32> @llvm.vector.extract.nxv4i32.nxv16i32(<vscale x 16 x i32> [[VAL]], i64 4)
+// CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 4 x i32> @llvm.vector.extract.nxv4i32.nxv16i32(<vscale x 16 x i32> [[VAL]], i64 8)
+// CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 4 x i32> @llvm.vector.extract.nxv4i32.nxv16i32(<vscale x 16 x i32> [[VAL]], i64 12)
+// CHECK-NEXT:    tail call void @llvm.aarch64.sme.write.hor.vg4.nxv4i32(i32 3, i32 [[BASE:%.*]], <vscale x 4 x i32> [[TMP0]], <vscale x 4 x i32> [[TMP1]], <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> [[TMP3]])
+// CHECK-NEXT:    ret void
+//
+// CPP-CHECK-LABEL: @_Z29test_svwrite_hor_za32_u32_vg4j12svuint32x4_t(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 4 x i32> @llvm.vector.extract.nxv4i32.nxv16i32(<vscale x 16 x i32> [[VAL:%.*]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 4 x i32> @llvm.vector.extract.nxv4i32.nxv16i32(<vscale x 16 x i32> [[VAL]], i64 4)
+// CPP-CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 4 x i32> @llvm.vector.extract.nxv4i32.nxv16i32(<vscale x 16 x i32> [[VAL]], i64 8)
+// CPP-CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 4 x i32> @llvm.vector.extract.nxv4i32.nxv16i32(<vscale x 16 x i32> [[VAL]], i64 12)
+// CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.write.hor.vg4.nxv4i32(i32 3, i32 [[BASE:%.*]], <vscale x 4 x i32> [[TMP0]], <vscale x 4 x i32> [[TMP1]], <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> [[TMP3]])
+// CPP-CHECK-NEXT:    ret void
+//
+void test_svwrite_hor_za32_u32_vg4(uint32_t base, svuint32x4_t val) {
+  SVE_ACLE_FUNC(svwrite_hor_za32,_u32,_vg4,)(3, base, val);
+}
+
+// CHECK-LABEL: @test_svwrite_hor_za32_f32_vg4(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 4 x float> @llvm.vector.extract.nxv4f32.nxv16f32(<vscale x 16 x float> [[VAL:%.*]], i64 0)
+// CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 4 x float> @llvm.vector.extract.nxv4f32.nxv16f32(<vscale x 16 x float> [[VAL]], i64 4)
+// CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 4 x float> @llvm.vector.extract.nxv4f32.nxv16f32(<vscale x 16 x float> [[VAL]], i64 8)
+// CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 4 x float> @llvm.vector.extract.nxv4f32.nxv16f32(<vscale x 16 x float> [[VAL]], i64 12)
+// CHECK-NEXT:    tail call void @llvm.aarch64.sme.write.hor.vg4.nxv4f32(i32 3, i32 [[BASE:%.*]], <vscale x 4 x float> [[TMP0]], <vscale x 4 x float> [[TMP1]], <vscale x 4 x float> [[TMP2]], <vscale x 4 x float> [[TMP3]])
+// CHECK-NEXT:    ret void
+//
+// CPP-CHECK-LABEL: @_Z29test_svwrite_hor_za32_f32_vg4j13svfloat32x4_t(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 4 x float> @llvm.vector.extract.nxv4f32.nxv16f32(<vscale x 16 x float> [[VAL:%.*]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 4 x float> @llvm.vector.extract.nxv4f32.nxv16f32(<vscale x 16 x float> [[VAL]], i64 4)
+// CPP-CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 4 x float> @llvm.vector.extract.nxv4f32.nxv16f32(<vscale x 16 x float> [[VAL]], i64 8)
+// CPP-CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 4 x float> @llvm.vector.extract.nxv4f32.nxv16f32(<vscale x 16 x float> [[VAL]], i64 12)
+// CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.write.hor.vg4.nxv4f32(i32 3, i32 [[BASE:%.*]], <vscale x 4 x float> [[TMP0]], <vscale x 4 x float> [[TMP1]], <vscale x 4 x float> [[TMP2]], <vscale x 4 x float> [[TMP3]])
+// CPP-CHECK-NEXT:    ret void
+//
+void test_svwrite_hor_za32_f32_vg4(uint32_t base, svfloat32x4_t val) {
+  SVE_ACLE_FUNC(svwrite_hor_za32,_f32,_vg4,)(3, base, val);
+}
+
+// CHECK-LABEL: @test_svwrite_hor_za32_s32_vg4(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 4 x i32> @llvm.vector.extract.nxv4i32.nxv16i32(<vscale x 16 x i32> [[VAL:%.*]], i64 0)
+// CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 4 x i32> @llvm.vector.extract.nxv4i32.nxv16i32(<vscale x 16 x i32> [[VAL]], i64 4)
+// CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 4 x i32> @llvm.vector.extract.nxv4i32.nxv16i32(<vscale x 16 x i32> [[VAL]], i64 8)
+// CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 4 x i32> @llvm.vector.extract.nxv4i32.nxv16i32(<vscale x 16 x i32> [[VAL]], i64 12)
+// CHECK-NEXT:    tail call void @llvm.aarch64.sme.write.hor.vg4.nxv4i32(i32 3, i32 [[BASE:%.*]], <vscale x 4 x i32> [[TMP0]], <vscale x 4 x i32> [[TMP1]], <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> [[TMP3]])
+// CHECK-NEXT:    ret void
+//
+// CPP-CHECK-LABEL: @_Z29test_svwrite_hor_za32_s32_vg4j11svint32x4_t(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 4 x i32> @llvm.vector.extract.nxv4i32.nxv16i32(<vscale x 16 x i32> [[VAL:%.*]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 4 x i32> @llvm.vector.extract.nxv4i32.nxv16i32(<vscale x 16 x i32> [[VAL]], i64 4)
+// CPP-CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 4 x i32> @llvm.vector.extract.nxv4i32.nxv16i32(<vscale x 16 x i32> [[VAL]], i64 8)
+// CPP-CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 4 x i32> @llvm.vector.extract.nxv4i32.nxv16i32(<vscale x 16 x i32> [[VAL]], i64 12)
+// CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.write.hor.vg4.nxv4i32(i32 3, i32 [[BASE:%.*]], <vscale x 4 x i32> [[TMP0]], <vscale x 4 x i32> [[TMP1]], <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> [[TMP3]])
+// CPP-CHECK-NEXT:    ret void
+//
+void test_svwrite_hor_za32_s32_vg4(uint32_t base, svint32x4_t val) {
+  SVE_ACLE_FUNC(svwrite_hor_za32,_s32,_vg4,)(3, base, val);
+}
+
+// CHECK-LABEL: @test_svwrite_ver_za32_u32_vg4(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 4 x i32> @llvm.vector.extract.nxv4i32.nxv16i32(<vscale x 16 x i32> [[VAL:%.*]], i64 0)
+// CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 4 x i32> @llvm.vector.extract.nxv4i32.nxv16i32(<vscale x 16 x i32> [[VAL]], i64 4)
+// CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 4 x i32> @llvm.vector.extract.nxv4i32.nxv16i32(<vscale x 16 x i32> [[VAL]], i64 8)
+// CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 4 x i32> @llvm.vector.extract.nxv4i32.nxv16i32(<vscale x 16 x i32> [[VAL]], i64 12)
+// CHECK-NEXT:    tail call void @llvm.aarch64.sme.write.ver.vg4.nxv4i32(i32 3, i32 [[BASE:%.*]], <vscale x 4 x i32> [[TMP0]], <vscale x 4 x i32> [[TMP1]], <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> [[TMP3]])
+// CHECK-NEXT:    ret void
+//
+// CPP-CHECK-LABEL: @_Z29test_svwrite_ver_za32_u32_vg4j12svuint32x4_t(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 4 x i32> @llvm.vector.extract.nxv4i32.nxv16i32(<vscale x 16 x i32> [[VAL:%.*]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 4 x i32> @llvm.vector.extract.nxv4i32.nxv16i32(<vscale x 16 x i32> [[VAL]], i64 4)
+// CPP-CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 4 x i32> @llvm.vector.extract.nxv4i32.nxv16i32(<vscale x 16 x i32> [[VAL]], i64 8)
+// CPP-CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 4 x i32> @llvm.vector.extract.nxv4i32.nxv16i32(<vscale x 16 x i32> [[VAL]], i64 12)
+// CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.write.ver.vg4.nxv4i32(i32 3, i32 [[BASE:%.*]], <vscale x 4 x i32> [[TMP0]], <vscale x 4 x i32> [[TMP1]], <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> [[TMP3]])
+// CPP-CHECK-NEXT:    ret void
+//
+void test_svwrite_ver_za32_u32_vg4(uint32_t base, svuint32x4_t val) {
+  SVE_ACLE_FUNC(svwrite_ver_za32,_u32,_vg4,)(3, base, val);
+}
+
+// CHECK-LABEL: @test_svwrite_ver_za32_f32_vg4(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 4 x float> @llvm.vector.extract.nxv4f32.nxv16f32(<vscale x 16 x float> [[VAL:%.*]], i64 0)
+// CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 4 x float> @llvm.vector.extract.nxv4f32.nxv16f32(<vscale x 16 x float> [[VAL]], i64 4)
+// CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 4 x float> @llvm.vector.extract.nxv4f32.nxv16f32(<vscale x 16 x float> [[VAL]], i64 8)
+// CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 4 x float> @llvm.vector.extract.nxv4f32.nxv16f32(<vscale x 16 x float> [[VAL]], i64 12)
+// CHECK-NEXT:    tail call void @llvm.aarch64.sme.write.ver.vg4.nxv4f32(i32 3, i32 [[BASE:%.*]], <vscale x 4 x float> [[TMP0]], <vscale x 4 x float> [[TMP1]], <vscale x 4 x float> [[TMP2]], <vscale x 4 x float> [[TMP3]])
+// CHECK-NEXT:    ret void
+//
+// CPP-CHECK-LABEL: @_Z29test_svwrite_ver_za32_f32_vg4j13svfloat32x4_t(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 4 x float> @llvm.vector.extract.nxv4f32.nxv16f32(<vscale x 16 x float> [[VAL:%.*]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 4 x float> @llvm.vector.extract.nxv4f32.nxv16f32(<vscale x 16 x float> [[VAL]], i64 4)
+// CPP-CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 4 x float> @llvm.vector.extract.nxv4f32.nxv16f32(<vscale x 16 x float> [[VAL]], i64 8)
+// CPP-CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 4 x float> @llvm.vector.extract.nxv4f32.nxv16f32(<vscale x 16 x float> [[VAL]], i64 12)
+// CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.write.ver.vg4.nxv4f32(i32 3, i32 [[BASE:%.*]], <vscale x 4 x float> [[TMP0]], <vscale x 4 x float> [[TMP1]], <vscale x 4 x float> [[TMP2]], <vscale x 4 x float> [[TMP3]])
+// CPP-CHECK-NEXT:    ret void
+//
+void test_svwrite_ver_za32_f32_vg4(uint32_t base, svfloat32x4_t val) {
+  SVE_ACLE_FUNC(svwrite_ver_za32,_f32,_vg4,)(3, base, val);
+}
+
+// CHECK-LABEL: @test_svwrite_ver_za32_s32_vg4(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 4 x i32> @llvm.vector.extract.nxv4i32.nxv16i32(<vscale x 16 x i32> [[VAL:%.*]], i64 0)
+// CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 4 x i32> @llvm.vector.extract.nxv4i32.nxv16i32(<vscale x 16 x i32> [[VAL]], i64 4)
+// CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 4 x i32> @llvm.vector.extract.nxv4i32.nxv16i32(<vscale x 16 x i32> [[VAL]], i64 8)
+// CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 4 x i32> @llvm.vector.extract.nxv4i32.nxv16i32(<vscale x 16 x i32> [[VAL]], i64 12)
+// CHECK-NEXT:    tail call void @llvm.aarch64.sme.write.ver.vg4.nxv4i32(i32 3, i32 [[BASE:%.*]], <vscale x 4 x i32> [[TMP0]], <vscale x 4 x i32> [[TMP1]], <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> [[TMP3]])
+// CHECK-NEXT:    ret void
+//
+// CPP-CHECK-LABEL: @_Z29test_svwrite_ver_za32_s32_vg4j11svint32x4_t(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 4 x i32> @llvm.vector.extract.nxv4i32.nxv16i32(<vscale x 16 x i32> [[VAL:%.*]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 4 x i32> @llvm.vector.extract.nxv4i32.nxv16i32(<vscale x 16 x i32> [[VAL]], i64 4)
+// CPP-CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 4 x i32> @llvm.vector.extract.nxv4i32.nxv16i32(<vscale x 16 x i32> [[VAL]], i64 8)
+// CPP-CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 4 x i32> @llvm.vector.extract.nxv4i32.nxv16i32(<vscale x 16 x i32> [[VAL]], i64 12)
+// CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.write.ver.vg4.nxv4i32(i32 3, i32 [[BASE:%.*]], <vscale x 4 x i32> [[TMP0]], <vscale x 4 x i32> [[TMP1]], <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> [[TMP3]])
+// CPP-CHECK-NEXT:    ret void
+//
+void test_svwrite_ver_za32_s32_vg4(uint32_t base, svint32x4_t val) {
+  SVE_ACLE_FUNC(svwrite_ver_za32,_s32,_vg4,)(3, base, val);
+}
+
+// CHECK-LABEL: @test_svwrite_hor_za64_u64_vg2(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 2 x i64> @llvm.vector.extract.nxv2i64.nxv4i64(<vscale x 4 x i64> [[VAL:%.*]], i64 0)
+// CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 2 x i64> @llvm.vector.extract.nxv2i64.nxv4i64(<vscale x 4 x i64> [[VAL]], i64 2)
+// CHECK-NEXT:    tail call void @llvm.aarch64.sme.write.hor.vg2.nxv2i64(i32 7, i32 [[BASE:%.*]], <vscale x 2 x i64> [[TMP0]], <vscale x 2 x i64> [[TMP1]])
+// CHECK-NEXT:    ret void
+//
+// CPP-CHECK-LABEL: @_Z29test_svwrite_hor_za64_u64_vg2j12svuint64x2_t(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 2 x i64> @llvm.vector.extract.nxv2i64.nxv4i64(<vscale x 4 x i64> [[VAL:%.*]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 2 x i64> @llvm.vector.extract.nxv2i64.nxv4i64(<vscale x 4 x i64> [[VAL]], i64 2)
+// CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.write.hor.vg2.nxv2i64(i32 7, i32 [[BASE:%.*]], <vscale x 2 x i64> [[TMP0]], <vscale x 2 x i64> [[TMP1]])
+// CPP-CHECK-NEXT:    ret void
+//
+void test_svwrite_hor_za64_u64_vg2(uint32_t base, svuint64x2_t val) {
+  SVE_ACLE_FUNC(svwrite_hor_za64,_u64,_vg2,)(7, base, val);
+}
+
+// CHECK-LABEL: @test_svwrite_hor_za64_f64_vg2(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 2 x double> @llvm.vector.extract.nxv2f64.nxv4f64(<vscale x 4 x double> [[VAL:%.*]], i64 0)
+// CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 2 x double> @llvm.vector.extract.nxv2f64.nxv4f64(<vscale x 4 x double> [[VAL]], i64 2)
+// CHECK-NEXT:    tail call void @llvm.aarch64.sme.write.hor.vg2.nxv2f64(i32 7, i32 [[BASE:%.*]], <vscale x 2 x double> [[TMP0]], <vscale x 2 x double> [[TMP1]])
+// CHECK-NEXT:    ret void
+//
+// CPP-CHECK-LABEL: @_Z29test_svwrite_hor_za64_f64_vg2j13svfloat64x2_t(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 2 x double> @llvm.vector.extract.nxv2f64.nxv4f64(<vscale x 4 x double> [[VAL:%.*]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 2 x double> @llvm.vector.extract.nxv2f64.nxv4f64(<vscale x 4 x double> [[VAL]], i64 2)
+// CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.write.hor.vg2.nxv2f64(i32 7, i32 [[BASE:%.*]], <vscale x 2 x double> [[TMP0]], <vscale x 2 x double> [[TMP1]])
+// CPP-CHECK-NEXT:    ret void
+//
+void test_svwrite_hor_za64_f64_vg2(uint32_t base, svfloat64x2_t val) {
+  SVE_ACLE_FUNC(svwrite_hor_za64,_f64,_vg2,)(7, base, val);
+}
+
+// CHECK-LABEL: @test_svwrite_hor_za64_s64_vg2(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 2 x i64> @llvm.vector.extract.nxv2i64.nxv4i64(<vscale x 4 x i64> [[VAL:%.*]], i64 0)
+// CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 2 x i64> @llvm.vector.extract.nxv2i64.nxv4i64(<vscale x 4 x i64> [[VAL]], i64 2)
+// CHECK-NEXT:    tail call void @llvm.aarch64.sme.write.hor.vg2.nxv2i64(i32 7, i32 [[BASE:%.*]], <vscale x 2 x i64> [[TMP0]], <vscale x 2 x i64> [[TMP1]])
+// CHECK-NEXT:    ret void
+//
+// CPP-CHECK-LABEL: @_Z29test_svwrite_hor_za64_s64_vg2j11svint64x2_t(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 2 x i64> @llvm.vector.extract.nxv2i64.nxv4i64(<vscale x 4 x i64> [[VAL:%.*]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 2 x i64> @llvm.vector.extract.nxv2i64.nxv4i64(<vscale x 4 x i64> [[VAL]], i64 2)
+// CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.write.hor.vg2.nxv2i64(i32 7, i32 [[BASE:%.*]], <vscale x 2 x i64> [[TMP0]], <vscale x 2 x i64> [[TMP1]])
+// CPP-CHECK-NEXT:    ret void
+//
+void test_svwrite_hor_za64_s64_vg2(uint32_t base, svint64x2_t val) {
+  SVE_ACLE_FUNC(svwrite_hor_za64,_s64,_vg2,)(7, base, val);
+}
+
+// CHECK-LABEL: @test_svwrite_ver_za64_u64_vg2(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 2 x i64> @llvm.vector.extract.nxv2i64.nxv4i64(<vscale x 4 x i64> [[VAL:%.*]], i64 0)
+// CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 2 x i64> @llvm.vector.extract.nxv2i64.nxv4i64(<vscale x 4 x i64> [[VAL]], i64 2)
+// CHECK-NEXT:    tail call void @llvm.aarch64.sme.write.ver.vg2.nxv2i64(i32 7, i32 [[BASE:%.*]], <vscale x 2 x i64> [[TMP0]], <vscale x 2 x i64> [[TMP1]])
+// CHECK-NEXT:    ret void
+//
+// CPP-CHECK-LABEL: @_Z29test_svwrite_ver_za64_u64_vg2j12svuint64x2_t(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 2 x i64> @llvm.vector.extract.nxv2i64.nxv4i64(<vscale x 4 x i64> [[VAL:%.*]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 2 x i64> @llvm.vector.extract.nxv2i64.nxv4i64(<vscale x 4 x i64> [[VAL]], i64 2)
+// CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.write.ver.vg2.nxv2i64(i32 7, i32 [[BASE:%.*]], <vscale x 2 x i64> [[TMP0]], <vscale x 2 x i64> [[TMP1]])
+// CPP-CHECK-NEXT:    ret void
+//
+void test_svwrite_ver_za64_u64_vg2(uint32_t base, svuint64x2_t val) {
+  SVE_ACLE_FUNC(svwrite_ver_za64,_u64,_vg2,)(7, base, val);
+}
+
+// CHECK-LABEL: @test_svwrite_ver_za64_f64_vg2(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 2 x double> @llvm.vector.extract.nxv2f64.nxv4f64(<vscale x 4 x double> [[VAL:%.*]], i64 0)
+// CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 2 x double> @llvm.vector.extract.nxv2f64.nxv4f64(<vscale x 4 x double> [[VAL]], i64 2)
+// CHECK-NEXT:    tail call void @llvm.aarch64.sme.write.ver.vg2.nxv2f64(i32 7, i32 [[BASE:%.*]], <vscale x 2 x double> [[TMP0]], <vscale x 2 x double> [[TMP1]])
+// CHECK-NEXT:    ret void
+//
+// CPP-CHECK-LABEL: @_Z29test_svwrite_ver_za64_f64_vg2j13svfloat64x2_t(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 2 x double> @llvm.vector.extract.nxv2f64.nxv4f64(<vscale x 4 x double> [[VAL:%.*]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 2 x double> @llvm.vector.extract.nxv2f64.nxv4f64(<vscale x 4 x double> [[VAL]], i64 2)
+// CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.write.ver.vg2.nxv2f64(i32 7, i32 [[BASE:%.*]], <vscale x 2 x double> [[TMP0]], <vscale x 2 x double> [[TMP1]])
+// CPP-CHECK-NEXT:    ret void
+//
+void test_svwrite_ver_za64_f64_vg2(uint32_t base, svfloat64x2_t val) {
+  SVE_ACLE_FUNC(svwrite_ver_za64,_f64,_vg2,)(7, base, val);
+}
+
+// CHECK-LABEL: @test_svwrite_ver_za64_s64_vg2(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 2 x i64> @llvm.vector.extract.nxv2i64.nxv4i64(<vscale x 4 x i64> [[VAL:%.*]], i64 0)
+// CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 2 x i64> @llvm.vector.extract.nxv2i64.nxv4i64(<vscale x 4 x i64> [[VAL]], i64 2)
+// CHECK-NEXT:    tail call void @llvm.aarch64.sme.write.ver.vg2.nxv2i64(i32 7, i32 [[BASE:%.*]], <vscale x 2 x i64> [[TMP0]], <vscale x 2 x i64> [[TMP1]])
+// CHECK-NEXT:    ret void
+//
+// CPP-CHECK-LABEL: @_Z29test_svwrite_ver_za64_s64_vg2j11svint64x2_t(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 2 x i64> @llvm.vector.extract.nxv2i64.nxv4i64(<vscale x 4 x i64> [[VAL:%.*]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 2 x i64> @llvm.vector.extract.nxv2i64.nxv4i64(<vscale x 4 x i64> [[VAL]], i64 2)
+// CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.write.ver.vg2.nxv2i64(i32 7, i32 [[BASE:%.*]], <vscale x 2 x i64> [[TMP0]], <vscale x 2 x i64> [[TMP1]])
+// CPP-CHECK-NEXT:    ret void
+//
+void test_svwrite_ver_za64_s64_vg2(uint32_t base, svint64x2_t val) {
+  SVE_ACLE_FUNC(svwrite_ver_za64,_s64,_vg2,)(7, base, val);
+}
+
+// CHECK-LABEL: @test_svwrite_hor_za64_u64_vg4(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 2 x i64> @llvm.vector.extract.nxv2i64.nxv8i64(<vscale x 8 x i64> [[VAL:%.*]], i64 0)
+// CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 2 x i64> @llvm.vector.extract.nxv2i64.nxv8i64(<vscale x 8 x i64> [[VAL]], i64 2)
+// CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 2 x i64> @llvm.vector.extract.nxv2i64.nxv8i64(<vscale x 8 x i64> [[VAL]], i64 4)
+// CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 2 x i64> @llvm.vector.extract.nxv2i64.nxv8i64(<vscale x 8 x i64> [[VAL]], i64 6)
+// CHECK-NEXT:    tail call void @llvm.aarch64.sme.write.hor.vg4.nxv2i64(i32 7, i32 [[BASE:%.*]], <vscale x 2 x i64> [[TMP0]], <vscale x 2 x i64> [[TMP1]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP3]])
+// CHECK-NEXT:    ret void
+//
+// CPP-CHECK-LABEL: @_Z29test_svwrite_hor_za64_u64_vg4j12svuint64x4_t(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 2 x i64> @llvm.vector.extract.nxv2i64.nxv8i64(<vscale x 8 x i64> [[VAL:%.*]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 2 x i64> @llvm.vector.extract.nxv2i64.nxv8i64(<vscale x 8 x i64> [[VAL]], i64 2)
+// CPP-CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 2 x i64> @llvm.vector.extract.nxv2i64.nxv8i64(<vscale x 8 x i64> [[VAL]], i64 4)
+// CPP-CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 2 x i64> @llvm.vector.extract.nxv2i64.nxv8i64(<vscale x 8 x i64> [[VAL]], i64 6)
+// CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.write.hor.vg4.nxv2i64(i32 7, i32 [[BASE:%.*]], <vscale x 2 x i64> [[TMP0]], <vscale x 2 x i64> [[TMP1]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP3]])
+// CPP-CHECK-NEXT:    ret void
+//
+void test_svwrite_hor_za64_u64_vg4(uint32_t base, svuint64x4_t val) {
+  SVE_ACLE_FUNC(svwrite_hor_za64,_u64,_vg4,)(7, base, val);
+}
+
+// CHECK-LABEL: @test_svwrite_ver_za64_u64_vg4(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 2 x i64> @llvm.vector.extract.nxv2i64.nxv8i64(<vscale x 8 x i64> [[VAL:%.*]], i64 0)
+// CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 2 x i64> @llvm.vector.extract.nxv2i64.nxv8i64(<vscale x 8 x i64> [[VAL]], i64 2)
+// CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 2 x i64> @llvm.vector.extract.nxv2i64.nxv8i64(<vscale x 8 x i64> [[VAL]], i64 4)
+// CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 2 x i64> @llvm.vector.extract.nxv2i64.nxv8i64(<vscale x 8 x i64> [[VAL]], i64 6)
+// CHECK-NEXT:    tail call void @llvm.aarch64.sme.write.ver.vg4.nxv2i64(i32 7, i32 [[BASE:%.*]], <vscale x 2 x i64> [[TMP0]], <vscale x 2 x i64> [[TMP1]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP3]])
+// CHECK-NEXT:    ret void
+//
+// CPP-CHECK-LABEL: @_Z29test_svwrite_ver_za64_u64_vg4j12svuint64x4_t(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 2 x i64> @llvm.vector.extract.nxv2i64.nxv8i64(<vscale x 8 x i64> [[VAL:%.*]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 2 x i64> @llvm.vector.extract.nxv2i64.nxv8i64(<vscale x 8 x i64> [[VAL]], i64 2)
+// CPP-CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 2 x i64> @llvm.vector.extract.nxv2i64.nxv8i64(<vscale x 8 x i64> [[VAL]], i64 4)
+// CPP-CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 2 x i64> @llvm.vector.extract.nxv2i64.nxv8i64(<vscale x 8 x i64> [[VAL]], i64 6)
+// CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.write.ver.vg4.nxv2i64(i32 7, i32 [[BASE:%.*]], <vscale x 2 x i64> [[TMP0]], <vscale x 2 x i64> [[TMP1]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP3]])
+// CPP-CHECK-NEXT:    ret void
+//
+void test_svwrite_ver_za64_u64_vg4(uint32_t base, svuint64x4_t val) {
+  SVE_ACLE_FUNC(svwrite_ver_za64,_u64,_vg4,)(7, base, val);
+}
+
+// CHECK-LABEL: @test_svwrite_hor_za64_f64_vg4(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 2 x double> @llvm.vector.extract.nxv2f64.nxv8f64(<vscale x 8 x double> [[VAL:%.*]], i64 0)
+// CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 2 x double> @llvm.vector.extract.nxv2f64.nxv8f64(<vscale x 8 x double> [[VAL]], i64 2)
+// CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 2 x double> @llvm.vector.extract.nxv2f64.nxv8f64(<vscale x 8 x double> [[VAL]], i64 4)
+// CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 2 x double> @llvm.vector.extract.nxv2f64.nxv8f64(<vscale x 8 x double> [[VAL]], i64 6)
+// CHECK-NEXT:    tail call void @llvm.aarch64.sme.write.hor.vg4.nxv2f64(i32 7, i32 [[BASE:%.*]], <vscale x 2 x double> [[TMP0]], <vscale x 2 x double> [[TMP1]], <vscale x 2 x double> [[TMP2]], <vscale x 2 x double> [[TMP3]])
+// CHECK-NEXT:    ret void
+//
+// CPP-CHECK-LABEL: @_Z29test_svwrite_hor_za64_f64_vg4j13svfloat64x4_t(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 2 x double> @llvm.vector.extract.nxv2f64.nxv8f64(<vscale x 8 x double> [[VAL:%.*]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 2 x double> @llvm.vector.extract.nxv2f64.nxv8f64(<vscale x 8 x double> [[VAL]], i64 2)
+// CPP-CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 2 x double> @llvm.vector.extract.nxv2f64.nxv8f64(<vscale x 8 x double> [[VAL]], i64 4)
+// CPP-CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 2 x double> @llvm.vector.extract.nxv2f64.nxv8f64(<vscale x 8 x double> [[VAL]], i64 6)
+// CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.write.hor.vg4.nxv2f64(i32 7, i32 [[BASE:%.*]], <vscale x 2 x double> [[TMP0]], <vscale x 2 x double> [[TMP1]], <vscale x 2 x double> [[TMP2]], <vscale x 2 x double> [[TMP3]])
+// CPP-CHECK-NEXT:    ret void
+//
+void test_svwrite_hor_za64_f64_vg4(uint32_t base, svfloat64x4_t val) {
+  SVE_ACLE_FUNC(svwrite_hor_za64,_f64,_vg4,)(7, base, val);
+}
+
+// CHECK-LABEL: @test_svwrite_hor_za64_s64_vg4(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 2 x i64> @llvm.vector.extract.nxv2i64.nxv8i64(<vscale x 8 x i64> [[VAL:%.*]], i64 0)
+// CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 2 x i64> @llvm.vector.extract.nxv2i64.nxv8i64(<vscale x 8 x i64> [[VAL]], i64 2)
+// CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 2 x i64> @llvm.vector.extract.nxv2i64.nxv8i64(<vscale x 8 x i64> [[VAL]], i64 4)
+// CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 2 x i64> @llvm.vector.extract.nxv2i64.nxv8i64(<vscale x 8 x i64> [[VAL]], i64 6)
+// CHECK-NEXT:    tail call void @llvm.aarch64.sme.write.hor.vg4.nxv2i64(i32 7, i32 [[BASE:%.*]], <vscale x 2 x i64> [[TMP0]], <vscale x 2 x i64> [[TMP1]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP3]])
+// CHECK-NEXT:    ret void
+//
+// CPP-CHECK-LABEL: @_Z29test_svwrite_hor_za64_s64_vg4j11svint64x4_t(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 2 x i64> @llvm.vector.extract.nxv2i64.nxv8i64(<vscale x 8 x i64> [[VAL:%.*]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 2 x i64> @llvm.vector.extract.nxv2i64.nxv8i64(<vscale x 8 x i64> [[VAL]], i64 2)
+// CPP-CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 2 x i64> @llvm.vector.extract.nxv2i64.nxv8i64(<vscale x 8 x i64> [[VAL]], i64 4)
+// CPP-CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 2 x i64> @llvm.vector.extract.nxv2i64.nxv8i64(<vscale x 8 x i64> [[VAL]], i64 6)
+// CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.write.hor.vg4.nxv2i64(i32 7, i32 [[BASE:%.*]], <vscale x 2 x i64> [[TMP0]], <vscale x 2 x i64> [[TMP1]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP3]])
+// CPP-CHECK-NEXT:    ret void
+//
+void test_svwrite_hor_za64_s64_vg4(uint32_t base, svint64x4_t val) {
+  SVE_ACLE_FUNC(svwrite_hor_za64,_s64,_vg4,)(7, base, val);
+}
+
+// CHECK-LABEL: @test_svwrite_ver_za64_f64_vg4(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 2 x double> @llvm.vector.extract.nxv2f64.nxv8f64(<vscale x 8 x double> [[VAL:%.*]], i64 0)
+// CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 2 x double> @llvm.vector.extract.nxv2f64.nxv8f64(<vscale x 8 x double> [[VAL]], i64 2)
+// CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 2 x double> @llvm.vector.extract.nxv2f64.nxv8f64(<vscale x 8 x double> [[VAL]], i64 4)
+// CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 2 x double> @llvm.vector.extract.nxv2f64.nxv8f64(<vscale x 8 x double> [[VAL]], i64 6)
+// CHECK-NEXT:    tail call void @llvm.aarch64.sme.write.ver.vg4.nxv2f64(i32 7, i32 [[BASE:%.*]], <vscale x 2 x double> [[TMP0]], <vscale x 2 x double> [[TMP1]], <vscale x 2 x double> [[TMP2]], <vscale x 2 x double> [[TMP3]])
+// CHECK-NEXT:    ret void
+//
+// CPP-CHECK-LABEL: @_Z29test_svwrite_ver_za64_f64_vg4j13svfloat64x4_t(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 2 x double> @llvm.vector.extract.nxv2f64.nxv8f64(<vscale x 8 x double> [[VAL:%.*]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 2 x double> @llvm.vector.extract.nxv2f64.nxv8f64(<vscale x 8 x double> [[VAL]], i64 2)
+// CPP-CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 2 x double> @llvm.vector.extract.nxv2f64.nxv8f64(<vscale x 8 x double> [[VAL]], i64 4)
+// CPP-CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 2 x double> @llvm.vector.extract.nxv2f64.nxv8f64(<vscale x 8 x double> [[VAL]], i64 6)
+// CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.write.ver.vg4.nxv2f64(i32 7, i32 [[BASE:%.*]], <vscale x 2 x double> [[TMP0]], <vscale x 2 x double> [[TMP1]], <vscale x 2 x double> [[TMP2]], <vscale x 2 x double> [[TMP3]])
+// CPP-CHECK-NEXT:    ret void
+//
+void test_svwrite_ver_za64_f64_vg4(uint32_t base, svfloat64x4_t val) {
+  SVE_ACLE_FUNC(svwrite_ver_za64,_f64,_vg4,)(7, base, val);
+}
+
+// CHECK-LABEL: @test_svwrite_ver_za64_s64_vg4(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 2 x i64> @llvm.vector.extract.nxv2i64.nxv8i64(<vscale x 8 x i64> [[VAL:%.*]], i64 0)
+// CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 2 x i64> @llvm.vector.extract.nxv2i64.nxv8i64(<vscale x 8 x i64> [[VAL]], i64 2)
+// CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 2 x i64> @llvm.vector.extract.nxv2i64.nxv8i64(<vscale x 8 x i64> [[VAL]], i64 4)
+// CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 2 x i64> @llvm.vector.extract.nxv2i64.nxv8i64(<vscale x 8 x i64> [[VAL]], i64 6)
+// CHECK-NEXT:    tail call void @llvm.aarch64.sme.write.ver.vg4.nxv2i64(i32 7, i32 [[BASE:%.*]], <vscale x 2 x i64> [[TMP0]], <vscale x 2 x i64> [[TMP1]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP3]])
+// CHECK-NEXT:    ret void
+//
+// CPP-CHECK-LABEL: @_Z29test_svwrite_ver_za64_s64_vg4j11svint64x4_t(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 2 x i64> @llvm.vector.extract.nxv2i64.nxv8i64(<vscale x 8 x i64> [[VAL:%.*]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 2 x i64> @llvm.vector.extract.nxv2i64.nxv8i64(<vscale x 8 x i64> [[VAL]], i64 2)
+// CPP-CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 2 x i64> @llvm.vector.extract.nxv2i64.nxv8i64(<vscale x 8 x i64> [[VAL]], i64 4)
+// CPP-CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 2 x i64> @llvm.vector.extract.nxv2i64.nxv8i64(<vscale x 8 x i64> [[VAL]], i64 6)
+// CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.write.ver.vg4.nxv2i64(i32 7, i32 [[BASE:%.*]], <vscale x 2 x i64> [[TMP0]], <vscale x 2 x i64> [[TMP1]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP3]])
+// CPP-CHECK-NEXT:    ret void
+//
+void test_svwrite_ver_za64_s64_vg4(uint32_t base, svint64x4_t val) {
+  SVE_ACLE_FUNC(svwrite_ver_za64,_s64,_vg4,)(7, base, val);
+}
+
+// CHECK-LABEL: @test_svwrite_za64_u64_vg1x2(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 2 x i64> @llvm.vector.extract.nxv2i64.nxv4i64(<vscale x 4 x i64> [[VAL:%.*]], i64 0)
+// CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 2 x i64> @llvm.vector.extract.nxv2i64.nxv4i64(<vscale x 4 x i64> [[VAL]], i64 2)
+// CHECK-NEXT:    tail call void @llvm.aarch64.sme.write.vg1x2.nxv2i64(i32 [[BASE:%.*]], <vscale x 2 x i64> [[TMP0]], <vscale x 2 x i64> [[TMP1]])
+// CHECK-NEXT:    ret void
+//
+// CPP-CHECK-LABEL: @_Z27test_svwrite_za64_u64_vg1x2j12svuint64x2_t(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 2 x i64> @llvm.vector.extract.nxv2i64.nxv4i64(<vscale x 4 x i64> [[VAL:%.*]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 2 x i64> @llvm.vector.extract.nxv2i64.nxv4i64(<vscale x 4 x i64> [[VAL]], i64 2)
+// CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.write.vg1x2.nxv2i64(i32 [[BASE:%.*]], <vscale x 2 x i64> [[TMP0]], <vscale x 2 x i64> [[TMP1]])
+// CPP-CHECK-NEXT:    ret void
+//
+void test_svwrite_za64_u64_vg1x2(uint32_t base, svuint64x2_t val) {
+  SVE_ACLE_FUNC(svwrite_za64,_u64,_vg1x2,)(base, val);
+}
+
+// CHECK-LABEL: @test_svwrite_za64_f64_vg1x2(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 2 x double> @llvm.vector.extract.nxv2f64.nxv4f64(<vscale x 4 x double> [[VAL:%.*]], i64 0)
+// CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 2 x double> @llvm.vector.extract.nxv2f64.nxv4f64(<vscale x 4 x double> [[VAL]], i64 2)
+// CHECK-NEXT:    tail call void @llvm.aarch64.sme.write.vg1x2.nxv2f64(i32 [[BASE:%.*]], <vscale x 2 x double> [[TMP0]], <vscale x 2 x double> [[TMP1]])
+// CHECK-NEXT:    ret void
+//
+// CPP-CHECK-LABEL: @_Z27test_svwrite_za64_f64_vg1x2j13svfloat64x2_t(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 2 x double> @llvm.vector.extract.nxv2f64.nxv4f64(<vscale x 4 x double> [[VAL:%.*]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 2 x double> @llvm.vector.extract.nxv2f64.nxv4f64(<vscale x 4 x double> [[VAL]], i64 2)
+// CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.write.vg1x2.nxv2f64(i32 [[BASE:%.*]], <vscale x 2 x double> [[TMP0]], <vscale x 2 x double> [[TMP1]])
+// CPP-CHECK-NEXT:    ret void
+//
+void test_svwrite_za64_f64_vg1x2(uint32_t base, svfloat64x2_t val) {
+  SVE_ACLE_FUNC(svwrite_za64,_f64,_vg1x2,)(base, val);
+}
+
+// CHECK-LABEL: @test_svwrite_za64_s64_vg1x2(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 2 x i64> @llvm.vector.extract.nxv2i64.nxv4i64(<vscale x 4 x i64> [[VAL:%.*]], i64 0)
+// CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 2 x i64> @llvm.vector.extract.nxv2i64.nxv4i64(<vscale x 4 x i64> [[VAL]], i64 2)
+// CHECK-NEXT:    tail call void @llvm.aarch64.sme.write.vg1x2.nxv2i64(i32 [[BASE:%.*]], <vscale x 2 x i64> [[TMP0]], <vscale x 2 x i64> [[TMP1]])
+// CHECK-NEXT:    ret void
+//
+// CPP-CHECK-LABEL: @_Z27test_svwrite_za64_s64_vg1x2j11svint64x2_t(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 2 x i64> @llvm.vector.extract.nxv2i64.nxv4i64(<vscale x 4 x i64> [[VAL:%.*]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 2 x i64> @llvm.vector.extract.nxv2i64.nxv4i64(<vscale x 4 x i64> [[VAL]], i64 2)
+// CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.write.vg1x2.nxv2i64(i32 [[BASE:%.*]], <vscale x 2 x i64> [[TMP0]], <vscale x 2 x i64> [[TMP1]])
+// CPP-CHECK-NEXT:    ret void
+//
+void test_svwrite_za64_s64_vg1x2(uint32_t base, svint64x2_t val) {
+  SVE_ACLE_FUNC(svwrite_za64,_s64,_vg1x2,)(base, val);
+}
+
+// CHECK-LABEL: @test_svwrite_za64_u64_vg1x4(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 2 x i64> @llvm.vector.extract.nxv2i64.nxv8i64(<vscale x 8 x i64> [[VAL:%.*]], i64 0)
+// CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 2 x i64> @llvm.vector.extract.nxv2i64.nxv8i64(<vscale x 8 x i64> [[VAL]], i64 2)
+// CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 2 x i64> @llvm.vector.extract.nxv2i64.nxv8i64(<vscale x 8 x i64> [[VAL]], i64 4)
+// CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 2 x i64> @llvm.vector.extract.nxv2i64.nxv8i64(<vscale x 8 x i64> [[VAL]], i64 6)
+// CHECK-NEXT:    tail call void @llvm.aarch64.sme.write.vg1x4.nxv2i64(i32 [[BASE:%.*]], <vscale x 2 x i64> [[TMP0]], <vscale x 2 x i64> [[TMP1]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP3]])
+// CHECK-NEXT:    ret void
+//
+// CPP-CHECK-LABEL: @_Z27test_svwrite_za64_u64_vg1x4j12svuint64x4_t(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 2 x i64> @llvm.vector.extract.nxv2i64.nxv8i64(<vscale x 8 x i64> [[VAL:%.*]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 2 x i64> @llvm.vector.extract.nxv2i64.nxv8i64(<vscale x 8 x i64> [[VAL]], i64 2)
+// CPP-CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 2 x i64> @llvm.vector.extract.nxv2i64.nxv8i64(<vscale x 8 x i64> [[VAL]], i64 4)
+// CPP-CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 2 x i64> @llvm.vector.extract.nxv2i64.nxv8i64(<vscale x 8 x i64> [[VAL]], i64 6)
+// CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.write.vg1x4.nxv2i64(i32 [[BASE:%.*]], <vscale x 2 x i64> [[TMP0]], <vscale x 2 x i64> [[TMP1]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP3]])
+// CPP-CHECK-NEXT:    ret void
+//
+void test_svwrite_za64_u64_vg1x4(uint32_t base, svuint64x4_t val) {
+  SVE_ACLE_FUNC(svwrite_za64,_u64,_vg1x4,)(base, val);
+}
+
+// CHECK-LABEL: @test_svwrite_za64_f64_vg1x4(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 2 x double> @llvm.vector.extract.nxv2f64.nxv8f64(<vscale x 8 x double> [[VAL:%.*]], i64 0)
+// CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 2 x double> @llvm.vector.extract.nxv2f64.nxv8f64(<vscale x 8 x double> [[VAL]], i64 2)
+// CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 2 x double> @llvm.vector.extract.nxv2f64.nxv8f64(<vscale x 8 x double> [[VAL]], i64 4)
+// CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 2 x double> @llvm.vector.extract.nxv2f64.nxv8f64(<vscale x 8 x double> [[VAL]], i64 6)
+// CHECK-NEXT:    tail call void @llvm.aarch64.sme.write.vg1x4.nxv2f64(i32 [[BASE:%.*]], <vscale x 2 x double> [[TMP0]], <vscale x 2 x double> [[TMP1]], <vscale x 2 x double> [[TMP2]], <vscale x 2 x double> [[TMP3]])
+// CHECK-NEXT:    ret void
+//
+// CPP-CHECK-LABEL: @_Z27test_svwrite_za64_f64_vg1x4j13svfloat64x4_t(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 2 x double> @llvm.vector.extract.nxv2f64.nxv8f64(<vscale x 8 x double> [[VAL:%.*]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 2 x double> @llvm.vector.extract.nxv2f64.nxv8f64(<vscale x 8 x double> [[VAL]], i64 2)
+// CPP-CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 2 x double> @llvm.vector.extract.nxv2f64.nxv8f64(<vscale x 8 x double> [[VAL]], i64 4)
+// CPP-CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 2 x double> @llvm.vector.extract.nxv2f64.nxv8f64(<vscale x 8 x double> [[VAL]], i64 6)
+// CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.write.vg1x4.nxv2f64(i32 [[BASE:%.*]], <vscale x 2 x double> [[TMP0]], <vscale x 2 x double> [[TMP1]], <vscale x 2 x double> [[TMP2]], <vscale x 2 x double> [[TMP3]])
+// CPP-CHECK-NEXT:    ret void
+//
+void test_svwrite_za64_f64_vg1x4(uint32_t base, svfloat64x4_t val) {
+  SVE_ACLE_FUNC(svwrite_za64,_f64,_vg1x4,)(base, val);
+}
+
+// CHECK-LABEL: @test_svwrite_za64_s64_vg1x4(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 2 x i64> @llvm.vector.extract.nxv2i64.nxv8i64(<vscale x 8 x i64> [[VAL:%.*]], i64 0)
+// CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 2 x i64> @llvm.vector.extract.nxv2i64.nxv8i64(<vscale x 8 x i64> [[VAL]], i64 2)
+// CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 2 x i64> @llvm.vector.extract.nxv2i64.nxv8i64(<vscale x 8 x i64> [[VAL]], i64 4)
+// CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 2 x i64> @llvm.vector.extract.nxv2i64.nxv8i64(<vscale x 8 x i64> [[VAL]], i64 6)
+// CHECK-NEXT:    tail call void @llvm.aarch64.sme.write.vg1x4.nxv2i64(i32 [[BASE:%.*]], <vscale x 2 x i64> [[TMP0]], <vscale x 2 x i64> [[TMP1]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP3]])
+// CHECK-NEXT:    ret void
+//
+// CPP-CHECK-LABEL: @_Z27test_svwrite_za64_s64_vg1x4j11svint64x4_t(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 2 x i64> @llvm.vector.extract.nxv2i64.nxv8i64(<vscale x 8 x i64> [[VAL:%.*]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 2 x i64> @llvm.vector.extract.nxv2i64.nxv8i64(<vscale x 8 x i64> [[VAL]], i64 2)
+// CPP-CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 2 x i64> @llvm.vector.extract.nxv2i64.nxv8i64(<vscale x 8 x i64> [[VAL]], i64 4)
+// CPP-CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 2 x i64> @llvm.vector.extract.nxv2i64.nxv8i64(<vscale x 8 x i64> [[VAL]], i64 6)
+// CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.write.vg1x4.nxv2i64(i32 [[BASE:%.*]], <vscale x 2 x i64> [[TMP0]], <vscale x 2 x i64> [[TMP1]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP3]])
+// CPP-CHECK-NEXT:    ret void
+//
+void test_svwrite_za64_s64_vg1x4(uint32_t base, svint64x4_t val) {
+  SVE_ACLE_FUNC(svwrite_za64,_s64,_vg1x4,)(base, val);
+}
diff --git a/clang/test/Sema/aarch64-sme2-intrinsics/acle_sme2_imm.cpp b/clang/test/Sema/aarch64-sme2-intrinsics/acle_sme2_imm.cpp
index 70987ad395f73..77b1133f55101 100644
--- a/clang/test/Sema/aarch64-sme2-intrinsics/acle_sme2_imm.cpp
+++ b/clang/test/Sema/aarch64-sme2-intrinsics/acle_sme2_imm.cpp
@@ -5,6 +5,57 @@
 
 #include <arm_sme_draft_spec_subject_to_change.h>
 
+void test_multivector_read(uint32_t base) {
+
+  // Test Tile Range
+  svread_hor_za8_u8_vg2(1, base); // expected-error {{argument value 1 is outside the valid range [0, 0]}}
+  svread_ver_za8_u8_vg2(1, base); // expected-error {{argument value 1 is outside the valid range [0, 0]}}
+  svread_hor_za8_u8_vg4(1, base); // expected-error {{argument value 1 is outside the valid range [0, 0]}}
+  svread_ver_za8_u8_vg4(1, base); // expected-error {{argument value 1 is outside the valid range [0, 0]}}
+
+  svread_hor_za16_u16_vg2(2, base); // expected-error {{argument value 2 is outside the valid range [0, 1]}}
+  svread_ver_za16_u16_vg2(2, base); // expected-error {{argument value 2 is outside the valid range [0, 1]}}
+  svread_hor_za16_u16_vg4(2, base); // expected-error {{argument value 2 is outside the valid range [0, 1]}}
+  svread_ver_za16_u16_vg4(2, base); // expected-error {{argument value 2 is outside the valid range [0, 1]}}
+
+  svread_hor_za32_u32_vg2(4, base); // expected-error {{argument value 4 is outside the valid range [0, 3]}}
+  svread_ver_za32_u32_vg2(4, base); // expected-error {{argument value 4 is outside the valid range [0, 3]}}
+  svread_hor_za32_u32_vg4(4, base); // expected-error {{argument value 4 is outside the valid range [0, 3]}}
+  svread_ver_za32_u32_vg4(4, base); // expected-error {{argument value 4 is outside the valid range [0, 3]}}
+
+  svread_hor_za64_u64_vg2(8, base); // expected-error {{argument value 8 is outside the valid range [0, 7]}}
+  svread_ver_za64_u64_vg2(8, base); // expected-error {{argument value 8 is outside the valid range [0, 7]}}
+  svread_hor_za64_u64_vg4(8, base); // expected-error {{argument value 8 is outside the valid range [0, 7]}}
+  svread_ver_za64_u64_vg4(8, base); // expected-error {{argument value 8 is outside the valid range [0, 7]}}
+}
+
+void test_multivector_write(uint32_t base, svuint8x2_t v8x2, svuint8x4_t v8x4,
+                            svuint16x2_t v16x2, svuint16x4_t v16x4,
+                            svuint32x2_t v32x2, svuint32x4_t v32x4,
+                            svuint64x2_t v64x2, svuint64x4_t v64x4) {
+
+  // Test Tile Range
+  svwrite_hor_za8_u8_vg2(1, base, v8x2); // expected-error {{argument value 1 is outside the valid range [0, 0]}}
+  svwrite_ver_za8_u8_vg2(1, base, v8x2); // expected-error {{argument value 1 is outside the valid range [0, 0]}}
+  svwrite_hor_za8_u8_vg4(1, base, v8x4); // expected-error {{argument value 1 is outside the valid range [0, 0]}}
+  svwrite_ver_za8_u8_vg4(1, base, v8x4); // expected-error {{argument value 1 is outside the valid range [0, 0]}}
+
+  svwrite_hor_za16_u16_vg2(2, base, v16x2); // expected-error {{argument value 2 is outside the valid range [0, 1]}}
+  svwrite_ver_za16_u16_vg2(2, base, v16x2); // expected-error {{argument value 2 is outside the valid range [0, 1]}}
+  svwrite_hor_za16_u16_vg4(2, base, v16x4); // expected-error {{argument value 2 is outside the valid range [0, 1]}}
+  svwrite_ver_za16_u16_vg4(2, base, v16x4); // expected-error {{argument value 2 is outside the valid range [0, 1]}}
+
+  svwrite_hor_za32_u32_vg2(4, base, v32x2); // expected-error {{argument value 4 is outside the valid range [0, 3]}}
+  svwrite_ver_za32_u32_vg2(4, base, v32x2); // expected-error {{argument value 4 is outside the valid range [0, 3]}}
+  svwrite_hor_za32_u32_vg4(4, base, v32x4); // expected-error {{argument value 4 is outside the valid range [0, 3]}}
+  svwrite_ver_za32_u32_vg4(4, base, v32x4); // expected-error {{argument value 4 is outside the valid range [0, 3]}}
+
+  svwrite_hor_za64_u64_vg2(8, base, v64x2); // expected-error {{argument value 8 is outside the valid range [0, 7]}}
+  svwrite_ver_za64_u64_vg2(8, base, v64x2); // expected-error {{argument value 8 is outside the valid range [0, 7]}}
+  svwrite_hor_za64_u64_vg4(8, base, v64x4); // expected-error {{argument value 8 is outside the valid range [0, 7]}}
+  svwrite_ver_za64_u64_vg4(8, base, v64x4); // expected-error {{argument value 8 is outside the valid range [0, 7]}}
+}
+
 void test_outer_product(svbool_t pred, svint16_t s16, svuint16_t u16, svint32_t s32, svuint32_t u32) __arm_streaming __arm_shared_za {
   // Test Tile Range
   svmopa_za32_u16_m(4, pred, pred, u16, u16); // expected-error {{argument value 4 is outside the valid range [0, 3]}}

From c369e641b58328fdde662d7b97557855a5ded3e5 Mon Sep 17 00:00:00 2001
From: Kerry McLaughlin <kerry.mclaughlin@arm.com>
Date: Mon, 13 Nov 2023 10:39:29 +0000
Subject: [PATCH 2/5] - Add arm_streaming arm_shared_za & arm_preserves_za
 attributes to tests

---
 .../aarch64-sme2-intrinsics/acle_sme2_read.c  | 108 +++++++++---------
 .../aarch64-sme2-intrinsics/acle_sme2_write.c | 108 +++++++++---------
 .../aarch64-sme2-intrinsics/acle_sme2_imm.cpp |   4 +-
 3 files changed, 110 insertions(+), 110 deletions(-)

diff --git a/clang/test/CodeGen/aarch64-sme2-intrinsics/acle_sme2_read.c b/clang/test/CodeGen/aarch64-sme2-intrinsics/acle_sme2_read.c
index 706403e5180ad..89700afca41e4 100644
--- a/clang/test/CodeGen/aarch64-sme2-intrinsics/acle_sme2_read.c
+++ b/clang/test/CodeGen/aarch64-sme2-intrinsics/acle_sme2_read.c
@@ -34,7 +34,7 @@
 // CPP-CHECK-NEXT:    [[TMP4:%.*]] = tail call <vscale x 32 x i8> @llvm.vector.insert.nxv32i8.nxv16i8(<vscale x 32 x i8> [[TMP2]], <vscale x 16 x i8> [[TMP3]], i64 16)
 // CPP-CHECK-NEXT:    ret <vscale x 32 x i8> [[TMP4]]
 //
-svuint8x2_t test_svread_ver_za8_u8_vg2(uint32_t base) {
+svuint8x2_t test_svread_ver_za8_u8_vg2(uint32_t base) __arm_streaming __arm_shared_za __arm_preserves_za {
   return svread_ver_za8_u8_vg2(0, base);
 }
 
@@ -56,7 +56,7 @@ svuint8x2_t test_svread_ver_za8_u8_vg2(uint32_t base) {
 // CPP-CHECK-NEXT:    [[TMP4:%.*]] = tail call <vscale x 32 x i8> @llvm.vector.insert.nxv32i8.nxv16i8(<vscale x 32 x i8> [[TMP2]], <vscale x 16 x i8> [[TMP3]], i64 16)
 // CPP-CHECK-NEXT:    ret <vscale x 32 x i8> [[TMP4]]
 //
-svint8x2_t test_svread_ver_za8_s8_vg2(uint32_t base) {
+svint8x2_t test_svread_ver_za8_s8_vg2(uint32_t base) __arm_streaming __arm_shared_za __arm_preserves_za {
   return svread_ver_za8_s8_vg2(0, base);
 }
 
@@ -78,7 +78,7 @@ svint8x2_t test_svread_ver_za8_s8_vg2(uint32_t base) {
 // CPP-CHECK-NEXT:    [[TMP4:%.*]] = tail call <vscale x 32 x i8> @llvm.vector.insert.nxv32i8.nxv16i8(<vscale x 32 x i8> [[TMP2]], <vscale x 16 x i8> [[TMP3]], i64 16)
 // CPP-CHECK-NEXT:    ret <vscale x 32 x i8> [[TMP4]]
 //
-svuint8x2_t test_svread_hor_za8_u8_vg2(uint32_t base) {
+svuint8x2_t test_svread_hor_za8_u8_vg2(uint32_t base) __arm_streaming __arm_shared_za __arm_preserves_za {
   return svread_hor_za8_u8_vg2(0, base);
 }
 
@@ -100,7 +100,7 @@ svuint8x2_t test_svread_hor_za8_u8_vg2(uint32_t base) {
 // CPP-CHECK-NEXT:    [[TMP4:%.*]] = tail call <vscale x 32 x i8> @llvm.vector.insert.nxv32i8.nxv16i8(<vscale x 32 x i8> [[TMP2]], <vscale x 16 x i8> [[TMP3]], i64 16)
 // CPP-CHECK-NEXT:    ret <vscale x 32 x i8> [[TMP4]]
 //
-svint8x2_t test_svread_hor_za8_s8_vg2(uint32_t base) {
+svint8x2_t test_svread_hor_za8_s8_vg2(uint32_t base) __arm_streaming __arm_shared_za __arm_preserves_za {
   return svread_hor_za8_s8_vg2(0, base);
 }
 
@@ -130,7 +130,7 @@ svint8x2_t test_svread_hor_za8_s8_vg2(uint32_t base) {
 // CPP-CHECK-NEXT:    [[TMP8:%.*]] = tail call <vscale x 64 x i8> @llvm.vector.insert.nxv64i8.nxv16i8(<vscale x 64 x i8> [[TMP6]], <vscale x 16 x i8> [[TMP7]], i64 48)
 // CPP-CHECK-NEXT:    ret <vscale x 64 x i8> [[TMP8]]
 //
-svuint8x4_t test_svread_hor_za8_u8_vg4(uint32_t base) {
+svuint8x4_t test_svread_hor_za8_u8_vg4(uint32_t base) __arm_streaming __arm_shared_za __arm_preserves_za {
   return svread_hor_za8_u8_vg4(0, base);
 }
 
@@ -160,7 +160,7 @@ svuint8x4_t test_svread_hor_za8_u8_vg4(uint32_t base) {
 // CPP-CHECK-NEXT:    [[TMP8:%.*]] = tail call <vscale x 64 x i8> @llvm.vector.insert.nxv64i8.nxv16i8(<vscale x 64 x i8> [[TMP6]], <vscale x 16 x i8> [[TMP7]], i64 48)
 // CPP-CHECK-NEXT:    ret <vscale x 64 x i8> [[TMP8]]
 //
-svint8x4_t test_svread_hor_za8_s8_vg4(uint32_t base) {
+svint8x4_t test_svread_hor_za8_s8_vg4(uint32_t base) __arm_streaming __arm_shared_za __arm_preserves_za {
   return svread_hor_za8_s8_vg4(0, base);
 }
 
@@ -190,7 +190,7 @@ svint8x4_t test_svread_hor_za8_s8_vg4(uint32_t base) {
 // CPP-CHECK-NEXT:    [[TMP8:%.*]] = tail call <vscale x 64 x i8> @llvm.vector.insert.nxv64i8.nxv16i8(<vscale x 64 x i8> [[TMP6]], <vscale x 16 x i8> [[TMP7]], i64 48)
 // CPP-CHECK-NEXT:    ret <vscale x 64 x i8> [[TMP8]]
 //
-svuint8x4_t test_svread_ver_za8_u8_vg4(uint32_t base) {
+svuint8x4_t test_svread_ver_za8_u8_vg4(uint32_t base) __arm_streaming __arm_shared_za __arm_preserves_za {
   return svread_ver_za8_u8_vg4(0, base);
 }
 
@@ -220,7 +220,7 @@ svuint8x4_t test_svread_ver_za8_u8_vg4(uint32_t base) {
 // CPP-CHECK-NEXT:    [[TMP8:%.*]] = tail call <vscale x 64 x i8> @llvm.vector.insert.nxv64i8.nxv16i8(<vscale x 64 x i8> [[TMP6]], <vscale x 16 x i8> [[TMP7]], i64 48)
 // CPP-CHECK-NEXT:    ret <vscale x 64 x i8> [[TMP8]]
 //
-svint8x4_t test_svread_ver_za8_s8_vg4(uint32_t base) {
+svint8x4_t test_svread_ver_za8_s8_vg4(uint32_t base) __arm_streaming __arm_shared_za __arm_preserves_za {
   return svread_ver_za8_s8_vg4(0, base);
 }
 
@@ -242,7 +242,7 @@ svint8x4_t test_svread_ver_za8_s8_vg4(uint32_t base) {
 // CPP-CHECK-NEXT:    [[TMP4:%.*]] = tail call <vscale x 16 x i16> @llvm.vector.insert.nxv16i16.nxv8i16(<vscale x 16 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP3]], i64 8)
 // CPP-CHECK-NEXT:    ret <vscale x 16 x i16> [[TMP4]]
 //
-svuint16x2_t test_svread_hor_za16_u16_vg2(uint32_t base) {
+svuint16x2_t test_svread_hor_za16_u16_vg2(uint32_t base) __arm_streaming __arm_shared_za __arm_preserves_za {
   return svread_hor_za16_u16_vg2(1, base);
 }
 
@@ -264,7 +264,7 @@ svuint16x2_t test_svread_hor_za16_u16_vg2(uint32_t base) {
 // CPP-CHECK-NEXT:    [[TMP4:%.*]] = tail call <vscale x 16 x bfloat> @llvm.vector.insert.nxv16bf16.nxv8bf16(<vscale x 16 x bfloat> [[TMP2]], <vscale x 8 x bfloat> [[TMP3]], i64 8)
 // CPP-CHECK-NEXT:    ret <vscale x 16 x bfloat> [[TMP4]]
 //
-svbfloat16x2_t test_svread_hor_za16_bf16_vg2(uint32_t base) {
+svbfloat16x2_t test_svread_hor_za16_bf16_vg2(uint32_t base) __arm_streaming __arm_shared_za __arm_preserves_za {
   return svread_hor_za16_bf16_vg2(1, base);
 }
 
@@ -286,7 +286,7 @@ svbfloat16x2_t test_svread_hor_za16_bf16_vg2(uint32_t base) {
 // CPP-CHECK-NEXT:    [[TMP4:%.*]] = tail call <vscale x 16 x half> @llvm.vector.insert.nxv16f16.nxv8f16(<vscale x 16 x half> [[TMP2]], <vscale x 8 x half> [[TMP3]], i64 8)
 // CPP-CHECK-NEXT:    ret <vscale x 16 x half> [[TMP4]]
 //
-svfloat16x2_t test_svread_hor_za16_f16_vg2(uint32_t base) {
+svfloat16x2_t test_svread_hor_za16_f16_vg2(uint32_t base) __arm_streaming __arm_shared_za __arm_preserves_za {
   return svread_hor_za16_f16_vg2(1, base);
 }
 
@@ -308,7 +308,7 @@ svfloat16x2_t test_svread_hor_za16_f16_vg2(uint32_t base) {
 // CPP-CHECK-NEXT:    [[TMP4:%.*]] = tail call <vscale x 16 x i16> @llvm.vector.insert.nxv16i16.nxv8i16(<vscale x 16 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP3]], i64 8)
 // CPP-CHECK-NEXT:    ret <vscale x 16 x i16> [[TMP4]]
 //
-svint16x2_t test_svread_hor_za16_s16_vg2(uint32_t base) {
+svint16x2_t test_svread_hor_za16_s16_vg2(uint32_t base) __arm_streaming __arm_shared_za __arm_preserves_za {
   return svread_hor_za16_s16_vg2(1, base);
 }
 
@@ -330,7 +330,7 @@ svint16x2_t test_svread_hor_za16_s16_vg2(uint32_t base) {
 // CPP-CHECK-NEXT:    [[TMP4:%.*]] = tail call <vscale x 16 x i16> @llvm.vector.insert.nxv16i16.nxv8i16(<vscale x 16 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP3]], i64 8)
 // CPP-CHECK-NEXT:    ret <vscale x 16 x i16> [[TMP4]]
 //
-svuint16x2_t test_svread_ver_za16_u16_vg2(uint32_t base) {
+svuint16x2_t test_svread_ver_za16_u16_vg2(uint32_t base) __arm_streaming __arm_shared_za __arm_preserves_za {
   return svread_ver_za16_u16_vg2(1, base);
 }
 
@@ -352,7 +352,7 @@ svuint16x2_t test_svread_ver_za16_u16_vg2(uint32_t base) {
 // CPP-CHECK-NEXT:    [[TMP4:%.*]] = tail call <vscale x 16 x bfloat> @llvm.vector.insert.nxv16bf16.nxv8bf16(<vscale x 16 x bfloat> [[TMP2]], <vscale x 8 x bfloat> [[TMP3]], i64 8)
 // CPP-CHECK-NEXT:    ret <vscale x 16 x bfloat> [[TMP4]]
 //
-svbfloat16x2_t test_svread_ver_za16_bf16_vg2(uint32_t base) {
+svbfloat16x2_t test_svread_ver_za16_bf16_vg2(uint32_t base) __arm_streaming __arm_shared_za __arm_preserves_za {
   return svread_ver_za16_bf16_vg2(1, base);
 }
 
@@ -374,7 +374,7 @@ svbfloat16x2_t test_svread_ver_za16_bf16_vg2(uint32_t base) {
 // CPP-CHECK-NEXT:    [[TMP4:%.*]] = tail call <vscale x 16 x half> @llvm.vector.insert.nxv16f16.nxv8f16(<vscale x 16 x half> [[TMP2]], <vscale x 8 x half> [[TMP3]], i64 8)
 // CPP-CHECK-NEXT:    ret <vscale x 16 x half> [[TMP4]]
 //
-svfloat16x2_t test_svread_ver_za16_f16_vg2(uint32_t base) {
+svfloat16x2_t test_svread_ver_za16_f16_vg2(uint32_t base) __arm_streaming __arm_shared_za __arm_preserves_za {
   return svread_ver_za16_f16_vg2(1, base);
 }
 
@@ -396,7 +396,7 @@ svfloat16x2_t test_svread_ver_za16_f16_vg2(uint32_t base) {
 // CPP-CHECK-NEXT:    [[TMP4:%.*]] = tail call <vscale x 16 x i16> @llvm.vector.insert.nxv16i16.nxv8i16(<vscale x 16 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP3]], i64 8)
 // CPP-CHECK-NEXT:    ret <vscale x 16 x i16> [[TMP4]]
 //
-svint16x2_t test_svread_ver_za16_s16_vg2(uint32_t base) {
+svint16x2_t test_svread_ver_za16_s16_vg2(uint32_t base) __arm_streaming __arm_shared_za __arm_preserves_za {
   return svread_ver_za16_s16_vg2(1, base);
 }
 
@@ -426,7 +426,7 @@ svint16x2_t test_svread_ver_za16_s16_vg2(uint32_t base) {
 // CPP-CHECK-NEXT:    [[TMP8:%.*]] = tail call <vscale x 32 x i16> @llvm.vector.insert.nxv32i16.nxv8i16(<vscale x 32 x i16> [[TMP6]], <vscale x 8 x i16> [[TMP7]], i64 24)
 // CPP-CHECK-NEXT:    ret <vscale x 32 x i16> [[TMP8]]
 //
-svuint16x4_t test_svread_hor_za16_u16_vg4(uint32_t base) {
+svuint16x4_t test_svread_hor_za16_u16_vg4(uint32_t base) __arm_streaming __arm_shared_za __arm_preserves_za {
   return svread_hor_za16_u16_vg4(1, base);
 }
 
@@ -456,7 +456,7 @@ svuint16x4_t test_svread_hor_za16_u16_vg4(uint32_t base) {
 // CPP-CHECK-NEXT:    [[TMP8:%.*]] = tail call <vscale x 32 x bfloat> @llvm.vector.insert.nxv32bf16.nxv8bf16(<vscale x 32 x bfloat> [[TMP6]], <vscale x 8 x bfloat> [[TMP7]], i64 24)
 // CPP-CHECK-NEXT:    ret <vscale x 32 x bfloat> [[TMP8]]
 //
-svbfloat16x4_t test_svread_hor_za16_bf16_vg4(uint32_t base) {
+svbfloat16x4_t test_svread_hor_za16_bf16_vg4(uint32_t base) __arm_streaming __arm_shared_za __arm_preserves_za {
   return svread_hor_za16_bf16_vg4(1, base);
 }
 
@@ -486,7 +486,7 @@ svbfloat16x4_t test_svread_hor_za16_bf16_vg4(uint32_t base) {
 // CPP-CHECK-NEXT:    [[TMP8:%.*]] = tail call <vscale x 32 x half> @llvm.vector.insert.nxv32f16.nxv8f16(<vscale x 32 x half> [[TMP6]], <vscale x 8 x half> [[TMP7]], i64 24)
 // CPP-CHECK-NEXT:    ret <vscale x 32 x half> [[TMP8]]
 //
-svfloat16x4_t test_svread_hor_za16_f16_vg4(uint32_t base) {
+svfloat16x4_t test_svread_hor_za16_f16_vg4(uint32_t base) __arm_streaming __arm_shared_za __arm_preserves_za {
   return svread_hor_za16_f16_vg4(1, base);
 }
 
@@ -516,7 +516,7 @@ svfloat16x4_t test_svread_hor_za16_f16_vg4(uint32_t base) {
 // CPP-CHECK-NEXT:    [[TMP8:%.*]] = tail call <vscale x 32 x i16> @llvm.vector.insert.nxv32i16.nxv8i16(<vscale x 32 x i16> [[TMP6]], <vscale x 8 x i16> [[TMP7]], i64 24)
 // CPP-CHECK-NEXT:    ret <vscale x 32 x i16> [[TMP8]]
 //
-svint16x4_t test_svread_hor_za16_s16_vg4(uint32_t base) {
+svint16x4_t test_svread_hor_za16_s16_vg4(uint32_t base) __arm_streaming __arm_shared_za __arm_preserves_za {
   return svread_hor_za16_s16_vg4(1, base);
 }
 
@@ -546,7 +546,7 @@ svint16x4_t test_svread_hor_za16_s16_vg4(uint32_t base) {
 // CPP-CHECK-NEXT:    [[TMP8:%.*]] = tail call <vscale x 32 x i16> @llvm.vector.insert.nxv32i16.nxv8i16(<vscale x 32 x i16> [[TMP6]], <vscale x 8 x i16> [[TMP7]], i64 24)
 // CPP-CHECK-NEXT:    ret <vscale x 32 x i16> [[TMP8]]
 //
-svuint16x4_t test_svread_ver_za16_u16_vg4(uint32_t base) {
+svuint16x4_t test_svread_ver_za16_u16_vg4(uint32_t base) __arm_streaming __arm_shared_za __arm_preserves_za {
   return svread_ver_za16_u16_vg4(1, base);
 }
 
@@ -576,7 +576,7 @@ svuint16x4_t test_svread_ver_za16_u16_vg4(uint32_t base) {
 // CPP-CHECK-NEXT:    [[TMP8:%.*]] = tail call <vscale x 32 x bfloat> @llvm.vector.insert.nxv32bf16.nxv8bf16(<vscale x 32 x bfloat> [[TMP6]], <vscale x 8 x bfloat> [[TMP7]], i64 24)
 // CPP-CHECK-NEXT:    ret <vscale x 32 x bfloat> [[TMP8]]
 //
-svbfloat16x4_t test_svread_ver_za16_bf16_vg4(uint32_t base) {
+svbfloat16x4_t test_svread_ver_za16_bf16_vg4(uint32_t base) __arm_streaming __arm_shared_za __arm_preserves_za {
   return svread_ver_za16_bf16_vg4(1, base);
 }
 
@@ -606,7 +606,7 @@ svbfloat16x4_t test_svread_ver_za16_bf16_vg4(uint32_t base) {
 // CPP-CHECK-NEXT:    [[TMP8:%.*]] = tail call <vscale x 32 x half> @llvm.vector.insert.nxv32f16.nxv8f16(<vscale x 32 x half> [[TMP6]], <vscale x 8 x half> [[TMP7]], i64 24)
 // CPP-CHECK-NEXT:    ret <vscale x 32 x half> [[TMP8]]
 //
-svfloat16x4_t test_svread_ver_za16_f16_vg4(uint32_t base) {
+svfloat16x4_t test_svread_ver_za16_f16_vg4(uint32_t base) __arm_streaming __arm_shared_za __arm_preserves_za {
   return svread_ver_za16_f16_vg4(1, base);
 }
 
@@ -636,7 +636,7 @@ svfloat16x4_t test_svread_ver_za16_f16_vg4(uint32_t base) {
 // CPP-CHECK-NEXT:    [[TMP8:%.*]] = tail call <vscale x 32 x i16> @llvm.vector.insert.nxv32i16.nxv8i16(<vscale x 32 x i16> [[TMP6]], <vscale x 8 x i16> [[TMP7]], i64 24)
 // CPP-CHECK-NEXT:    ret <vscale x 32 x i16> [[TMP8]]
 //
-svint16x4_t test_svread_ver_za16_s16_vg4(uint32_t base) {
+svint16x4_t test_svread_ver_za16_s16_vg4(uint32_t base) __arm_streaming __arm_shared_za __arm_preserves_za {
   return svread_ver_za16_s16_vg4(1, base);
 }
 
@@ -658,7 +658,7 @@ svint16x4_t test_svread_ver_za16_s16_vg4(uint32_t base) {
 // CPP-CHECK-NEXT:    [[TMP4:%.*]] = tail call <vscale x 8 x i32> @llvm.vector.insert.nxv8i32.nxv4i32(<vscale x 8 x i32> [[TMP2]], <vscale x 4 x i32> [[TMP3]], i64 4)
 // CPP-CHECK-NEXT:    ret <vscale x 8 x i32> [[TMP4]]
 //
-svuint32x2_t test_svread_hor_za32_u32_vg2(uint32_t base) {
+svuint32x2_t test_svread_hor_za32_u32_vg2(uint32_t base) __arm_streaming __arm_shared_za __arm_preserves_za {
   return svread_hor_za32_u32_vg2(3, base);
 }
 
@@ -680,7 +680,7 @@ svuint32x2_t test_svread_hor_za32_u32_vg2(uint32_t base) {
 // CPP-CHECK-NEXT:    [[TMP4:%.*]] = tail call <vscale x 8 x float> @llvm.vector.insert.nxv8f32.nxv4f32(<vscale x 8 x float> [[TMP2]], <vscale x 4 x float> [[TMP3]], i64 4)
 // CPP-CHECK-NEXT:    ret <vscale x 8 x float> [[TMP4]]
 //
-svfloat32x2_t test_svread_hor_za32_f32_vg2(uint32_t base) {
+svfloat32x2_t test_svread_hor_za32_f32_vg2(uint32_t base) __arm_streaming __arm_shared_za __arm_preserves_za {
   return svread_hor_za32_f32_vg2(3, base);
 }
 
@@ -702,7 +702,7 @@ svfloat32x2_t test_svread_hor_za32_f32_vg2(uint32_t base) {
 // CPP-CHECK-NEXT:    [[TMP4:%.*]] = tail call <vscale x 8 x i32> @llvm.vector.insert.nxv8i32.nxv4i32(<vscale x 8 x i32> [[TMP2]], <vscale x 4 x i32> [[TMP3]], i64 4)
 // CPP-CHECK-NEXT:    ret <vscale x 8 x i32> [[TMP4]]
 //
-svint32x2_t test_svread_hor_za32_s32_vg2(uint32_t base) {
+svint32x2_t test_svread_hor_za32_s32_vg2(uint32_t base) __arm_streaming __arm_shared_za __arm_preserves_za {
   return svread_hor_za32_s32_vg2(3, base);
 }
 
@@ -724,7 +724,7 @@ svint32x2_t test_svread_hor_za32_s32_vg2(uint32_t base) {
 // CPP-CHECK-NEXT:    [[TMP4:%.*]] = tail call <vscale x 8 x i32> @llvm.vector.insert.nxv8i32.nxv4i32(<vscale x 8 x i32> [[TMP2]], <vscale x 4 x i32> [[TMP3]], i64 4)
 // CPP-CHECK-NEXT:    ret <vscale x 8 x i32> [[TMP4]]
 //
-svuint32x2_t test_svread_ver_za32_u32_vg2(uint32_t base) {
+svuint32x2_t test_svread_ver_za32_u32_vg2(uint32_t base) __arm_streaming __arm_shared_za __arm_preserves_za {
   return svread_ver_za32_u32_vg2(3, base);
 }
 
@@ -746,7 +746,7 @@ svuint32x2_t test_svread_ver_za32_u32_vg2(uint32_t base) {
 // CPP-CHECK-NEXT:    [[TMP4:%.*]] = tail call <vscale x 8 x float> @llvm.vector.insert.nxv8f32.nxv4f32(<vscale x 8 x float> [[TMP2]], <vscale x 4 x float> [[TMP3]], i64 4)
 // CPP-CHECK-NEXT:    ret <vscale x 8 x float> [[TMP4]]
 //
-svfloat32x2_t test_svread_ver_za32_f32_vg2(uint32_t base) {
+svfloat32x2_t test_svread_ver_za32_f32_vg2(uint32_t base) __arm_streaming __arm_shared_za __arm_preserves_za {
   return svread_ver_za32_f32_vg2(3, base);
 }
 
@@ -768,7 +768,7 @@ svfloat32x2_t test_svread_ver_za32_f32_vg2(uint32_t base) {
 // CPP-CHECK-NEXT:    [[TMP4:%.*]] = tail call <vscale x 8 x i32> @llvm.vector.insert.nxv8i32.nxv4i32(<vscale x 8 x i32> [[TMP2]], <vscale x 4 x i32> [[TMP3]], i64 4)
 // CPP-CHECK-NEXT:    ret <vscale x 8 x i32> [[TMP4]]
 //
-svint32x2_t test_svread_ver_za32_s32_vg2(uint32_t base) {
+svint32x2_t test_svread_ver_za32_s32_vg2(uint32_t base) __arm_streaming __arm_shared_za __arm_preserves_za {
   return svread_ver_za32_s32_vg2(3, base);
 }
 
@@ -798,7 +798,7 @@ svint32x2_t test_svread_ver_za32_s32_vg2(uint32_t base) {
 // CPP-CHECK-NEXT:    [[TMP8:%.*]] = tail call <vscale x 16 x i32> @llvm.vector.insert.nxv16i32.nxv4i32(<vscale x 16 x i32> [[TMP6]], <vscale x 4 x i32> [[TMP7]], i64 12)
 // CPP-CHECK-NEXT:    ret <vscale x 16 x i32> [[TMP8]]
 //
-svuint32x4_t test_svread_hor_za32_u32_vg4(uint32_t base) {
+svuint32x4_t test_svread_hor_za32_u32_vg4(uint32_t base) __arm_streaming __arm_shared_za __arm_preserves_za {
   return svread_hor_za32_u32_vg4(3, base);
 }
 
@@ -828,7 +828,7 @@ svuint32x4_t test_svread_hor_za32_u32_vg4(uint32_t base) {
 // CPP-CHECK-NEXT:    [[TMP8:%.*]] = tail call <vscale x 16 x float> @llvm.vector.insert.nxv16f32.nxv4f32(<vscale x 16 x float> [[TMP6]], <vscale x 4 x float> [[TMP7]], i64 12)
 // CPP-CHECK-NEXT:    ret <vscale x 16 x float> [[TMP8]]
 //
-svfloat32x4_t test_svread_hor_za32_f32_vg4(uint32_t base) {
+svfloat32x4_t test_svread_hor_za32_f32_vg4(uint32_t base) __arm_streaming __arm_shared_za __arm_preserves_za {
   return svread_hor_za32_f32_vg4(3, base);
 }
 
@@ -858,7 +858,7 @@ svfloat32x4_t test_svread_hor_za32_f32_vg4(uint32_t base) {
 // CPP-CHECK-NEXT:    [[TMP8:%.*]] = tail call <vscale x 16 x i32> @llvm.vector.insert.nxv16i32.nxv4i32(<vscale x 16 x i32> [[TMP6]], <vscale x 4 x i32> [[TMP7]], i64 12)
 // CPP-CHECK-NEXT:    ret <vscale x 16 x i32> [[TMP8]]
 //
-svint32x4_t test_svread_hor_za32_s32_vg4(uint32_t base) {
+svint32x4_t test_svread_hor_za32_s32_vg4(uint32_t base) __arm_streaming __arm_shared_za __arm_preserves_za {
   return svread_hor_za32_s32_vg4(3, base);
 }
 
@@ -888,7 +888,7 @@ svint32x4_t test_svread_hor_za32_s32_vg4(uint32_t base) {
 // CPP-CHECK-NEXT:    [[TMP8:%.*]] = tail call <vscale x 16 x i32> @llvm.vector.insert.nxv16i32.nxv4i32(<vscale x 16 x i32> [[TMP6]], <vscale x 4 x i32> [[TMP7]], i64 12)
 // CPP-CHECK-NEXT:    ret <vscale x 16 x i32> [[TMP8]]
 //
-svuint32x4_t test_svread_ver_za32_u32_vg4(uint32_t base) {
+svuint32x4_t test_svread_ver_za32_u32_vg4(uint32_t base) __arm_streaming __arm_shared_za __arm_preserves_za {
   return svread_ver_za32_u32_vg4(3, base);
 }
 
@@ -918,7 +918,7 @@ svuint32x4_t test_svread_ver_za32_u32_vg4(uint32_t base) {
 // CPP-CHECK-NEXT:    [[TMP8:%.*]] = tail call <vscale x 16 x float> @llvm.vector.insert.nxv16f32.nxv4f32(<vscale x 16 x float> [[TMP6]], <vscale x 4 x float> [[TMP7]], i64 12)
 // CPP-CHECK-NEXT:    ret <vscale x 16 x float> [[TMP8]]
 //
-svfloat32x4_t test_svread_ver_za32_f32_vg4(uint32_t base) {
+svfloat32x4_t test_svread_ver_za32_f32_vg4(uint32_t base) __arm_streaming __arm_shared_za __arm_preserves_za {
   return svread_ver_za32_f32_vg4(3, base);
 }
 
@@ -948,7 +948,7 @@ svfloat32x4_t test_svread_ver_za32_f32_vg4(uint32_t base) {
 // CPP-CHECK-NEXT:    [[TMP8:%.*]] = tail call <vscale x 16 x i32> @llvm.vector.insert.nxv16i32.nxv4i32(<vscale x 16 x i32> [[TMP6]], <vscale x 4 x i32> [[TMP7]], i64 12)
 // CPP-CHECK-NEXT:    ret <vscale x 16 x i32> [[TMP8]]
 //
-svint32x4_t test_svread_ver_za32_s32_vg4(uint32_t base) {
+svint32x4_t test_svread_ver_za32_s32_vg4(uint32_t base) __arm_streaming __arm_shared_za __arm_preserves_za {
   return svread_ver_za32_s32_vg4(3, base);
 }
 
@@ -970,7 +970,7 @@ svint32x4_t test_svread_ver_za32_s32_vg4(uint32_t base) {
 // CPP-CHECK-NEXT:    [[TMP4:%.*]] = tail call <vscale x 4 x i64> @llvm.vector.insert.nxv4i64.nxv2i64(<vscale x 4 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP3]], i64 2)
 // CPP-CHECK-NEXT:    ret <vscale x 4 x i64> [[TMP4]]
 //
-svuint64x2_t test_svread_hor_za64_u64_vg2(uint32_t base) {
+svuint64x2_t test_svread_hor_za64_u64_vg2(uint32_t base) __arm_streaming __arm_shared_za __arm_preserves_za {
   return svread_hor_za64_u64_vg2(7, base);
 }
 
@@ -992,7 +992,7 @@ svuint64x2_t test_svread_hor_za64_u64_vg2(uint32_t base) {
 // CPP-CHECK-NEXT:    [[TMP4:%.*]] = tail call <vscale x 4 x double> @llvm.vector.insert.nxv4f64.nxv2f64(<vscale x 4 x double> [[TMP2]], <vscale x 2 x double> [[TMP3]], i64 2)
 // CPP-CHECK-NEXT:    ret <vscale x 4 x double> [[TMP4]]
 //
-svfloat64x2_t test_svread_hor_za64_f64_vg2(uint32_t base) {
+svfloat64x2_t test_svread_hor_za64_f64_vg2(uint32_t base) __arm_streaming __arm_shared_za __arm_preserves_za {
   return svread_hor_za64_f64_vg2(7, base);
 }
 
@@ -1014,7 +1014,7 @@ svfloat64x2_t test_svread_hor_za64_f64_vg2(uint32_t base) {
 // CPP-CHECK-NEXT:    [[TMP4:%.*]] = tail call <vscale x 4 x i64> @llvm.vector.insert.nxv4i64.nxv2i64(<vscale x 4 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP3]], i64 2)
 // CPP-CHECK-NEXT:    ret <vscale x 4 x i64> [[TMP4]]
 //
-svint64x2_t test_svread_hor_za64_s64_vg2(uint32_t base) {
+svint64x2_t test_svread_hor_za64_s64_vg2(uint32_t base) __arm_streaming __arm_shared_za __arm_preserves_za {
   return svread_hor_za64_s64_vg2(7, base);
 }
 
@@ -1036,7 +1036,7 @@ svint64x2_t test_svread_hor_za64_s64_vg2(uint32_t base) {
 // CPP-CHECK-NEXT:    [[TMP4:%.*]] = tail call <vscale x 4 x i64> @llvm.vector.insert.nxv4i64.nxv2i64(<vscale x 4 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP3]], i64 2)
 // CPP-CHECK-NEXT:    ret <vscale x 4 x i64> [[TMP4]]
 //
-svuint64x2_t test_svread_ver_za64_u64_vg2(uint32_t base) {
+svuint64x2_t test_svread_ver_za64_u64_vg2(uint32_t base) __arm_streaming __arm_shared_za __arm_preserves_za {
   return svread_ver_za64_u64_vg2(7, base);
 }
 
@@ -1058,7 +1058,7 @@ svuint64x2_t test_svread_ver_za64_u64_vg2(uint32_t base) {
 // CPP-CHECK-NEXT:    [[TMP4:%.*]] = tail call <vscale x 4 x double> @llvm.vector.insert.nxv4f64.nxv2f64(<vscale x 4 x double> [[TMP2]], <vscale x 2 x double> [[TMP3]], i64 2)
 // CPP-CHECK-NEXT:    ret <vscale x 4 x double> [[TMP4]]
 //
-svfloat64x2_t test_svread_ver_za64_f64_vg2(uint32_t base) {
+svfloat64x2_t test_svread_ver_za64_f64_vg2(uint32_t base) __arm_streaming __arm_shared_za __arm_preserves_za {
   return svread_ver_za64_f64_vg2(7, base);
 }
 
@@ -1080,7 +1080,7 @@ svfloat64x2_t test_svread_ver_za64_f64_vg2(uint32_t base) {
 // CPP-CHECK-NEXT:    [[TMP4:%.*]] = tail call <vscale x 4 x i64> @llvm.vector.insert.nxv4i64.nxv2i64(<vscale x 4 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP3]], i64 2)
 // CPP-CHECK-NEXT:    ret <vscale x 4 x i64> [[TMP4]]
 //
-svint64x2_t test_svread_ver_za64_s64_vg2(uint32_t base) {
+svint64x2_t test_svread_ver_za64_s64_vg2(uint32_t base) __arm_streaming __arm_shared_za __arm_preserves_za {
   return svread_ver_za64_s64_vg2(7, base);
 }
 
@@ -1110,7 +1110,7 @@ svint64x2_t test_svread_ver_za64_s64_vg2(uint32_t base) {
 // CPP-CHECK-NEXT:    [[TMP8:%.*]] = tail call <vscale x 8 x i64> @llvm.vector.insert.nxv8i64.nxv2i64(<vscale x 8 x i64> [[TMP6]], <vscale x 2 x i64> [[TMP7]], i64 6)
 // CPP-CHECK-NEXT:    ret <vscale x 8 x i64> [[TMP8]]
 //
-svuint64x4_t test_svread_hor_za64_u64_vg4(uint32_t base) {
+svuint64x4_t test_svread_hor_za64_u64_vg4(uint32_t base) __arm_streaming __arm_shared_za __arm_preserves_za {
   return svread_hor_za64_u64_vg4(7, base);
 }
 
@@ -1140,7 +1140,7 @@ svuint64x4_t test_svread_hor_za64_u64_vg4(uint32_t base) {
 // CPP-CHECK-NEXT:    [[TMP8:%.*]] = tail call <vscale x 8 x double> @llvm.vector.insert.nxv8f64.nxv2f64(<vscale x 8 x double> [[TMP6]], <vscale x 2 x double> [[TMP7]], i64 6)
 // CPP-CHECK-NEXT:    ret <vscale x 8 x double> [[TMP8]]
 //
-svfloat64x4_t test_svread_hor_za64_f64_vg4(uint32_t base) {
+svfloat64x4_t test_svread_hor_za64_f64_vg4(uint32_t base) __arm_streaming __arm_shared_za __arm_preserves_za {
   return svread_hor_za64_f64_vg4(7, base);
 }
 
@@ -1170,7 +1170,7 @@ svfloat64x4_t test_svread_hor_za64_f64_vg4(uint32_t base) {
 // CPP-CHECK-NEXT:    [[TMP8:%.*]] = tail call <vscale x 8 x i64> @llvm.vector.insert.nxv8i64.nxv2i64(<vscale x 8 x i64> [[TMP6]], <vscale x 2 x i64> [[TMP7]], i64 6)
 // CPP-CHECK-NEXT:    ret <vscale x 8 x i64> [[TMP8]]
 //
-svint64x4_t test_svread_hor_za64_s64_vg4(uint32_t base) {
+svint64x4_t test_svread_hor_za64_s64_vg4(uint32_t base) __arm_streaming __arm_shared_za __arm_preserves_za {
   return svread_hor_za64_s64_vg4(7, base);
 }
 
@@ -1200,7 +1200,7 @@ svint64x4_t test_svread_hor_za64_s64_vg4(uint32_t base) {
 // CPP-CHECK-NEXT:    [[TMP8:%.*]] = tail call <vscale x 8 x i64> @llvm.vector.insert.nxv8i64.nxv2i64(<vscale x 8 x i64> [[TMP6]], <vscale x 2 x i64> [[TMP7]], i64 6)
 // CPP-CHECK-NEXT:    ret <vscale x 8 x i64> [[TMP8]]
 //
-svuint64x4_t test_svread_ver_za64_u64_vg4(uint32_t base) {
+svuint64x4_t test_svread_ver_za64_u64_vg4(uint32_t base) __arm_streaming __arm_shared_za __arm_preserves_za {
   return svread_ver_za64_u64_vg4(7, base);
 }
 
@@ -1230,7 +1230,7 @@ svuint64x4_t test_svread_ver_za64_u64_vg4(uint32_t base) {
 // CPP-CHECK-NEXT:    [[TMP8:%.*]] = tail call <vscale x 8 x double> @llvm.vector.insert.nxv8f64.nxv2f64(<vscale x 8 x double> [[TMP6]], <vscale x 2 x double> [[TMP7]], i64 6)
 // CPP-CHECK-NEXT:    ret <vscale x 8 x double> [[TMP8]]
 //
-svfloat64x4_t test_svread_ver_za64_f64_vg4(uint32_t base) {
+svfloat64x4_t test_svread_ver_za64_f64_vg4(uint32_t base) __arm_streaming __arm_shared_za __arm_preserves_za {
   return svread_ver_za64_f64_vg4(7, base);
 }
 
@@ -1260,7 +1260,7 @@ svfloat64x4_t test_svread_ver_za64_f64_vg4(uint32_t base) {
 // CPP-CHECK-NEXT:    [[TMP8:%.*]] = tail call <vscale x 8 x i64> @llvm.vector.insert.nxv8i64.nxv2i64(<vscale x 8 x i64> [[TMP6]], <vscale x 2 x i64> [[TMP7]], i64 6)
 // CPP-CHECK-NEXT:    ret <vscale x 8 x i64> [[TMP8]]
 //
-svint64x4_t test_svread_ver_za64_s64_vg4(uint32_t base) {
+svint64x4_t test_svread_ver_za64_s64_vg4(uint32_t base) __arm_streaming __arm_shared_za __arm_preserves_za {
   return svread_ver_za64_s64_vg4(7, base);
 }
 
@@ -1282,7 +1282,7 @@ svint64x4_t test_svread_ver_za64_s64_vg4(uint32_t base) {
 // CPP-CHECK-NEXT:    [[TMP4:%.*]] = tail call <vscale x 4 x i64> @llvm.vector.insert.nxv4i64.nxv2i64(<vscale x 4 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP3]], i64 2)
 // CPP-CHECK-NEXT:    ret <vscale x 4 x i64> [[TMP4]]
 //
-svuint64x2_t test_svread_za64_u64_vg1x2(uint32_t base) {
+svuint64x2_t test_svread_za64_u64_vg1x2(uint32_t base) __arm_streaming __arm_shared_za __arm_preserves_za {
   return svread_za64_u64_vg1x2(base);
 }
 
@@ -1304,7 +1304,7 @@ svuint64x2_t test_svread_za64_u64_vg1x2(uint32_t base) {
 // CPP-CHECK-NEXT:    [[TMP4:%.*]] = tail call <vscale x 4 x double> @llvm.vector.insert.nxv4f64.nxv2f64(<vscale x 4 x double> [[TMP2]], <vscale x 2 x double> [[TMP3]], i64 2)
 // CPP-CHECK-NEXT:    ret <vscale x 4 x double> [[TMP4]]
 //
-svfloat64x2_t test_svread_za64_f64_vg1x2(uint32_t base) {
+svfloat64x2_t test_svread_za64_f64_vg1x2(uint32_t base) __arm_streaming __arm_shared_za __arm_preserves_za {
   return svread_za64_f64_vg1x2(base);
 }
 
@@ -1326,7 +1326,7 @@ svfloat64x2_t test_svread_za64_f64_vg1x2(uint32_t base) {
 // CPP-CHECK-NEXT:    [[TMP4:%.*]] = tail call <vscale x 4 x i64> @llvm.vector.insert.nxv4i64.nxv2i64(<vscale x 4 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP3]], i64 2)
 // CPP-CHECK-NEXT:    ret <vscale x 4 x i64> [[TMP4]]
 //
-svint64x2_t test_svread_za64_s64_vg1x2(uint32_t base) {
+svint64x2_t test_svread_za64_s64_vg1x2(uint32_t base) __arm_streaming __arm_shared_za __arm_preserves_za {
   return svread_za64_s64_vg1x2(base);
 }
 
@@ -1356,7 +1356,7 @@ svint64x2_t test_svread_za64_s64_vg1x2(uint32_t base) {
 // CPP-CHECK-NEXT:    [[TMP8:%.*]] = tail call <vscale x 8 x i64> @llvm.vector.insert.nxv8i64.nxv2i64(<vscale x 8 x i64> [[TMP6]], <vscale x 2 x i64> [[TMP7]], i64 6)
 // CPP-CHECK-NEXT:    ret <vscale x 8 x i64> [[TMP8]]
 //
-svuint64x4_t test_svread_za64_u64_vg1x4(uint32_t base) {
+svuint64x4_t test_svread_za64_u64_vg1x4(uint32_t base) __arm_streaming __arm_shared_za __arm_preserves_za {
   return svread_za64_u64_vg1x4(base);
 }
 
@@ -1386,7 +1386,7 @@ svuint64x4_t test_svread_za64_u64_vg1x4(uint32_t base) {
 // CPP-CHECK-NEXT:    [[TMP8:%.*]] = tail call <vscale x 8 x double> @llvm.vector.insert.nxv8f64.nxv2f64(<vscale x 8 x double> [[TMP6]], <vscale x 2 x double> [[TMP7]], i64 6)
 // CPP-CHECK-NEXT:    ret <vscale x 8 x double> [[TMP8]]
 //
-svfloat64x4_t test_svread_za64_f64_vg1x4(uint32_t base) {
+svfloat64x4_t test_svread_za64_f64_vg1x4(uint32_t base) __arm_streaming __arm_shared_za __arm_preserves_za {
   return svread_za64_f64_vg1x4(base);
 }
 
@@ -1416,7 +1416,7 @@ svfloat64x4_t test_svread_za64_f64_vg1x4(uint32_t base) {
 // CPP-CHECK-NEXT:    [[TMP8:%.*]] = tail call <vscale x 8 x i64> @llvm.vector.insert.nxv8i64.nxv2i64(<vscale x 8 x i64> [[TMP6]], <vscale x 2 x i64> [[TMP7]], i64 6)
 // CPP-CHECK-NEXT:    ret <vscale x 8 x i64> [[TMP8]]
 //
-svint64x4_t test_svread_za64_s64_vg1x4(uint32_t base) {
+svint64x4_t test_svread_za64_s64_vg1x4(uint32_t base) __arm_streaming __arm_shared_za __arm_preserves_za {
   return svread_za64_s64_vg1x4(base);
 }
 
diff --git a/clang/test/CodeGen/aarch64-sme2-intrinsics/acle_sme2_write.c b/clang/test/CodeGen/aarch64-sme2-intrinsics/acle_sme2_write.c
index 745eb7f3edb56..99aa221834724 100644
--- a/clang/test/CodeGen/aarch64-sme2-intrinsics/acle_sme2_write.c
+++ b/clang/test/CodeGen/aarch64-sme2-intrinsics/acle_sme2_write.c
@@ -30,7 +30,7 @@
 // CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.write.ver.vg2.nxv16i8(i32 0, i32 [[BASE:%.*]], <vscale x 16 x i8> [[TMP0]], <vscale x 16 x i8> [[TMP1]])
 // CPP-CHECK-NEXT:    ret void
 //
-void test_svwrite_ver_za8_u8_vg2(uint32_t base, svuint8x2_t val) {
+void test_svwrite_ver_za8_u8_vg2(uint32_t base, svuint8x2_t val) __arm_streaming __arm_shared_za {
   SVE_ACLE_FUNC(svwrite_ver_za8,_u8,_vg2,)(0, base, val);
 }
 
@@ -48,7 +48,7 @@ void test_svwrite_ver_za8_u8_vg2(uint32_t base, svuint8x2_t val) {
 // CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.write.ver.vg2.nxv16i8(i32 0, i32 [[BASE:%.*]], <vscale x 16 x i8> [[TMP0]], <vscale x 16 x i8> [[TMP1]])
 // CPP-CHECK-NEXT:    ret void
 //
-void test_svwrite_ver_za8_s8_vg2(uint32_t base, svint8x2_t val) {
+void test_svwrite_ver_za8_s8_vg2(uint32_t base, svint8x2_t val) __arm_streaming __arm_shared_za {
   SVE_ACLE_FUNC(svwrite_ver_za8,_s8,_vg2,)(0, base, val);
 }
 
@@ -66,7 +66,7 @@ void test_svwrite_ver_za8_s8_vg2(uint32_t base, svint8x2_t val) {
 // CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.write.hor.vg2.nxv16i8(i32 0, i32 [[BASE:%.*]], <vscale x 16 x i8> [[TMP0]], <vscale x 16 x i8> [[TMP1]])
 // CPP-CHECK-NEXT:    ret void
 //
-void test_svwrite_hor_za8_u8_vg2(uint32_t base, svuint8x2_t val) {
+void test_svwrite_hor_za8_u8_vg2(uint32_t base, svuint8x2_t val) __arm_streaming __arm_shared_za {
   SVE_ACLE_FUNC(svwrite_hor_za8,_u8,_vg2,)(0, base, val);
 }
 
@@ -84,7 +84,7 @@ void test_svwrite_hor_za8_u8_vg2(uint32_t base, svuint8x2_t val) {
 // CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.write.hor.vg2.nxv16i8(i32 0, i32 [[BASE:%.*]], <vscale x 16 x i8> [[TMP0]], <vscale x 16 x i8> [[TMP1]])
 // CPP-CHECK-NEXT:    ret void
 //
-void test_svwrite_hor_za8_s8_vg2(uint32_t base, svint8x2_t val) {
+void test_svwrite_hor_za8_s8_vg2(uint32_t base, svint8x2_t val) __arm_streaming __arm_shared_za {
   SVE_ACLE_FUNC(svwrite_hor_za8,_s8,_vg2,)(0, base, val);
 }
 
@@ -106,7 +106,7 @@ void test_svwrite_hor_za8_s8_vg2(uint32_t base, svint8x2_t val) {
 // CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.write.hor.vg4.nxv16i8(i32 0, i32 [[BASE:%.*]], <vscale x 16 x i8> [[TMP0]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP2]], <vscale x 16 x i8> [[TMP3]])
 // CPP-CHECK-NEXT:    ret void
 //
-void test_svwrite_hor_za8_u8_vg4(uint32_t base, svuint8x4_t val) {
+void test_svwrite_hor_za8_u8_vg4(uint32_t base, svuint8x4_t val) __arm_streaming __arm_shared_za {
   SVE_ACLE_FUNC(svwrite_hor_za8,_u8,_vg4,)(0, base, val);
 }
 
@@ -128,7 +128,7 @@ void test_svwrite_hor_za8_u8_vg4(uint32_t base, svuint8x4_t val) {
 // CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.write.hor.vg4.nxv16i8(i32 0, i32 [[BASE:%.*]], <vscale x 16 x i8> [[TMP0]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP2]], <vscale x 16 x i8> [[TMP3]])
 // CPP-CHECK-NEXT:    ret void
 //
-void test_svwrite_hor_za8_s8_vg4(uint32_t base, svint8x4_t val) {
+void test_svwrite_hor_za8_s8_vg4(uint32_t base, svint8x4_t val) __arm_streaming __arm_shared_za {
   SVE_ACLE_FUNC(svwrite_hor_za8,_s8,_vg4,)(0, base, val);
 }
 
@@ -150,7 +150,7 @@ void test_svwrite_hor_za8_s8_vg4(uint32_t base, svint8x4_t val) {
 // CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.write.ver.vg4.nxv16i8(i32 0, i32 [[BASE:%.*]], <vscale x 16 x i8> [[TMP0]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP2]], <vscale x 16 x i8> [[TMP3]])
 // CPP-CHECK-NEXT:    ret void
 //
-void test_svwrite_ver_za8_u8_vg4(uint32_t base, svuint8x4_t val) {
+void test_svwrite_ver_za8_u8_vg4(uint32_t base, svuint8x4_t val) __arm_streaming __arm_shared_za {
   SVE_ACLE_FUNC(svwrite_ver_za8,_u8,_vg4,)(0, base, val);
 }
 
@@ -172,7 +172,7 @@ void test_svwrite_ver_za8_u8_vg4(uint32_t base, svuint8x4_t val) {
 // CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.write.ver.vg4.nxv16i8(i32 0, i32 [[BASE:%.*]], <vscale x 16 x i8> [[TMP0]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP2]], <vscale x 16 x i8> [[TMP3]])
 // CPP-CHECK-NEXT:    ret void
 //
-void test_svwrite_ver_za8_s8_vg4(uint32_t base, svint8x4_t val) {
+void test_svwrite_ver_za8_s8_vg4(uint32_t base, svint8x4_t val) __arm_streaming __arm_shared_za {
   SVE_ACLE_FUNC(svwrite_ver_za8,_s8,_vg4,)(0, base, val);
 }
 
@@ -190,7 +190,7 @@ void test_svwrite_ver_za8_s8_vg4(uint32_t base, svint8x4_t val) {
 // CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.write.hor.vg2.nxv8i16(i32 1, i32 [[BASE:%.*]], <vscale x 8 x i16> [[TMP0]], <vscale x 8 x i16> [[TMP1]])
 // CPP-CHECK-NEXT:    ret void
 //
-void test_svwrite_hor_za16_u16_vg2(uint32_t base, svuint16x2_t val) {
+void test_svwrite_hor_za16_u16_vg2(uint32_t base, svuint16x2_t val) __arm_streaming __arm_shared_za {
   SVE_ACLE_FUNC(svwrite_hor_za16,_u16,_vg2,)(1, base, val);
 }
 
@@ -208,7 +208,7 @@ void test_svwrite_hor_za16_u16_vg2(uint32_t base, svuint16x2_t val) {
 // CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.write.hor.vg2.nxv8bf16(i32 1, i32 [[BASE:%.*]], <vscale x 8 x bfloat> [[TMP0]], <vscale x 8 x bfloat> [[TMP1]])
 // CPP-CHECK-NEXT:    ret void
 //
-void test_svwrite_hor_za16_bf16_vg2(uint32_t base, svbfloat16x2_t val) {
+void test_svwrite_hor_za16_bf16_vg2(uint32_t base, svbfloat16x2_t val) __arm_streaming __arm_shared_za {
   SVE_ACLE_FUNC(svwrite_hor_za16,_bf16,_vg2,)(1, base, val);
 }
 
@@ -226,7 +226,7 @@ void test_svwrite_hor_za16_bf16_vg2(uint32_t base, svbfloat16x2_t val) {
 // CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.write.hor.vg2.nxv8f16(i32 1, i32 [[BASE:%.*]], <vscale x 8 x half> [[TMP0]], <vscale x 8 x half> [[TMP1]])
 // CPP-CHECK-NEXT:    ret void
 //
-void test_svwrite_hor_za16_f16_vg2(uint32_t base, svfloat16x2_t val) {
+void test_svwrite_hor_za16_f16_vg2(uint32_t base, svfloat16x2_t val) __arm_streaming __arm_shared_za {
   SVE_ACLE_FUNC(svwrite_hor_za16,_f16,_vg2,)(1, base, val);
 }
 
@@ -244,7 +244,7 @@ void test_svwrite_hor_za16_f16_vg2(uint32_t base, svfloat16x2_t val) {
 // CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.write.hor.vg2.nxv8i16(i32 1, i32 [[BASE:%.*]], <vscale x 8 x i16> [[TMP0]], <vscale x 8 x i16> [[TMP1]])
 // CPP-CHECK-NEXT:    ret void
 //
-void test_svwrite_hor_za16_s16_vg2(uint32_t base, svint16x2_t val) {
+void test_svwrite_hor_za16_s16_vg2(uint32_t base, svint16x2_t val) __arm_streaming __arm_shared_za {
   SVE_ACLE_FUNC(svwrite_hor_za16,_s16,_vg2,)(1, base, val);
 }
 
@@ -262,7 +262,7 @@ void test_svwrite_hor_za16_s16_vg2(uint32_t base, svint16x2_t val) {
 // CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.write.ver.vg2.nxv8i16(i32 1, i32 [[BASE:%.*]], <vscale x 8 x i16> [[TMP0]], <vscale x 8 x i16> [[TMP1]])
 // CPP-CHECK-NEXT:    ret void
 //
-void test_svwrite_ver_za16_u16_vg2(uint32_t base, svuint16x2_t val) {
+void test_svwrite_ver_za16_u16_vg2(uint32_t base, svuint16x2_t val) __arm_streaming __arm_shared_za {
   SVE_ACLE_FUNC(svwrite_ver_za16,_u16,_vg2,)(1, base, val);
 }
 
@@ -280,7 +280,7 @@ void test_svwrite_ver_za16_u16_vg2(uint32_t base, svuint16x2_t val) {
 // CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.write.ver.vg2.nxv8bf16(i32 1, i32 [[BASE:%.*]], <vscale x 8 x bfloat> [[TMP0]], <vscale x 8 x bfloat> [[TMP1]])
 // CPP-CHECK-NEXT:    ret void
 //
-void test_svwrite_ver_za16_bf16_vg2(uint32_t base, svbfloat16x2_t val) {
+void test_svwrite_ver_za16_bf16_vg2(uint32_t base, svbfloat16x2_t val) __arm_streaming __arm_shared_za {
   SVE_ACLE_FUNC(svwrite_ver_za16,_bf16,_vg2,)(1, base, val);
 }
 
@@ -298,7 +298,7 @@ void test_svwrite_ver_za16_bf16_vg2(uint32_t base, svbfloat16x2_t val) {
 // CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.write.ver.vg2.nxv8f16(i32 1, i32 [[BASE:%.*]], <vscale x 8 x half> [[TMP0]], <vscale x 8 x half> [[TMP1]])
 // CPP-CHECK-NEXT:    ret void
 //
-void test_svwrite_ver_za16_f16_vg2(uint32_t base, svfloat16x2_t val) {
+void test_svwrite_ver_za16_f16_vg2(uint32_t base, svfloat16x2_t val) __arm_streaming __arm_shared_za {
   SVE_ACLE_FUNC(svwrite_ver_za16,_f16,_vg2,)(1, base, val);
 }
 
@@ -316,7 +316,7 @@ void test_svwrite_ver_za16_f16_vg2(uint32_t base, svfloat16x2_t val) {
 // CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.write.ver.vg2.nxv8i16(i32 1, i32 [[BASE:%.*]], <vscale x 8 x i16> [[TMP0]], <vscale x 8 x i16> [[TMP1]])
 // CPP-CHECK-NEXT:    ret void
 //
-void test_svwrite_ver_za16_s16_vg2(uint32_t base, svint16x2_t val) {
+void test_svwrite_ver_za16_s16_vg2(uint32_t base, svint16x2_t val) __arm_streaming __arm_shared_za {
   SVE_ACLE_FUNC(svwrite_ver_za16,_s16,_vg2,)(1, base, val);
 }
 
@@ -338,7 +338,7 @@ void test_svwrite_ver_za16_s16_vg2(uint32_t base, svint16x2_t val) {
 // CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.write.hor.vg4.nxv8i16(i32 1, i32 [[BASE:%.*]], <vscale x 8 x i16> [[TMP0]], <vscale x 8 x i16> [[TMP1]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP3]])
 // CPP-CHECK-NEXT:    ret void
 //
-void test_svwrite_hor_za16_u16_vg4(uint32_t base, svuint16x4_t val) {
+void test_svwrite_hor_za16_u16_vg4(uint32_t base, svuint16x4_t val) __arm_streaming __arm_shared_za {
   SVE_ACLE_FUNC(svwrite_hor_za16,_u16,_vg4,)(1, base, val);
 }
 
@@ -360,7 +360,7 @@ void test_svwrite_hor_za16_u16_vg4(uint32_t base, svuint16x4_t val) {
 // CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.write.hor.vg4.nxv8bf16(i32 1, i32 [[BASE:%.*]], <vscale x 8 x bfloat> [[TMP0]], <vscale x 8 x bfloat> [[TMP1]], <vscale x 8 x bfloat> [[TMP2]], <vscale x 8 x bfloat> [[TMP3]])
 // CPP-CHECK-NEXT:    ret void
 //
-void test_svwrite_hor_za16_bf16_vg4(uint32_t base, svbfloat16x4_t val) {
+void test_svwrite_hor_za16_bf16_vg4(uint32_t base, svbfloat16x4_t val) __arm_streaming __arm_shared_za {
   SVE_ACLE_FUNC(svwrite_hor_za16,_bf16,_vg4,)(1, base, val);
 }
 
@@ -382,7 +382,7 @@ void test_svwrite_hor_za16_bf16_vg4(uint32_t base, svbfloat16x4_t val) {
 // CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.write.hor.vg4.nxv8f16(i32 1, i32 [[BASE:%.*]], <vscale x 8 x half> [[TMP0]], <vscale x 8 x half> [[TMP1]], <vscale x 8 x half> [[TMP2]], <vscale x 8 x half> [[TMP3]])
 // CPP-CHECK-NEXT:    ret void
 //
-void test_svwrite_hor_za16_f16_vg4(uint32_t base, svfloat16x4_t val) {
+void test_svwrite_hor_za16_f16_vg4(uint32_t base, svfloat16x4_t val) __arm_streaming __arm_shared_za {
   SVE_ACLE_FUNC(svwrite_hor_za16,_f16,_vg4,)(1, base, val);
 }
 
@@ -404,7 +404,7 @@ void test_svwrite_hor_za16_f16_vg4(uint32_t base, svfloat16x4_t val) {
 // CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.write.hor.vg4.nxv8i16(i32 1, i32 [[BASE:%.*]], <vscale x 8 x i16> [[TMP0]], <vscale x 8 x i16> [[TMP1]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP3]])
 // CPP-CHECK-NEXT:    ret void
 //
-void test_svwrite_hor_za16_s16_vg4(uint32_t base, svint16x4_t val) {
+void test_svwrite_hor_za16_s16_vg4(uint32_t base, svint16x4_t val) __arm_streaming __arm_shared_za {
   SVE_ACLE_FUNC(svwrite_hor_za16,_s16,_vg4,)(1, base, val);
 }
 
@@ -426,7 +426,7 @@ void test_svwrite_hor_za16_s16_vg4(uint32_t base, svint16x4_t val) {
 // CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.write.ver.vg4.nxv8i16(i32 1, i32 [[BASE:%.*]], <vscale x 8 x i16> [[TMP0]], <vscale x 8 x i16> [[TMP1]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP3]])
 // CPP-CHECK-NEXT:    ret void
 //
-void test_svwrite_ver_za16_u16_vg4(uint32_t base, svuint16x4_t val) {
+void test_svwrite_ver_za16_u16_vg4(uint32_t base, svuint16x4_t val) __arm_streaming __arm_shared_za {
   SVE_ACLE_FUNC(svwrite_ver_za16,_u16,_vg4,)(1, base, val);
 }
 
@@ -448,7 +448,7 @@ void test_svwrite_ver_za16_u16_vg4(uint32_t base, svuint16x4_t val) {
 // CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.write.ver.vg4.nxv8bf16(i32 1, i32 [[BASE:%.*]], <vscale x 8 x bfloat> [[TMP0]], <vscale x 8 x bfloat> [[TMP1]], <vscale x 8 x bfloat> [[TMP2]], <vscale x 8 x bfloat> [[TMP3]])
 // CPP-CHECK-NEXT:    ret void
 //
-void test_svwrite_ver_za16_bf16_vg4(uint32_t base, svbfloat16x4_t val) {
+void test_svwrite_ver_za16_bf16_vg4(uint32_t base, svbfloat16x4_t val) __arm_streaming __arm_shared_za {
   SVE_ACLE_FUNC(svwrite_ver_za16,_bf16,_vg4,)(1, base, val);
 }
 
@@ -470,7 +470,7 @@ void test_svwrite_ver_za16_bf16_vg4(uint32_t base, svbfloat16x4_t val) {
 // CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.write.ver.vg4.nxv8f16(i32 1, i32 [[BASE:%.*]], <vscale x 8 x half> [[TMP0]], <vscale x 8 x half> [[TMP1]], <vscale x 8 x half> [[TMP2]], <vscale x 8 x half> [[TMP3]])
 // CPP-CHECK-NEXT:    ret void
 //
-void test_svwrite_ver_za16_f16_vg4(uint32_t base, svfloat16x4_t val) {
+void test_svwrite_ver_za16_f16_vg4(uint32_t base, svfloat16x4_t val) __arm_streaming __arm_shared_za {
   SVE_ACLE_FUNC(svwrite_ver_za16,_f16,_vg4,)(1, base, val);
 }
 
@@ -492,7 +492,7 @@ void test_svwrite_ver_za16_f16_vg4(uint32_t base, svfloat16x4_t val) {
 // CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.write.ver.vg4.nxv8i16(i32 1, i32 [[BASE:%.*]], <vscale x 8 x i16> [[TMP0]], <vscale x 8 x i16> [[TMP1]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP3]])
 // CPP-CHECK-NEXT:    ret void
 //
-void test_svwrite_ver_za16_s16_vg4(uint32_t base, svint16x4_t val) {
+void test_svwrite_ver_za16_s16_vg4(uint32_t base, svint16x4_t val) __arm_streaming __arm_shared_za {
   SVE_ACLE_FUNC(svwrite_ver_za16,_s16,_vg4,)(1, base, val);
 }
 
@@ -510,7 +510,7 @@ void test_svwrite_ver_za16_s16_vg4(uint32_t base, svint16x4_t val) {
 // CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.write.hor.vg2.nxv4i32(i32 3, i32 [[BASE:%.*]], <vscale x 4 x i32> [[TMP0]], <vscale x 4 x i32> [[TMP1]])
 // CPP-CHECK-NEXT:    ret void
 //
-void test_svwrite_hor_za32_u32_vg2(uint32_t base, svuint32x2_t val) {
+void test_svwrite_hor_za32_u32_vg2(uint32_t base, svuint32x2_t val) __arm_streaming __arm_shared_za {
   SVE_ACLE_FUNC(svwrite_hor_za32,_u32,_vg2,)(3, base, val);
 }
 
@@ -528,7 +528,7 @@ void test_svwrite_hor_za32_u32_vg2(uint32_t base, svuint32x2_t val) {
 // CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.write.hor.vg2.nxv4f32(i32 3, i32 [[BASE:%.*]], <vscale x 4 x float> [[TMP0]], <vscale x 4 x float> [[TMP1]])
 // CPP-CHECK-NEXT:    ret void
 //
-void test_svwrite_hor_za32_f32_vg2(uint32_t base, svfloat32x2_t val) {
+void test_svwrite_hor_za32_f32_vg2(uint32_t base, svfloat32x2_t val) __arm_streaming __arm_shared_za {
   SVE_ACLE_FUNC(svwrite_hor_za32,_f32,_vg2,)(3, base, val);
 }
 
@@ -546,7 +546,7 @@ void test_svwrite_hor_za32_f32_vg2(uint32_t base, svfloat32x2_t val) {
 // CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.write.hor.vg2.nxv4i32(i32 3, i32 [[BASE:%.*]], <vscale x 4 x i32> [[TMP0]], <vscale x 4 x i32> [[TMP1]])
 // CPP-CHECK-NEXT:    ret void
 //
-void test_svwrite_hor_za32_s32_vg2(uint32_t base, svint32x2_t val) {
+void test_svwrite_hor_za32_s32_vg2(uint32_t base, svint32x2_t val) __arm_streaming __arm_shared_za {
   SVE_ACLE_FUNC(svwrite_hor_za32,_s32,_vg2,)(3, base, val);
 }
 
@@ -564,7 +564,7 @@ void test_svwrite_hor_za32_s32_vg2(uint32_t base, svint32x2_t val) {
 // CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.write.ver.vg2.nxv4i32(i32 3, i32 [[BASE:%.*]], <vscale x 4 x i32> [[TMP0]], <vscale x 4 x i32> [[TMP1]])
 // CPP-CHECK-NEXT:    ret void
 //
-void test_svwrite_ver_za32_u32_vg2(uint32_t base, svuint32x2_t val) {
+void test_svwrite_ver_za32_u32_vg2(uint32_t base, svuint32x2_t val) __arm_streaming __arm_shared_za {
   SVE_ACLE_FUNC(svwrite_ver_za32,_u32,_vg2,)(3, base, val);
 }
 
@@ -582,7 +582,7 @@ void test_svwrite_ver_za32_u32_vg2(uint32_t base, svuint32x2_t val) {
 // CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.write.ver.vg2.nxv4f32(i32 3, i32 [[BASE:%.*]], <vscale x 4 x float> [[TMP0]], <vscale x 4 x float> [[TMP1]])
 // CPP-CHECK-NEXT:    ret void
 //
-void test_svwrite_ver_za32_f32_vg2(uint32_t base, svfloat32x2_t val) {
+void test_svwrite_ver_za32_f32_vg2(uint32_t base, svfloat32x2_t val) __arm_streaming __arm_shared_za {
   SVE_ACLE_FUNC(svwrite_ver_za32,_f32,_vg2,)(3, base, val);
 }
 
@@ -600,7 +600,7 @@ void test_svwrite_ver_za32_f32_vg2(uint32_t base, svfloat32x2_t val) {
 // CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.write.ver.vg2.nxv4i32(i32 3, i32 [[BASE:%.*]], <vscale x 4 x i32> [[TMP0]], <vscale x 4 x i32> [[TMP1]])
 // CPP-CHECK-NEXT:    ret void
 //
-void test_svwrite_ver_za32_s32_vg2(uint32_t base, svint32x2_t val) {
+void test_svwrite_ver_za32_s32_vg2(uint32_t base, svint32x2_t val) __arm_streaming __arm_shared_za {
   SVE_ACLE_FUNC(svwrite_ver_za32,_s32,_vg2,)(3, base, val);
 }
 
@@ -622,7 +622,7 @@ void test_svwrite_ver_za32_s32_vg2(uint32_t base, svint32x2_t val) {
 // CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.write.hor.vg4.nxv4i32(i32 3, i32 [[BASE:%.*]], <vscale x 4 x i32> [[TMP0]], <vscale x 4 x i32> [[TMP1]], <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> [[TMP3]])
 // CPP-CHECK-NEXT:    ret void
 //
-void test_svwrite_hor_za32_u32_vg4(uint32_t base, svuint32x4_t val) {
+void test_svwrite_hor_za32_u32_vg4(uint32_t base, svuint32x4_t val) __arm_streaming __arm_shared_za {
   SVE_ACLE_FUNC(svwrite_hor_za32,_u32,_vg4,)(3, base, val);
 }
 
@@ -644,7 +644,7 @@ void test_svwrite_hor_za32_u32_vg4(uint32_t base, svuint32x4_t val) {
 // CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.write.hor.vg4.nxv4f32(i32 3, i32 [[BASE:%.*]], <vscale x 4 x float> [[TMP0]], <vscale x 4 x float> [[TMP1]], <vscale x 4 x float> [[TMP2]], <vscale x 4 x float> [[TMP3]])
 // CPP-CHECK-NEXT:    ret void
 //
-void test_svwrite_hor_za32_f32_vg4(uint32_t base, svfloat32x4_t val) {
+void test_svwrite_hor_za32_f32_vg4(uint32_t base, svfloat32x4_t val) __arm_streaming __arm_shared_za {
   SVE_ACLE_FUNC(svwrite_hor_za32,_f32,_vg4,)(3, base, val);
 }
 
@@ -666,7 +666,7 @@ void test_svwrite_hor_za32_f32_vg4(uint32_t base, svfloat32x4_t val) {
 // CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.write.hor.vg4.nxv4i32(i32 3, i32 [[BASE:%.*]], <vscale x 4 x i32> [[TMP0]], <vscale x 4 x i32> [[TMP1]], <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> [[TMP3]])
 // CPP-CHECK-NEXT:    ret void
 //
-void test_svwrite_hor_za32_s32_vg4(uint32_t base, svint32x4_t val) {
+void test_svwrite_hor_za32_s32_vg4(uint32_t base, svint32x4_t val) __arm_streaming __arm_shared_za {
   SVE_ACLE_FUNC(svwrite_hor_za32,_s32,_vg4,)(3, base, val);
 }
 
@@ -688,7 +688,7 @@ void test_svwrite_hor_za32_s32_vg4(uint32_t base, svint32x4_t val) {
 // CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.write.ver.vg4.nxv4i32(i32 3, i32 [[BASE:%.*]], <vscale x 4 x i32> [[TMP0]], <vscale x 4 x i32> [[TMP1]], <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> [[TMP3]])
 // CPP-CHECK-NEXT:    ret void
 //
-void test_svwrite_ver_za32_u32_vg4(uint32_t base, svuint32x4_t val) {
+void test_svwrite_ver_za32_u32_vg4(uint32_t base, svuint32x4_t val) __arm_streaming __arm_shared_za {
   SVE_ACLE_FUNC(svwrite_ver_za32,_u32,_vg4,)(3, base, val);
 }
 
@@ -710,7 +710,7 @@ void test_svwrite_ver_za32_u32_vg4(uint32_t base, svuint32x4_t val) {
 // CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.write.ver.vg4.nxv4f32(i32 3, i32 [[BASE:%.*]], <vscale x 4 x float> [[TMP0]], <vscale x 4 x float> [[TMP1]], <vscale x 4 x float> [[TMP2]], <vscale x 4 x float> [[TMP3]])
 // CPP-CHECK-NEXT:    ret void
 //
-void test_svwrite_ver_za32_f32_vg4(uint32_t base, svfloat32x4_t val) {
+void test_svwrite_ver_za32_f32_vg4(uint32_t base, svfloat32x4_t val) __arm_streaming __arm_shared_za {
   SVE_ACLE_FUNC(svwrite_ver_za32,_f32,_vg4,)(3, base, val);
 }
 
@@ -732,7 +732,7 @@ void test_svwrite_ver_za32_f32_vg4(uint32_t base, svfloat32x4_t val) {
 // CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.write.ver.vg4.nxv4i32(i32 3, i32 [[BASE:%.*]], <vscale x 4 x i32> [[TMP0]], <vscale x 4 x i32> [[TMP1]], <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> [[TMP3]])
 // CPP-CHECK-NEXT:    ret void
 //
-void test_svwrite_ver_za32_s32_vg4(uint32_t base, svint32x4_t val) {
+void test_svwrite_ver_za32_s32_vg4(uint32_t base, svint32x4_t val) __arm_streaming __arm_shared_za {
   SVE_ACLE_FUNC(svwrite_ver_za32,_s32,_vg4,)(3, base, val);
 }
 
@@ -750,7 +750,7 @@ void test_svwrite_ver_za32_s32_vg4(uint32_t base, svint32x4_t val) {
 // CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.write.hor.vg2.nxv2i64(i32 7, i32 [[BASE:%.*]], <vscale x 2 x i64> [[TMP0]], <vscale x 2 x i64> [[TMP1]])
 // CPP-CHECK-NEXT:    ret void
 //
-void test_svwrite_hor_za64_u64_vg2(uint32_t base, svuint64x2_t val) {
+void test_svwrite_hor_za64_u64_vg2(uint32_t base, svuint64x2_t val) __arm_streaming __arm_shared_za {
   SVE_ACLE_FUNC(svwrite_hor_za64,_u64,_vg2,)(7, base, val);
 }
 
@@ -768,7 +768,7 @@ void test_svwrite_hor_za64_u64_vg2(uint32_t base, svuint64x2_t val) {
 // CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.write.hor.vg2.nxv2f64(i32 7, i32 [[BASE:%.*]], <vscale x 2 x double> [[TMP0]], <vscale x 2 x double> [[TMP1]])
 // CPP-CHECK-NEXT:    ret void
 //
-void test_svwrite_hor_za64_f64_vg2(uint32_t base, svfloat64x2_t val) {
+void test_svwrite_hor_za64_f64_vg2(uint32_t base, svfloat64x2_t val) __arm_streaming __arm_shared_za {
   SVE_ACLE_FUNC(svwrite_hor_za64,_f64,_vg2,)(7, base, val);
 }
 
@@ -786,7 +786,7 @@ void test_svwrite_hor_za64_f64_vg2(uint32_t base, svfloat64x2_t val) {
 // CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.write.hor.vg2.nxv2i64(i32 7, i32 [[BASE:%.*]], <vscale x 2 x i64> [[TMP0]], <vscale x 2 x i64> [[TMP1]])
 // CPP-CHECK-NEXT:    ret void
 //
-void test_svwrite_hor_za64_s64_vg2(uint32_t base, svint64x2_t val) {
+void test_svwrite_hor_za64_s64_vg2(uint32_t base, svint64x2_t val) __arm_streaming __arm_shared_za {
   SVE_ACLE_FUNC(svwrite_hor_za64,_s64,_vg2,)(7, base, val);
 }
 
@@ -804,7 +804,7 @@ void test_svwrite_hor_za64_s64_vg2(uint32_t base, svint64x2_t val) {
 // CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.write.ver.vg2.nxv2i64(i32 7, i32 [[BASE:%.*]], <vscale x 2 x i64> [[TMP0]], <vscale x 2 x i64> [[TMP1]])
 // CPP-CHECK-NEXT:    ret void
 //
-void test_svwrite_ver_za64_u64_vg2(uint32_t base, svuint64x2_t val) {
+void test_svwrite_ver_za64_u64_vg2(uint32_t base, svuint64x2_t val) __arm_streaming __arm_shared_za {
   SVE_ACLE_FUNC(svwrite_ver_za64,_u64,_vg2,)(7, base, val);
 }
 
@@ -822,7 +822,7 @@ void test_svwrite_ver_za64_u64_vg2(uint32_t base, svuint64x2_t val) {
 // CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.write.ver.vg2.nxv2f64(i32 7, i32 [[BASE:%.*]], <vscale x 2 x double> [[TMP0]], <vscale x 2 x double> [[TMP1]])
 // CPP-CHECK-NEXT:    ret void
 //
-void test_svwrite_ver_za64_f64_vg2(uint32_t base, svfloat64x2_t val) {
+void test_svwrite_ver_za64_f64_vg2(uint32_t base, svfloat64x2_t val) __arm_streaming __arm_shared_za {
   SVE_ACLE_FUNC(svwrite_ver_za64,_f64,_vg2,)(7, base, val);
 }
 
@@ -840,7 +840,7 @@ void test_svwrite_ver_za64_f64_vg2(uint32_t base, svfloat64x2_t val) {
 // CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.write.ver.vg2.nxv2i64(i32 7, i32 [[BASE:%.*]], <vscale x 2 x i64> [[TMP0]], <vscale x 2 x i64> [[TMP1]])
 // CPP-CHECK-NEXT:    ret void
 //
-void test_svwrite_ver_za64_s64_vg2(uint32_t base, svint64x2_t val) {
+void test_svwrite_ver_za64_s64_vg2(uint32_t base, svint64x2_t val) __arm_streaming __arm_shared_za {
   SVE_ACLE_FUNC(svwrite_ver_za64,_s64,_vg2,)(7, base, val);
 }
 
@@ -862,7 +862,7 @@ void test_svwrite_ver_za64_s64_vg2(uint32_t base, svint64x2_t val) {
 // CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.write.hor.vg4.nxv2i64(i32 7, i32 [[BASE:%.*]], <vscale x 2 x i64> [[TMP0]], <vscale x 2 x i64> [[TMP1]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP3]])
 // CPP-CHECK-NEXT:    ret void
 //
-void test_svwrite_hor_za64_u64_vg4(uint32_t base, svuint64x4_t val) {
+void test_svwrite_hor_za64_u64_vg4(uint32_t base, svuint64x4_t val) __arm_streaming __arm_shared_za {
   SVE_ACLE_FUNC(svwrite_hor_za64,_u64,_vg4,)(7, base, val);
 }
 
@@ -884,7 +884,7 @@ void test_svwrite_hor_za64_u64_vg4(uint32_t base, svuint64x4_t val) {
 // CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.write.ver.vg4.nxv2i64(i32 7, i32 [[BASE:%.*]], <vscale x 2 x i64> [[TMP0]], <vscale x 2 x i64> [[TMP1]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP3]])
 // CPP-CHECK-NEXT:    ret void
 //
-void test_svwrite_ver_za64_u64_vg4(uint32_t base, svuint64x4_t val) {
+void test_svwrite_ver_za64_u64_vg4(uint32_t base, svuint64x4_t val) __arm_streaming __arm_shared_za {
   SVE_ACLE_FUNC(svwrite_ver_za64,_u64,_vg4,)(7, base, val);
 }
 
@@ -906,7 +906,7 @@ void test_svwrite_ver_za64_u64_vg4(uint32_t base, svuint64x4_t val) {
 // CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.write.hor.vg4.nxv2f64(i32 7, i32 [[BASE:%.*]], <vscale x 2 x double> [[TMP0]], <vscale x 2 x double> [[TMP1]], <vscale x 2 x double> [[TMP2]], <vscale x 2 x double> [[TMP3]])
 // CPP-CHECK-NEXT:    ret void
 //
-void test_svwrite_hor_za64_f64_vg4(uint32_t base, svfloat64x4_t val) {
+void test_svwrite_hor_za64_f64_vg4(uint32_t base, svfloat64x4_t val) __arm_streaming __arm_shared_za {
   SVE_ACLE_FUNC(svwrite_hor_za64,_f64,_vg4,)(7, base, val);
 }
 
@@ -928,7 +928,7 @@ void test_svwrite_hor_za64_f64_vg4(uint32_t base, svfloat64x4_t val) {
 // CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.write.hor.vg4.nxv2i64(i32 7, i32 [[BASE:%.*]], <vscale x 2 x i64> [[TMP0]], <vscale x 2 x i64> [[TMP1]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP3]])
 // CPP-CHECK-NEXT:    ret void
 //
-void test_svwrite_hor_za64_s64_vg4(uint32_t base, svint64x4_t val) {
+void test_svwrite_hor_za64_s64_vg4(uint32_t base, svint64x4_t val) __arm_streaming __arm_shared_za {
   SVE_ACLE_FUNC(svwrite_hor_za64,_s64,_vg4,)(7, base, val);
 }
 
@@ -950,7 +950,7 @@ void test_svwrite_hor_za64_s64_vg4(uint32_t base, svint64x4_t val) {
 // CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.write.ver.vg4.nxv2f64(i32 7, i32 [[BASE:%.*]], <vscale x 2 x double> [[TMP0]], <vscale x 2 x double> [[TMP1]], <vscale x 2 x double> [[TMP2]], <vscale x 2 x double> [[TMP3]])
 // CPP-CHECK-NEXT:    ret void
 //
-void test_svwrite_ver_za64_f64_vg4(uint32_t base, svfloat64x4_t val) {
+void test_svwrite_ver_za64_f64_vg4(uint32_t base, svfloat64x4_t val) __arm_streaming __arm_shared_za {
   SVE_ACLE_FUNC(svwrite_ver_za64,_f64,_vg4,)(7, base, val);
 }
 
@@ -972,7 +972,7 @@ void test_svwrite_ver_za64_f64_vg4(uint32_t base, svfloat64x4_t val) {
 // CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.write.ver.vg4.nxv2i64(i32 7, i32 [[BASE:%.*]], <vscale x 2 x i64> [[TMP0]], <vscale x 2 x i64> [[TMP1]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP3]])
 // CPP-CHECK-NEXT:    ret void
 //
-void test_svwrite_ver_za64_s64_vg4(uint32_t base, svint64x4_t val) {
+void test_svwrite_ver_za64_s64_vg4(uint32_t base, svint64x4_t val) __arm_streaming __arm_shared_za {
   SVE_ACLE_FUNC(svwrite_ver_za64,_s64,_vg4,)(7, base, val);
 }
 
@@ -990,7 +990,7 @@ void test_svwrite_ver_za64_s64_vg4(uint32_t base, svint64x4_t val) {
 // CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.write.vg1x2.nxv2i64(i32 [[BASE:%.*]], <vscale x 2 x i64> [[TMP0]], <vscale x 2 x i64> [[TMP1]])
 // CPP-CHECK-NEXT:    ret void
 //
-void test_svwrite_za64_u64_vg1x2(uint32_t base, svuint64x2_t val) {
+void test_svwrite_za64_u64_vg1x2(uint32_t base, svuint64x2_t val) __arm_streaming __arm_shared_za {
   SVE_ACLE_FUNC(svwrite_za64,_u64,_vg1x2,)(base, val);
 }
 
@@ -1008,7 +1008,7 @@ void test_svwrite_za64_u64_vg1x2(uint32_t base, svuint64x2_t val) {
 // CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.write.vg1x2.nxv2f64(i32 [[BASE:%.*]], <vscale x 2 x double> [[TMP0]], <vscale x 2 x double> [[TMP1]])
 // CPP-CHECK-NEXT:    ret void
 //
-void test_svwrite_za64_f64_vg1x2(uint32_t base, svfloat64x2_t val) {
+void test_svwrite_za64_f64_vg1x2(uint32_t base, svfloat64x2_t val) __arm_streaming __arm_shared_za {
   SVE_ACLE_FUNC(svwrite_za64,_f64,_vg1x2,)(base, val);
 }
 
@@ -1026,7 +1026,7 @@ void test_svwrite_za64_f64_vg1x2(uint32_t base, svfloat64x2_t val) {
 // CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.write.vg1x2.nxv2i64(i32 [[BASE:%.*]], <vscale x 2 x i64> [[TMP0]], <vscale x 2 x i64> [[TMP1]])
 // CPP-CHECK-NEXT:    ret void
 //
-void test_svwrite_za64_s64_vg1x2(uint32_t base, svint64x2_t val) {
+void test_svwrite_za64_s64_vg1x2(uint32_t base, svint64x2_t val) __arm_streaming __arm_shared_za {
   SVE_ACLE_FUNC(svwrite_za64,_s64,_vg1x2,)(base, val);
 }
 
@@ -1048,7 +1048,7 @@ void test_svwrite_za64_s64_vg1x2(uint32_t base, svint64x2_t val) {
 // CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.write.vg1x4.nxv2i64(i32 [[BASE:%.*]], <vscale x 2 x i64> [[TMP0]], <vscale x 2 x i64> [[TMP1]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP3]])
 // CPP-CHECK-NEXT:    ret void
 //
-void test_svwrite_za64_u64_vg1x4(uint32_t base, svuint64x4_t val) {
+void test_svwrite_za64_u64_vg1x4(uint32_t base, svuint64x4_t val) __arm_streaming __arm_shared_za {
   SVE_ACLE_FUNC(svwrite_za64,_u64,_vg1x4,)(base, val);
 }
 
@@ -1070,7 +1070,7 @@ void test_svwrite_za64_u64_vg1x4(uint32_t base, svuint64x4_t val) {
 // CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.write.vg1x4.nxv2f64(i32 [[BASE:%.*]], <vscale x 2 x double> [[TMP0]], <vscale x 2 x double> [[TMP1]], <vscale x 2 x double> [[TMP2]], <vscale x 2 x double> [[TMP3]])
 // CPP-CHECK-NEXT:    ret void
 //
-void test_svwrite_za64_f64_vg1x4(uint32_t base, svfloat64x4_t val) {
+void test_svwrite_za64_f64_vg1x4(uint32_t base, svfloat64x4_t val) __arm_streaming __arm_shared_za {
   SVE_ACLE_FUNC(svwrite_za64,_f64,_vg1x4,)(base, val);
 }
 
@@ -1092,6 +1092,6 @@ void test_svwrite_za64_f64_vg1x4(uint32_t base, svfloat64x4_t val) {
 // CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.write.vg1x4.nxv2i64(i32 [[BASE:%.*]], <vscale x 2 x i64> [[TMP0]], <vscale x 2 x i64> [[TMP1]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP3]])
 // CPP-CHECK-NEXT:    ret void
 //
-void test_svwrite_za64_s64_vg1x4(uint32_t base, svint64x4_t val) {
+void test_svwrite_za64_s64_vg1x4(uint32_t base, svint64x4_t val) __arm_streaming __arm_shared_za {
   SVE_ACLE_FUNC(svwrite_za64,_s64,_vg1x4,)(base, val);
 }
diff --git a/clang/test/Sema/aarch64-sme2-intrinsics/acle_sme2_imm.cpp b/clang/test/Sema/aarch64-sme2-intrinsics/acle_sme2_imm.cpp
index 77b1133f55101..63a7a40c3273a 100644
--- a/clang/test/Sema/aarch64-sme2-intrinsics/acle_sme2_imm.cpp
+++ b/clang/test/Sema/aarch64-sme2-intrinsics/acle_sme2_imm.cpp
@@ -5,7 +5,7 @@
 
 #include <arm_sme_draft_spec_subject_to_change.h>
 
-void test_multivector_read(uint32_t base) {
+void test_multivector_read(uint32_t base) __arm_streaming __arm_shared_za __arm_preserves_za {
 
   // Test Tile Range
   svread_hor_za8_u8_vg2(1, base); // expected-error {{argument value 1 is outside the valid range [0, 0]}}
@@ -32,7 +32,7 @@ void test_multivector_read(uint32_t base) {
 void test_multivector_write(uint32_t base, svuint8x2_t v8x2, svuint8x4_t v8x4,
                             svuint16x2_t v16x2, svuint16x4_t v16x4,
                             svuint32x2_t v32x2, svuint32x4_t v32x4,
-                            svuint64x2_t v64x2, svuint64x4_t v64x4) {
+                            svuint64x2_t v64x2, svuint64x4_t v64x4) __arm_streaming __arm_shared_za {
 
   // Test Tile Range
   svwrite_hor_za8_u8_vg2(1, base, v8x2); // expected-error {{argument value 1 is outside the valid range [0, 0]}}

From 884f9e91dc8e3f674c5f8fb5ed5a7ae1e6e01bb4 Mon Sep 17 00:00:00 2001
From: Kerry McLaughlin <kerry.mclaughlin@arm.com>
Date: Tue, 5 Dec 2023 13:51:24 +0000
Subject: [PATCH 3/5] - Added overloaded types for svread & svwrite.   (See
 https://github.com/ARM-software/acle/pull/278)

---
 clang/include/clang/Basic/arm_sme.td          |  57 +--
 .../aarch64-sme2-intrinsics/acle_sme2_read.c  | 469 +++++++++++++++++-
 .../aarch64-sme2-intrinsics/acle_sme2_write.c | 360 ++++++++++++++
 llvm/lib/Target/AArch64/SMEInstrFormats.td    |  20 +-
 .../AArch64/sme2-intrinsics-extract-mova.ll   | 168 +++++++
 .../AArch64/sme2-intrinsics-insert-mova.ll    | 204 ++++++++
 6 files changed, 1241 insertions(+), 37 deletions(-)

diff --git a/clang/include/clang/Basic/arm_sme.td b/clang/include/clang/Basic/arm_sme.td
index b97629d943334..a46e2bdb34c6f 100644
--- a/clang/include/clang/Basic/arm_sme.td
+++ b/clang/include/clang/Basic/arm_sme.td
@@ -302,46 +302,39 @@ defm SVSUB : ZAAddSub<"sub">;
 // SME2 - MOVA
 
 //
-// 2 and 4 vector-group read/write intrinsics.
+// Single, 2 and 4 vector-group read/write intrinsics.
 //
 
-multiclass WriteHV_VG<string n, string t, string i, list<ImmCheck> checks> {
-  let TargetGuard = "sme2" in {
-    def NAME # _VG2_H : Inst<"svwrite_hor_" # n # "_vg2", "vim2", t, MergeNone, i # "_hor_vg2", [IsSharedZA, IsStreaming], checks>;
-    def NAME # _VG2_V : Inst<"svwrite_ver_" # n # "_vg2", "vim2", t, MergeNone, i # "_ver_vg2", [IsSharedZA, IsStreaming], checks>;
-    def NAME # _VG4_H : Inst<"svwrite_hor_" # n # "_vg4", "vim4", t, MergeNone, i # "_hor_vg4", [IsSharedZA, IsStreaming], checks>;
-    def NAME # _VG4_V : Inst<"svwrite_ver_" # n # "_vg4", "vim4", t, MergeNone, i # "_ver_vg4", [IsSharedZA, IsStreaming], checks>;
-  }
+multiclass ZAWrite_VG<string n, string t, string i, list<ImmCheck> checks> {
+  def NAME # _VG2_H : Inst<"svwrite_hor_" # n # "_vg2",   "vim2", t, MergeNone, i # "_hor_vg2", [IsSharedZA, IsStreaming], checks>;
+  def NAME # _VG2_V : Inst<"svwrite_ver_" # n # "_vg2",   "vim2", t, MergeNone, i # "_ver_vg2", [IsSharedZA, IsStreaming], checks>;
+  def NAME # _VG4_H : Inst<"svwrite_hor_" # n # "_vg4",   "vim4", t, MergeNone, i # "_hor_vg4", [IsSharedZA, IsStreaming], checks>;
+  def NAME # _VG4_V : Inst<"svwrite_ver_" # n # "_vg4",   "vim4", t, MergeNone, i # "_ver_vg4", [IsSharedZA, IsStreaming], checks>;
+  def NAME # _VG1x2 : Inst<"svwrite_"     # n # "_vg1x2", "vm2",  t, MergeNone, i # "_vg1x2",   [IsSharedZA, IsStreaming], []>;
+  def NAME # _VG1x4 : Inst<"svwrite_"     # n # "_vg1x4", "vm4",  t, MergeNone, i # "_vg1x4",   [IsSharedZA, IsStreaming], []>;
 }
 
-defm SVWRITE_ZA8  : WriteHV_VG<"za8[_{d}]",  "cUc",   "aarch64_sme_write", [ImmCheck<0, ImmCheck0_0>]>;
-defm SVWRITE_ZA16 : WriteHV_VG<"za16[_{d}]", "sUshb", "aarch64_sme_write", [ImmCheck<0, ImmCheck0_1>]>;
-defm SVWRITE_ZA32 : WriteHV_VG<"za32[_{d}]", "iUif",  "aarch64_sme_write", [ImmCheck<0, ImmCheck0_3>]>;
-defm SVWRITE_ZA64 : WriteHV_VG<"za64[_{d}]", "lUld",  "aarch64_sme_write", [ImmCheck<0, ImmCheck0_7>]>;
-
-multiclass ReadHV_VG<string n, string t, string i, list<ImmCheck> checks> {
-  let TargetGuard = "sme2" in {
-    def NAME # _VG2_H : Inst<"svread_hor_" # n # "_vg2", "2im", t, MergeNone, i # "_hor_vg2", [IsSharedZA, IsPreservesZA, IsStreaming], checks>;
-    def NAME # _VG2_V : Inst<"svread_ver_" # n # "_vg2", "2im", t, MergeNone, i # "_ver_vg2", [IsSharedZA, IsPreservesZA, IsStreaming], checks>;
-    def NAME # _VG4_H : Inst<"svread_hor_" # n # "_vg4", "4im", t, MergeNone, i # "_hor_vg4", [IsSharedZA, IsPreservesZA, IsStreaming], checks>;
-    def NAME # _VG4_V : Inst<"svread_ver_" # n # "_vg4", "4im", t, MergeNone, i # "_ver_vg4", [IsSharedZA, IsPreservesZA, IsStreaming], checks>;
-  }
+let TargetGuard = "sme2" in {
+  defm SVWRITE_ZA8  : ZAWrite_VG<"za8[_{d}]",  "cUc",   "aarch64_sme_write", [ImmCheck<0, ImmCheck0_0>]>;
+  defm SVWRITE_ZA16 : ZAWrite_VG<"za16[_{d}]", "sUshb", "aarch64_sme_write", [ImmCheck<0, ImmCheck0_1>]>;
+  defm SVWRITE_ZA32 : ZAWrite_VG<"za32[_{d}]", "iUif",  "aarch64_sme_write", [ImmCheck<0, ImmCheck0_3>]>;
+  defm SVWRITE_ZA64 : ZAWrite_VG<"za64[_{d}]", "lUld",  "aarch64_sme_write", [ImmCheck<0, ImmCheck0_7>]>;
 }
 
-defm SVREAD_ZA8  : ReadHV_VG<"za8_{d}",  "cUc",   "aarch64_sme_read", [ImmCheck<0, ImmCheck0_0>]>;
-defm SVREAD_ZA16 : ReadHV_VG<"za16_{d}", "sUshb", "aarch64_sme_read", [ImmCheck<0, ImmCheck0_1>]>;
-defm SVREAD_ZA32 : ReadHV_VG<"za32_{d}", "iUif",  "aarch64_sme_read", [ImmCheck<0, ImmCheck0_3>]>;
-defm SVREAD_ZA64 : ReadHV_VG<"za64_{d}", "lUld",  "aarch64_sme_read", [ImmCheck<0, ImmCheck0_7>]>;
-
-//
-// Single vector-group read/write intrinsics.
-//
+multiclass ZARead_VG<string n, string t, string i, list<ImmCheck> checks> {
+  def NAME # _VG2_H : Inst<"svread_hor_" # n # "_vg2",   "2im", t, MergeNone, i # "_hor_vg2", [IsSharedZA, IsPreservesZA, IsStreaming], checks>;
+  def NAME # _VG2_V : Inst<"svread_ver_" # n # "_vg2",   "2im", t, MergeNone, i # "_ver_vg2", [IsSharedZA, IsPreservesZA, IsStreaming], checks>;
+  def NAME # _VG4_H : Inst<"svread_hor_" # n # "_vg4",   "4im", t, MergeNone, i # "_hor_vg4", [IsSharedZA, IsPreservesZA, IsStreaming], checks>;
+  def NAME # _VG4_V : Inst<"svread_ver_" # n # "_vg4",   "4im", t, MergeNone, i # "_ver_vg4", [IsSharedZA, IsPreservesZA, IsStreaming], checks>;
+  def NAME # _VG1x2 : Inst<"svread_"     # n # "_vg1x2", "2m",  t, MergeNone, i # "_vg1x2",   [IsSharedZA, IsPreservesZA, IsStreaming], []>;
+  def NAME # _VG1x4 : Inst<"svread_"     # n # "_vg1x4", "4m",  t, MergeNone, i # "_vg1x4",   [IsSharedZA, IsPreservesZA, IsStreaming], []>;
+}
 
 let TargetGuard = "sme2" in {
-  def SVWRITE_ZA64_VG1x2 : Inst<"svwrite_za64[_{d}]_vg1x2", "vm2", "lUld", MergeNone, "aarch64_sme_write_vg1x2", [IsSharedZA, IsStreaming], []>;
-  def SVWRITE_ZA64_VG1x4 : Inst<"svwrite_za64[_{d}]_vg1x4", "vm4", "lUld", MergeNone, "aarch64_sme_write_vg1x4", [IsSharedZA, IsStreaming], []>;
-  def SVREAD_ZA64_VG1x2  : Inst<"svread_za64_{d}_vg1x2",    "2m",  "lUld", MergeNone, "aarch64_sme_read_vg1x2",  [IsSharedZA, IsPreservesZA, IsStreaming], []>;
-  def SVREAD_ZA64_VG1x4  : Inst<"svread_za64_{d}_vg1x4",    "4m",  "lUld", MergeNone, "aarch64_sme_read_vg1x4",  [IsSharedZA, IsPreservesZA, IsStreaming], []>;
+  defm SVREAD_ZA8  : ZARead_VG<"za8_{d}",  "cUc",   "aarch64_sme_read", [ImmCheck<0, ImmCheck0_0>]>;
+  defm SVREAD_ZA16 : ZARead_VG<"za16_{d}", "sUshb", "aarch64_sme_read", [ImmCheck<0, ImmCheck0_1>]>;
+  defm SVREAD_ZA32 : ZARead_VG<"za32_{d}", "iUif",  "aarch64_sme_read", [ImmCheck<0, ImmCheck0_3>]>;
+  defm SVREAD_ZA64 : ZARead_VG<"za64_{d}", "lUld",  "aarch64_sme_read", [ImmCheck<0, ImmCheck0_7>]>;
 }
 
 //
diff --git a/clang/test/CodeGen/aarch64-sme2-intrinsics/acle_sme2_read.c b/clang/test/CodeGen/aarch64-sme2-intrinsics/acle_sme2_read.c
index 89700afca41e4..63abc9acfdf9c 100644
--- a/clang/test/CodeGen/aarch64-sme2-intrinsics/acle_sme2_read.c
+++ b/clang/test/CodeGen/aarch64-sme2-intrinsics/acle_sme2_read.c
@@ -1264,6 +1264,204 @@ svint64x4_t test_svread_ver_za64_s64_vg4(uint32_t base) __arm_streaming __arm_sh
   return svread_ver_za64_s64_vg4(7, base);
 }
 
+// CHECK-LABEL: @test_svread_za8_s8_vg1x2(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call { <vscale x 16 x i8>, <vscale x 16 x i8> } @llvm.aarch64.sme.read.vg1x2.nxv16i8(i32 [[BASE:%.*]])
+// CHECK-NEXT:    [[TMP1:%.*]] = extractvalue { <vscale x 16 x i8>, <vscale x 16 x i8> } [[TMP0]], 0
+// CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 32 x i8> @llvm.vector.insert.nxv32i8.nxv16i8(<vscale x 32 x i8> poison, <vscale x 16 x i8> [[TMP1]], i64 0)
+// CHECK-NEXT:    [[TMP3:%.*]] = extractvalue { <vscale x 16 x i8>, <vscale x 16 x i8> } [[TMP0]], 1
+// CHECK-NEXT:    [[TMP4:%.*]] = tail call <vscale x 32 x i8> @llvm.vector.insert.nxv32i8.nxv16i8(<vscale x 32 x i8> [[TMP2]], <vscale x 16 x i8> [[TMP3]], i64 16)
+// CHECK-NEXT:    ret <vscale x 32 x i8> [[TMP4]]
+//
+// CPP-CHECK-LABEL: @_Z24test_svread_za8_s8_vg1x2j(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call { <vscale x 16 x i8>, <vscale x 16 x i8> } @llvm.aarch64.sme.read.vg1x2.nxv16i8(i32 [[BASE:%.*]])
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = extractvalue { <vscale x 16 x i8>, <vscale x 16 x i8> } [[TMP0]], 0
+// CPP-CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 32 x i8> @llvm.vector.insert.nxv32i8.nxv16i8(<vscale x 32 x i8> poison, <vscale x 16 x i8> [[TMP1]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP3:%.*]] = extractvalue { <vscale x 16 x i8>, <vscale x 16 x i8> } [[TMP0]], 1
+// CPP-CHECK-NEXT:    [[TMP4:%.*]] = tail call <vscale x 32 x i8> @llvm.vector.insert.nxv32i8.nxv16i8(<vscale x 32 x i8> [[TMP2]], <vscale x 16 x i8> [[TMP3]], i64 16)
+// CPP-CHECK-NEXT:    ret <vscale x 32 x i8> [[TMP4]]
+//
+svint8x2_t test_svread_za8_s8_vg1x2(uint32_t base) __arm_streaming __arm_shared_za __arm_preserves_za {
+  return svread_za8_s8_vg1x2(base);
+}
+
+// CHECK-LABEL: @test_svread_za8_u8_vg1x2(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call { <vscale x 16 x i8>, <vscale x 16 x i8> } @llvm.aarch64.sme.read.vg1x2.nxv16i8(i32 [[BASE:%.*]])
+// CHECK-NEXT:    [[TMP1:%.*]] = extractvalue { <vscale x 16 x i8>, <vscale x 16 x i8> } [[TMP0]], 0
+// CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 32 x i8> @llvm.vector.insert.nxv32i8.nxv16i8(<vscale x 32 x i8> poison, <vscale x 16 x i8> [[TMP1]], i64 0)
+// CHECK-NEXT:    [[TMP3:%.*]] = extractvalue { <vscale x 16 x i8>, <vscale x 16 x i8> } [[TMP0]], 1
+// CHECK-NEXT:    [[TMP4:%.*]] = tail call <vscale x 32 x i8> @llvm.vector.insert.nxv32i8.nxv16i8(<vscale x 32 x i8> [[TMP2]], <vscale x 16 x i8> [[TMP3]], i64 16)
+// CHECK-NEXT:    ret <vscale x 32 x i8> [[TMP4]]
+//
+// CPP-CHECK-LABEL: @_Z24test_svread_za8_u8_vg1x2j(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call { <vscale x 16 x i8>, <vscale x 16 x i8> } @llvm.aarch64.sme.read.vg1x2.nxv16i8(i32 [[BASE:%.*]])
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = extractvalue { <vscale x 16 x i8>, <vscale x 16 x i8> } [[TMP0]], 0
+// CPP-CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 32 x i8> @llvm.vector.insert.nxv32i8.nxv16i8(<vscale x 32 x i8> poison, <vscale x 16 x i8> [[TMP1]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP3:%.*]] = extractvalue { <vscale x 16 x i8>, <vscale x 16 x i8> } [[TMP0]], 1
+// CPP-CHECK-NEXT:    [[TMP4:%.*]] = tail call <vscale x 32 x i8> @llvm.vector.insert.nxv32i8.nxv16i8(<vscale x 32 x i8> [[TMP2]], <vscale x 16 x i8> [[TMP3]], i64 16)
+// CPP-CHECK-NEXT:    ret <vscale x 32 x i8> [[TMP4]]
+//
+svuint8x2_t test_svread_za8_u8_vg1x2(uint32_t base) __arm_streaming __arm_shared_za __arm_preserves_za {
+  return svread_za8_u8_vg1x2(base);
+}
+
+// CHECK-LABEL: @test_svread_za16_s16_vg1x2(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call { <vscale x 8 x i16>, <vscale x 8 x i16> } @llvm.aarch64.sme.read.vg1x2.nxv8i16(i32 [[BASE:%.*]])
+// CHECK-NEXT:    [[TMP1:%.*]] = extractvalue { <vscale x 8 x i16>, <vscale x 8 x i16> } [[TMP0]], 0
+// CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 16 x i16> @llvm.vector.insert.nxv16i16.nxv8i16(<vscale x 16 x i16> poison, <vscale x 8 x i16> [[TMP1]], i64 0)
+// CHECK-NEXT:    [[TMP3:%.*]] = extractvalue { <vscale x 8 x i16>, <vscale x 8 x i16> } [[TMP0]], 1
+// CHECK-NEXT:    [[TMP4:%.*]] = tail call <vscale x 16 x i16> @llvm.vector.insert.nxv16i16.nxv8i16(<vscale x 16 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP3]], i64 8)
+// CHECK-NEXT:    ret <vscale x 16 x i16> [[TMP4]]
+//
+// CPP-CHECK-LABEL: @_Z26test_svread_za16_s16_vg1x2j(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call { <vscale x 8 x i16>, <vscale x 8 x i16> } @llvm.aarch64.sme.read.vg1x2.nxv8i16(i32 [[BASE:%.*]])
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = extractvalue { <vscale x 8 x i16>, <vscale x 8 x i16> } [[TMP0]], 0
+// CPP-CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 16 x i16> @llvm.vector.insert.nxv16i16.nxv8i16(<vscale x 16 x i16> poison, <vscale x 8 x i16> [[TMP1]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP3:%.*]] = extractvalue { <vscale x 8 x i16>, <vscale x 8 x i16> } [[TMP0]], 1
+// CPP-CHECK-NEXT:    [[TMP4:%.*]] = tail call <vscale x 16 x i16> @llvm.vector.insert.nxv16i16.nxv8i16(<vscale x 16 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP3]], i64 8)
+// CPP-CHECK-NEXT:    ret <vscale x 16 x i16> [[TMP4]]
+//
+svint16x2_t test_svread_za16_s16_vg1x2(uint32_t base) __arm_streaming __arm_shared_za __arm_preserves_za {
+  return svread_za16_s16_vg1x2(base);
+}
+
+// CHECK-LABEL: @test_svread_za16_u16_vg1x2(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call { <vscale x 8 x i16>, <vscale x 8 x i16> } @llvm.aarch64.sme.read.vg1x2.nxv8i16(i32 [[BASE:%.*]])
+// CHECK-NEXT:    [[TMP1:%.*]] = extractvalue { <vscale x 8 x i16>, <vscale x 8 x i16> } [[TMP0]], 0
+// CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 16 x i16> @llvm.vector.insert.nxv16i16.nxv8i16(<vscale x 16 x i16> poison, <vscale x 8 x i16> [[TMP1]], i64 0)
+// CHECK-NEXT:    [[TMP3:%.*]] = extractvalue { <vscale x 8 x i16>, <vscale x 8 x i16> } [[TMP0]], 1
+// CHECK-NEXT:    [[TMP4:%.*]] = tail call <vscale x 16 x i16> @llvm.vector.insert.nxv16i16.nxv8i16(<vscale x 16 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP3]], i64 8)
+// CHECK-NEXT:    ret <vscale x 16 x i16> [[TMP4]]
+//
+// CPP-CHECK-LABEL: @_Z26test_svread_za16_u16_vg1x2j(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call { <vscale x 8 x i16>, <vscale x 8 x i16> } @llvm.aarch64.sme.read.vg1x2.nxv8i16(i32 [[BASE:%.*]])
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = extractvalue { <vscale x 8 x i16>, <vscale x 8 x i16> } [[TMP0]], 0
+// CPP-CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 16 x i16> @llvm.vector.insert.nxv16i16.nxv8i16(<vscale x 16 x i16> poison, <vscale x 8 x i16> [[TMP1]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP3:%.*]] = extractvalue { <vscale x 8 x i16>, <vscale x 8 x i16> } [[TMP0]], 1
+// CPP-CHECK-NEXT:    [[TMP4:%.*]] = tail call <vscale x 16 x i16> @llvm.vector.insert.nxv16i16.nxv8i16(<vscale x 16 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP3]], i64 8)
+// CPP-CHECK-NEXT:    ret <vscale x 16 x i16> [[TMP4]]
+//
+svuint16x2_t test_svread_za16_u16_vg1x2(uint32_t base) __arm_streaming __arm_shared_za __arm_preserves_za {
+  return svread_za16_u16_vg1x2(base);
+}
+
+// CHECK-LABEL: @test_svread_za16_bf16_vg1x2(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call { <vscale x 8 x bfloat>, <vscale x 8 x bfloat> } @llvm.aarch64.sme.read.vg1x2.nxv8bf16(i32 [[BASE:%.*]])
+// CHECK-NEXT:    [[TMP1:%.*]] = extractvalue { <vscale x 8 x bfloat>, <vscale x 8 x bfloat> } [[TMP0]], 0
+// CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 16 x bfloat> @llvm.vector.insert.nxv16bf16.nxv8bf16(<vscale x 16 x bfloat> poison, <vscale x 8 x bfloat> [[TMP1]], i64 0)
+// CHECK-NEXT:    [[TMP3:%.*]] = extractvalue { <vscale x 8 x bfloat>, <vscale x 8 x bfloat> } [[TMP0]], 1
+// CHECK-NEXT:    [[TMP4:%.*]] = tail call <vscale x 16 x bfloat> @llvm.vector.insert.nxv16bf16.nxv8bf16(<vscale x 16 x bfloat> [[TMP2]], <vscale x 8 x bfloat> [[TMP3]], i64 8)
+// CHECK-NEXT:    ret <vscale x 16 x bfloat> [[TMP4]]
+//
+// CPP-CHECK-LABEL: @_Z27test_svread_za16_bf16_vg1x2j(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call { <vscale x 8 x bfloat>, <vscale x 8 x bfloat> } @llvm.aarch64.sme.read.vg1x2.nxv8bf16(i32 [[BASE:%.*]])
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = extractvalue { <vscale x 8 x bfloat>, <vscale x 8 x bfloat> } [[TMP0]], 0
+// CPP-CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 16 x bfloat> @llvm.vector.insert.nxv16bf16.nxv8bf16(<vscale x 16 x bfloat> poison, <vscale x 8 x bfloat> [[TMP1]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP3:%.*]] = extractvalue { <vscale x 8 x bfloat>, <vscale x 8 x bfloat> } [[TMP0]], 1
+// CPP-CHECK-NEXT:    [[TMP4:%.*]] = tail call <vscale x 16 x bfloat> @llvm.vector.insert.nxv16bf16.nxv8bf16(<vscale x 16 x bfloat> [[TMP2]], <vscale x 8 x bfloat> [[TMP3]], i64 8)
+// CPP-CHECK-NEXT:    ret <vscale x 16 x bfloat> [[TMP4]]
+//
+svbfloat16x2_t test_svread_za16_bf16_vg1x2(uint32_t base) __arm_streaming __arm_shared_za __arm_preserves_za {
+  return svread_za16_bf16_vg1x2(base);
+}
+
+// CHECK-LABEL: @test_svread_za16_f16_vg1x2(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call { <vscale x 8 x half>, <vscale x 8 x half> } @llvm.aarch64.sme.read.vg1x2.nxv8f16(i32 [[BASE:%.*]])
+// CHECK-NEXT:    [[TMP1:%.*]] = extractvalue { <vscale x 8 x half>, <vscale x 8 x half> } [[TMP0]], 0
+// CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 16 x half> @llvm.vector.insert.nxv16f16.nxv8f16(<vscale x 16 x half> poison, <vscale x 8 x half> [[TMP1]], i64 0)
+// CHECK-NEXT:    [[TMP3:%.*]] = extractvalue { <vscale x 8 x half>, <vscale x 8 x half> } [[TMP0]], 1
+// CHECK-NEXT:    [[TMP4:%.*]] = tail call <vscale x 16 x half> @llvm.vector.insert.nxv16f16.nxv8f16(<vscale x 16 x half> [[TMP2]], <vscale x 8 x half> [[TMP3]], i64 8)
+// CHECK-NEXT:    ret <vscale x 16 x half> [[TMP4]]
+//
+// CPP-CHECK-LABEL: @_Z26test_svread_za16_f16_vg1x2j(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call { <vscale x 8 x half>, <vscale x 8 x half> } @llvm.aarch64.sme.read.vg1x2.nxv8f16(i32 [[BASE:%.*]])
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = extractvalue { <vscale x 8 x half>, <vscale x 8 x half> } [[TMP0]], 0
+// CPP-CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 16 x half> @llvm.vector.insert.nxv16f16.nxv8f16(<vscale x 16 x half> poison, <vscale x 8 x half> [[TMP1]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP3:%.*]] = extractvalue { <vscale x 8 x half>, <vscale x 8 x half> } [[TMP0]], 1
+// CPP-CHECK-NEXT:    [[TMP4:%.*]] = tail call <vscale x 16 x half> @llvm.vector.insert.nxv16f16.nxv8f16(<vscale x 16 x half> [[TMP2]], <vscale x 8 x half> [[TMP3]], i64 8)
+// CPP-CHECK-NEXT:    ret <vscale x 16 x half> [[TMP4]]
+//
+svfloat16x2_t test_svread_za16_f16_vg1x2(uint32_t base) __arm_streaming __arm_shared_za __arm_preserves_za {
+  return svread_za16_f16_vg1x2(base);
+}
+
+// CHECK-LABEL: @test_svread_za32_s32_vg1x2(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call { <vscale x 4 x i32>, <vscale x 4 x i32> } @llvm.aarch64.sme.read.vg1x2.nxv4i32(i32 [[BASE:%.*]])
+// CHECK-NEXT:    [[TMP1:%.*]] = extractvalue { <vscale x 4 x i32>, <vscale x 4 x i32> } [[TMP0]], 0
+// CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 8 x i32> @llvm.vector.insert.nxv8i32.nxv4i32(<vscale x 8 x i32> poison, <vscale x 4 x i32> [[TMP1]], i64 0)
+// CHECK-NEXT:    [[TMP3:%.*]] = extractvalue { <vscale x 4 x i32>, <vscale x 4 x i32> } [[TMP0]], 1
+// CHECK-NEXT:    [[TMP4:%.*]] = tail call <vscale x 8 x i32> @llvm.vector.insert.nxv8i32.nxv4i32(<vscale x 8 x i32> [[TMP2]], <vscale x 4 x i32> [[TMP3]], i64 4)
+// CHECK-NEXT:    ret <vscale x 8 x i32> [[TMP4]]
+//
+// CPP-CHECK-LABEL: @_Z26test_svread_za32_s32_vg1x2j(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call { <vscale x 4 x i32>, <vscale x 4 x i32> } @llvm.aarch64.sme.read.vg1x2.nxv4i32(i32 [[BASE:%.*]])
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = extractvalue { <vscale x 4 x i32>, <vscale x 4 x i32> } [[TMP0]], 0
+// CPP-CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 8 x i32> @llvm.vector.insert.nxv8i32.nxv4i32(<vscale x 8 x i32> poison, <vscale x 4 x i32> [[TMP1]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP3:%.*]] = extractvalue { <vscale x 4 x i32>, <vscale x 4 x i32> } [[TMP0]], 1
+// CPP-CHECK-NEXT:    [[TMP4:%.*]] = tail call <vscale x 8 x i32> @llvm.vector.insert.nxv8i32.nxv4i32(<vscale x 8 x i32> [[TMP2]], <vscale x 4 x i32> [[TMP3]], i64 4)
+// CPP-CHECK-NEXT:    ret <vscale x 8 x i32> [[TMP4]]
+//
+svint32x2_t test_svread_za32_s32_vg1x2(uint32_t base) __arm_streaming __arm_shared_za __arm_preserves_za {
+  return svread_za32_s32_vg1x2(base);
+}
+
+// CHECK-LABEL: @test_svread_za32_u32_vg1x2(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call { <vscale x 4 x i32>, <vscale x 4 x i32> } @llvm.aarch64.sme.read.vg1x2.nxv4i32(i32 [[BASE:%.*]])
+// CHECK-NEXT:    [[TMP1:%.*]] = extractvalue { <vscale x 4 x i32>, <vscale x 4 x i32> } [[TMP0]], 0
+// CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 8 x i32> @llvm.vector.insert.nxv8i32.nxv4i32(<vscale x 8 x i32> poison, <vscale x 4 x i32> [[TMP1]], i64 0)
+// CHECK-NEXT:    [[TMP3:%.*]] = extractvalue { <vscale x 4 x i32>, <vscale x 4 x i32> } [[TMP0]], 1
+// CHECK-NEXT:    [[TMP4:%.*]] = tail call <vscale x 8 x i32> @llvm.vector.insert.nxv8i32.nxv4i32(<vscale x 8 x i32> [[TMP2]], <vscale x 4 x i32> [[TMP3]], i64 4)
+// CHECK-NEXT:    ret <vscale x 8 x i32> [[TMP4]]
+//
+// CPP-CHECK-LABEL: @_Z26test_svread_za32_u32_vg1x2j(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call { <vscale x 4 x i32>, <vscale x 4 x i32> } @llvm.aarch64.sme.read.vg1x2.nxv4i32(i32 [[BASE:%.*]])
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = extractvalue { <vscale x 4 x i32>, <vscale x 4 x i32> } [[TMP0]], 0
+// CPP-CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 8 x i32> @llvm.vector.insert.nxv8i32.nxv4i32(<vscale x 8 x i32> poison, <vscale x 4 x i32> [[TMP1]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP3:%.*]] = extractvalue { <vscale x 4 x i32>, <vscale x 4 x i32> } [[TMP0]], 1
+// CPP-CHECK-NEXT:    [[TMP4:%.*]] = tail call <vscale x 8 x i32> @llvm.vector.insert.nxv8i32.nxv4i32(<vscale x 8 x i32> [[TMP2]], <vscale x 4 x i32> [[TMP3]], i64 4)
+// CPP-CHECK-NEXT:    ret <vscale x 8 x i32> [[TMP4]]
+//
+svuint32x2_t test_svread_za32_u32_vg1x2(uint32_t base) __arm_streaming __arm_shared_za __arm_preserves_za {
+  return svread_za32_u32_vg1x2(base);
+}
+
+// CHECK-LABEL: @test_svread_za32_f32_vg1x2(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call { <vscale x 4 x float>, <vscale x 4 x float> } @llvm.aarch64.sme.read.vg1x2.nxv4f32(i32 [[BASE:%.*]])
+// CHECK-NEXT:    [[TMP1:%.*]] = extractvalue { <vscale x 4 x float>, <vscale x 4 x float> } [[TMP0]], 0
+// CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 8 x float> @llvm.vector.insert.nxv8f32.nxv4f32(<vscale x 8 x float> poison, <vscale x 4 x float> [[TMP1]], i64 0)
+// CHECK-NEXT:    [[TMP3:%.*]] = extractvalue { <vscale x 4 x float>, <vscale x 4 x float> } [[TMP0]], 1
+// CHECK-NEXT:    [[TMP4:%.*]] = tail call <vscale x 8 x float> @llvm.vector.insert.nxv8f32.nxv4f32(<vscale x 8 x float> [[TMP2]], <vscale x 4 x float> [[TMP3]], i64 4)
+// CHECK-NEXT:    ret <vscale x 8 x float> [[TMP4]]
+//
+// CPP-CHECK-LABEL: @_Z26test_svread_za32_f32_vg1x2j(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call { <vscale x 4 x float>, <vscale x 4 x float> } @llvm.aarch64.sme.read.vg1x2.nxv4f32(i32 [[BASE:%.*]])
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = extractvalue { <vscale x 4 x float>, <vscale x 4 x float> } [[TMP0]], 0
+// CPP-CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 8 x float> @llvm.vector.insert.nxv8f32.nxv4f32(<vscale x 8 x float> poison, <vscale x 4 x float> [[TMP1]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP3:%.*]] = extractvalue { <vscale x 4 x float>, <vscale x 4 x float> } [[TMP0]], 1
+// CPP-CHECK-NEXT:    [[TMP4:%.*]] = tail call <vscale x 8 x float> @llvm.vector.insert.nxv8f32.nxv4f32(<vscale x 8 x float> [[TMP2]], <vscale x 4 x float> [[TMP3]], i64 4)
+// CPP-CHECK-NEXT:    ret <vscale x 8 x float> [[TMP4]]
+//
+svfloat32x2_t test_svread_za32_f32_vg1x2(uint32_t base) __arm_streaming __arm_shared_za __arm_preserves_za {
+  return svread_za32_f32_vg1x2(base);
+}
+
 // CHECK-LABEL: @test_svread_za64_u64_vg1x2(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = tail call { <vscale x 2 x i64>, <vscale x 2 x i64> } @llvm.aarch64.sme.read.vg1x2.nxv2i64(i32 [[BASE:%.*]])
@@ -1330,6 +1528,276 @@ svint64x2_t test_svread_za64_s64_vg1x2(uint32_t base) __arm_streaming __arm_shar
   return svread_za64_s64_vg1x2(base);
 }
 
+// CHECK-LABEL: @test_svread_za8_s8_vg1x4(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call { <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8> } @llvm.aarch64.sme.read.vg1x4.nxv16i8(i32 [[BASE:%.*]])
+// CHECK-NEXT:    [[TMP1:%.*]] = extractvalue { <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8> } [[TMP0]], 0
+// CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 64 x i8> @llvm.vector.insert.nxv64i8.nxv16i8(<vscale x 64 x i8> poison, <vscale x 16 x i8> [[TMP1]], i64 0)
+// CHECK-NEXT:    [[TMP3:%.*]] = extractvalue { <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8> } [[TMP0]], 1
+// CHECK-NEXT:    [[TMP4:%.*]] = tail call <vscale x 64 x i8> @llvm.vector.insert.nxv64i8.nxv16i8(<vscale x 64 x i8> [[TMP2]], <vscale x 16 x i8> [[TMP3]], i64 16)
+// CHECK-NEXT:    [[TMP5:%.*]] = extractvalue { <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8> } [[TMP0]], 2
+// CHECK-NEXT:    [[TMP6:%.*]] = tail call <vscale x 64 x i8> @llvm.vector.insert.nxv64i8.nxv16i8(<vscale x 64 x i8> [[TMP4]], <vscale x 16 x i8> [[TMP5]], i64 32)
+// CHECK-NEXT:    [[TMP7:%.*]] = extractvalue { <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8> } [[TMP0]], 3
+// CHECK-NEXT:    [[TMP8:%.*]] = tail call <vscale x 64 x i8> @llvm.vector.insert.nxv64i8.nxv16i8(<vscale x 64 x i8> [[TMP6]], <vscale x 16 x i8> [[TMP7]], i64 48)
+// CHECK-NEXT:    ret <vscale x 64 x i8> [[TMP8]]
+//
+// CPP-CHECK-LABEL: @_Z24test_svread_za8_s8_vg1x4j(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call { <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8> } @llvm.aarch64.sme.read.vg1x4.nxv16i8(i32 [[BASE:%.*]])
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = extractvalue { <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8> } [[TMP0]], 0
+// CPP-CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 64 x i8> @llvm.vector.insert.nxv64i8.nxv16i8(<vscale x 64 x i8> poison, <vscale x 16 x i8> [[TMP1]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP3:%.*]] = extractvalue { <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8> } [[TMP0]], 1
+// CPP-CHECK-NEXT:    [[TMP4:%.*]] = tail call <vscale x 64 x i8> @llvm.vector.insert.nxv64i8.nxv16i8(<vscale x 64 x i8> [[TMP2]], <vscale x 16 x i8> [[TMP3]], i64 16)
+// CPP-CHECK-NEXT:    [[TMP5:%.*]] = extractvalue { <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8> } [[TMP0]], 2
+// CPP-CHECK-NEXT:    [[TMP6:%.*]] = tail call <vscale x 64 x i8> @llvm.vector.insert.nxv64i8.nxv16i8(<vscale x 64 x i8> [[TMP4]], <vscale x 16 x i8> [[TMP5]], i64 32)
+// CPP-CHECK-NEXT:    [[TMP7:%.*]] = extractvalue { <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8> } [[TMP0]], 3
+// CPP-CHECK-NEXT:    [[TMP8:%.*]] = tail call <vscale x 64 x i8> @llvm.vector.insert.nxv64i8.nxv16i8(<vscale x 64 x i8> [[TMP6]], <vscale x 16 x i8> [[TMP7]], i64 48)
+// CPP-CHECK-NEXT:    ret <vscale x 64 x i8> [[TMP8]]
+//
+svint8x4_t test_svread_za8_s8_vg1x4(uint32_t base) __arm_streaming __arm_shared_za __arm_preserves_za {
+  return svread_za8_s8_vg1x4(base);
+}
+
+// CHECK-LABEL: @test_svread_za8_u8_vg1x4(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call { <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8> } @llvm.aarch64.sme.read.vg1x4.nxv16i8(i32 [[BASE:%.*]])
+// CHECK-NEXT:    [[TMP1:%.*]] = extractvalue { <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8> } [[TMP0]], 0
+// CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 64 x i8> @llvm.vector.insert.nxv64i8.nxv16i8(<vscale x 64 x i8> poison, <vscale x 16 x i8> [[TMP1]], i64 0)
+// CHECK-NEXT:    [[TMP3:%.*]] = extractvalue { <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8> } [[TMP0]], 1
+// CHECK-NEXT:    [[TMP4:%.*]] = tail call <vscale x 64 x i8> @llvm.vector.insert.nxv64i8.nxv16i8(<vscale x 64 x i8> [[TMP2]], <vscale x 16 x i8> [[TMP3]], i64 16)
+// CHECK-NEXT:    [[TMP5:%.*]] = extractvalue { <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8> } [[TMP0]], 2
+// CHECK-NEXT:    [[TMP6:%.*]] = tail call <vscale x 64 x i8> @llvm.vector.insert.nxv64i8.nxv16i8(<vscale x 64 x i8> [[TMP4]], <vscale x 16 x i8> [[TMP5]], i64 32)
+// CHECK-NEXT:    [[TMP7:%.*]] = extractvalue { <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8> } [[TMP0]], 3
+// CHECK-NEXT:    [[TMP8:%.*]] = tail call <vscale x 64 x i8> @llvm.vector.insert.nxv64i8.nxv16i8(<vscale x 64 x i8> [[TMP6]], <vscale x 16 x i8> [[TMP7]], i64 48)
+// CHECK-NEXT:    ret <vscale x 64 x i8> [[TMP8]]
+//
+// CPP-CHECK-LABEL: @_Z24test_svread_za8_u8_vg1x4j(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call { <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8> } @llvm.aarch64.sme.read.vg1x4.nxv16i8(i32 [[BASE:%.*]])
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = extractvalue { <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8> } [[TMP0]], 0
+// CPP-CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 64 x i8> @llvm.vector.insert.nxv64i8.nxv16i8(<vscale x 64 x i8> poison, <vscale x 16 x i8> [[TMP1]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP3:%.*]] = extractvalue { <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8> } [[TMP0]], 1
+// CPP-CHECK-NEXT:    [[TMP4:%.*]] = tail call <vscale x 64 x i8> @llvm.vector.insert.nxv64i8.nxv16i8(<vscale x 64 x i8> [[TMP2]], <vscale x 16 x i8> [[TMP3]], i64 16)
+// CPP-CHECK-NEXT:    [[TMP5:%.*]] = extractvalue { <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8> } [[TMP0]], 2
+// CPP-CHECK-NEXT:    [[TMP6:%.*]] = tail call <vscale x 64 x i8> @llvm.vector.insert.nxv64i8.nxv16i8(<vscale x 64 x i8> [[TMP4]], <vscale x 16 x i8> [[TMP5]], i64 32)
+// CPP-CHECK-NEXT:    [[TMP7:%.*]] = extractvalue { <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8> } [[TMP0]], 3
+// CPP-CHECK-NEXT:    [[TMP8:%.*]] = tail call <vscale x 64 x i8> @llvm.vector.insert.nxv64i8.nxv16i8(<vscale x 64 x i8> [[TMP6]], <vscale x 16 x i8> [[TMP7]], i64 48)
+// CPP-CHECK-NEXT:    ret <vscale x 64 x i8> [[TMP8]]
+//
+svuint8x4_t test_svread_za8_u8_vg1x4(uint32_t base) __arm_streaming __arm_shared_za __arm_preserves_za {
+  return svread_za8_u8_vg1x4(base);
+}
+
+// CHECK-LABEL: @test_svread_za16_s16_vg1x4(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call { <vscale x 8 x i16>, <vscale x 8 x i16>, <vscale x 8 x i16>, <vscale x 8 x i16> } @llvm.aarch64.sme.read.vg1x4.nxv8i16(i32 [[BASE:%.*]])
+// CHECK-NEXT:    [[TMP1:%.*]] = extractvalue { <vscale x 8 x i16>, <vscale x 8 x i16>, <vscale x 8 x i16>, <vscale x 8 x i16> } [[TMP0]], 0
+// CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 32 x i16> @llvm.vector.insert.nxv32i16.nxv8i16(<vscale x 32 x i16> poison, <vscale x 8 x i16> [[TMP1]], i64 0)
+// CHECK-NEXT:    [[TMP3:%.*]] = extractvalue { <vscale x 8 x i16>, <vscale x 8 x i16>, <vscale x 8 x i16>, <vscale x 8 x i16> } [[TMP0]], 1
+// CHECK-NEXT:    [[TMP4:%.*]] = tail call <vscale x 32 x i16> @llvm.vector.insert.nxv32i16.nxv8i16(<vscale x 32 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP3]], i64 8)
+// CHECK-NEXT:    [[TMP5:%.*]] = extractvalue { <vscale x 8 x i16>, <vscale x 8 x i16>, <vscale x 8 x i16>, <vscale x 8 x i16> } [[TMP0]], 2
+// CHECK-NEXT:    [[TMP6:%.*]] = tail call <vscale x 32 x i16> @llvm.vector.insert.nxv32i16.nxv8i16(<vscale x 32 x i16> [[TMP4]], <vscale x 8 x i16> [[TMP5]], i64 16)
+// CHECK-NEXT:    [[TMP7:%.*]] = extractvalue { <vscale x 8 x i16>, <vscale x 8 x i16>, <vscale x 8 x i16>, <vscale x 8 x i16> } [[TMP0]], 3
+// CHECK-NEXT:    [[TMP8:%.*]] = tail call <vscale x 32 x i16> @llvm.vector.insert.nxv32i16.nxv8i16(<vscale x 32 x i16> [[TMP6]], <vscale x 8 x i16> [[TMP7]], i64 24)
+// CHECK-NEXT:    ret <vscale x 32 x i16> [[TMP8]]
+//
+// CPP-CHECK-LABEL: @_Z26test_svread_za16_s16_vg1x4j(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call { <vscale x 8 x i16>, <vscale x 8 x i16>, <vscale x 8 x i16>, <vscale x 8 x i16> } @llvm.aarch64.sme.read.vg1x4.nxv8i16(i32 [[BASE:%.*]])
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = extractvalue { <vscale x 8 x i16>, <vscale x 8 x i16>, <vscale x 8 x i16>, <vscale x 8 x i16> } [[TMP0]], 0
+// CPP-CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 32 x i16> @llvm.vector.insert.nxv32i16.nxv8i16(<vscale x 32 x i16> poison, <vscale x 8 x i16> [[TMP1]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP3:%.*]] = extractvalue { <vscale x 8 x i16>, <vscale x 8 x i16>, <vscale x 8 x i16>, <vscale x 8 x i16> } [[TMP0]], 1
+// CPP-CHECK-NEXT:    [[TMP4:%.*]] = tail call <vscale x 32 x i16> @llvm.vector.insert.nxv32i16.nxv8i16(<vscale x 32 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP3]], i64 8)
+// CPP-CHECK-NEXT:    [[TMP5:%.*]] = extractvalue { <vscale x 8 x i16>, <vscale x 8 x i16>, <vscale x 8 x i16>, <vscale x 8 x i16> } [[TMP0]], 2
+// CPP-CHECK-NEXT:    [[TMP6:%.*]] = tail call <vscale x 32 x i16> @llvm.vector.insert.nxv32i16.nxv8i16(<vscale x 32 x i16> [[TMP4]], <vscale x 8 x i16> [[TMP5]], i64 16)
+// CPP-CHECK-NEXT:    [[TMP7:%.*]] = extractvalue { <vscale x 8 x i16>, <vscale x 8 x i16>, <vscale x 8 x i16>, <vscale x 8 x i16> } [[TMP0]], 3
+// CPP-CHECK-NEXT:    [[TMP8:%.*]] = tail call <vscale x 32 x i16> @llvm.vector.insert.nxv32i16.nxv8i16(<vscale x 32 x i16> [[TMP6]], <vscale x 8 x i16> [[TMP7]], i64 24)
+// CPP-CHECK-NEXT:    ret <vscale x 32 x i16> [[TMP8]]
+//
+svint16x4_t test_svread_za16_s16_vg1x4(uint32_t base) __arm_streaming __arm_shared_za __arm_preserves_za {
+  return svread_za16_s16_vg1x4(base);
+}
+
+// CHECK-LABEL: @test_svread_za16_u16_vg1x4(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call { <vscale x 8 x i16>, <vscale x 8 x i16>, <vscale x 8 x i16>, <vscale x 8 x i16> } @llvm.aarch64.sme.read.vg1x4.nxv8i16(i32 [[BASE:%.*]])
+// CHECK-NEXT:    [[TMP1:%.*]] = extractvalue { <vscale x 8 x i16>, <vscale x 8 x i16>, <vscale x 8 x i16>, <vscale x 8 x i16> } [[TMP0]], 0
+// CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 32 x i16> @llvm.vector.insert.nxv32i16.nxv8i16(<vscale x 32 x i16> poison, <vscale x 8 x i16> [[TMP1]], i64 0)
+// CHECK-NEXT:    [[TMP3:%.*]] = extractvalue { <vscale x 8 x i16>, <vscale x 8 x i16>, <vscale x 8 x i16>, <vscale x 8 x i16> } [[TMP0]], 1
+// CHECK-NEXT:    [[TMP4:%.*]] = tail call <vscale x 32 x i16> @llvm.vector.insert.nxv32i16.nxv8i16(<vscale x 32 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP3]], i64 8)
+// CHECK-NEXT:    [[TMP5:%.*]] = extractvalue { <vscale x 8 x i16>, <vscale x 8 x i16>, <vscale x 8 x i16>, <vscale x 8 x i16> } [[TMP0]], 2
+// CHECK-NEXT:    [[TMP6:%.*]] = tail call <vscale x 32 x i16> @llvm.vector.insert.nxv32i16.nxv8i16(<vscale x 32 x i16> [[TMP4]], <vscale x 8 x i16> [[TMP5]], i64 16)
+// CHECK-NEXT:    [[TMP7:%.*]] = extractvalue { <vscale x 8 x i16>, <vscale x 8 x i16>, <vscale x 8 x i16>, <vscale x 8 x i16> } [[TMP0]], 3
+// CHECK-NEXT:    [[TMP8:%.*]] = tail call <vscale x 32 x i16> @llvm.vector.insert.nxv32i16.nxv8i16(<vscale x 32 x i16> [[TMP6]], <vscale x 8 x i16> [[TMP7]], i64 24)
+// CHECK-NEXT:    ret <vscale x 32 x i16> [[TMP8]]
+//
+// CPP-CHECK-LABEL: @_Z26test_svread_za16_u16_vg1x4j(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call { <vscale x 8 x i16>, <vscale x 8 x i16>, <vscale x 8 x i16>, <vscale x 8 x i16> } @llvm.aarch64.sme.read.vg1x4.nxv8i16(i32 [[BASE:%.*]])
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = extractvalue { <vscale x 8 x i16>, <vscale x 8 x i16>, <vscale x 8 x i16>, <vscale x 8 x i16> } [[TMP0]], 0
+// CPP-CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 32 x i16> @llvm.vector.insert.nxv32i16.nxv8i16(<vscale x 32 x i16> poison, <vscale x 8 x i16> [[TMP1]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP3:%.*]] = extractvalue { <vscale x 8 x i16>, <vscale x 8 x i16>, <vscale x 8 x i16>, <vscale x 8 x i16> } [[TMP0]], 1
+// CPP-CHECK-NEXT:    [[TMP4:%.*]] = tail call <vscale x 32 x i16> @llvm.vector.insert.nxv32i16.nxv8i16(<vscale x 32 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP3]], i64 8)
+// CPP-CHECK-NEXT:    [[TMP5:%.*]] = extractvalue { <vscale x 8 x i16>, <vscale x 8 x i16>, <vscale x 8 x i16>, <vscale x 8 x i16> } [[TMP0]], 2
+// CPP-CHECK-NEXT:    [[TMP6:%.*]] = tail call <vscale x 32 x i16> @llvm.vector.insert.nxv32i16.nxv8i16(<vscale x 32 x i16> [[TMP4]], <vscale x 8 x i16> [[TMP5]], i64 16)
+// CPP-CHECK-NEXT:    [[TMP7:%.*]] = extractvalue { <vscale x 8 x i16>, <vscale x 8 x i16>, <vscale x 8 x i16>, <vscale x 8 x i16> } [[TMP0]], 3
+// CPP-CHECK-NEXT:    [[TMP8:%.*]] = tail call <vscale x 32 x i16> @llvm.vector.insert.nxv32i16.nxv8i16(<vscale x 32 x i16> [[TMP6]], <vscale x 8 x i16> [[TMP7]], i64 24)
+// CPP-CHECK-NEXT:    ret <vscale x 32 x i16> [[TMP8]]
+//
+svuint16x4_t test_svread_za16_u16_vg1x4(uint32_t base) __arm_streaming __arm_shared_za __arm_preserves_za {
+  return svread_za16_u16_vg1x4(base);
+}
+
+// CHECK-LABEL: @test_svread_za16_bf16_vg1x4(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call { <vscale x 8 x bfloat>, <vscale x 8 x bfloat>, <vscale x 8 x bfloat>, <vscale x 8 x bfloat> } @llvm.aarch64.sme.read.vg1x4.nxv8bf16(i32 [[BASE:%.*]])
+// CHECK-NEXT:    [[TMP1:%.*]] = extractvalue { <vscale x 8 x bfloat>, <vscale x 8 x bfloat>, <vscale x 8 x bfloat>, <vscale x 8 x bfloat> } [[TMP0]], 0
+// CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 32 x bfloat> @llvm.vector.insert.nxv32bf16.nxv8bf16(<vscale x 32 x bfloat> poison, <vscale x 8 x bfloat> [[TMP1]], i64 0)
+// CHECK-NEXT:    [[TMP3:%.*]] = extractvalue { <vscale x 8 x bfloat>, <vscale x 8 x bfloat>, <vscale x 8 x bfloat>, <vscale x 8 x bfloat> } [[TMP0]], 1
+// CHECK-NEXT:    [[TMP4:%.*]] = tail call <vscale x 32 x bfloat> @llvm.vector.insert.nxv32bf16.nxv8bf16(<vscale x 32 x bfloat> [[TMP2]], <vscale x 8 x bfloat> [[TMP3]], i64 8)
+// CHECK-NEXT:    [[TMP5:%.*]] = extractvalue { <vscale x 8 x bfloat>, <vscale x 8 x bfloat>, <vscale x 8 x bfloat>, <vscale x 8 x bfloat> } [[TMP0]], 2
+// CHECK-NEXT:    [[TMP6:%.*]] = tail call <vscale x 32 x bfloat> @llvm.vector.insert.nxv32bf16.nxv8bf16(<vscale x 32 x bfloat> [[TMP4]], <vscale x 8 x bfloat> [[TMP5]], i64 16)
+// CHECK-NEXT:    [[TMP7:%.*]] = extractvalue { <vscale x 8 x bfloat>, <vscale x 8 x bfloat>, <vscale x 8 x bfloat>, <vscale x 8 x bfloat> } [[TMP0]], 3
+// CHECK-NEXT:    [[TMP8:%.*]] = tail call <vscale x 32 x bfloat> @llvm.vector.insert.nxv32bf16.nxv8bf16(<vscale x 32 x bfloat> [[TMP6]], <vscale x 8 x bfloat> [[TMP7]], i64 24)
+// CHECK-NEXT:    ret <vscale x 32 x bfloat> [[TMP8]]
+//
+// CPP-CHECK-LABEL: @_Z27test_svread_za16_bf16_vg1x4j(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call { <vscale x 8 x bfloat>, <vscale x 8 x bfloat>, <vscale x 8 x bfloat>, <vscale x 8 x bfloat> } @llvm.aarch64.sme.read.vg1x4.nxv8bf16(i32 [[BASE:%.*]])
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = extractvalue { <vscale x 8 x bfloat>, <vscale x 8 x bfloat>, <vscale x 8 x bfloat>, <vscale x 8 x bfloat> } [[TMP0]], 0
+// CPP-CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 32 x bfloat> @llvm.vector.insert.nxv32bf16.nxv8bf16(<vscale x 32 x bfloat> poison, <vscale x 8 x bfloat> [[TMP1]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP3:%.*]] = extractvalue { <vscale x 8 x bfloat>, <vscale x 8 x bfloat>, <vscale x 8 x bfloat>, <vscale x 8 x bfloat> } [[TMP0]], 1
+// CPP-CHECK-NEXT:    [[TMP4:%.*]] = tail call <vscale x 32 x bfloat> @llvm.vector.insert.nxv32bf16.nxv8bf16(<vscale x 32 x bfloat> [[TMP2]], <vscale x 8 x bfloat> [[TMP3]], i64 8)
+// CPP-CHECK-NEXT:    [[TMP5:%.*]] = extractvalue { <vscale x 8 x bfloat>, <vscale x 8 x bfloat>, <vscale x 8 x bfloat>, <vscale x 8 x bfloat> } [[TMP0]], 2
+// CPP-CHECK-NEXT:    [[TMP6:%.*]] = tail call <vscale x 32 x bfloat> @llvm.vector.insert.nxv32bf16.nxv8bf16(<vscale x 32 x bfloat> [[TMP4]], <vscale x 8 x bfloat> [[TMP5]], i64 16)
+// CPP-CHECK-NEXT:    [[TMP7:%.*]] = extractvalue { <vscale x 8 x bfloat>, <vscale x 8 x bfloat>, <vscale x 8 x bfloat>, <vscale x 8 x bfloat> } [[TMP0]], 3
+// CPP-CHECK-NEXT:    [[TMP8:%.*]] = tail call <vscale x 32 x bfloat> @llvm.vector.insert.nxv32bf16.nxv8bf16(<vscale x 32 x bfloat> [[TMP6]], <vscale x 8 x bfloat> [[TMP7]], i64 24)
+// CPP-CHECK-NEXT:    ret <vscale x 32 x bfloat> [[TMP8]]
+//
+svbfloat16x4_t test_svread_za16_bf16_vg1x4(uint32_t base) __arm_streaming __arm_shared_za __arm_preserves_za {
+  return svread_za16_bf16_vg1x4(base);
+}
+
+// CHECK-LABEL: @test_svread_za16_f16_vg1x4(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call { <vscale x 8 x half>, <vscale x 8 x half>, <vscale x 8 x half>, <vscale x 8 x half> } @llvm.aarch64.sme.read.vg1x4.nxv8f16(i32 [[BASE:%.*]])
+// CHECK-NEXT:    [[TMP1:%.*]] = extractvalue { <vscale x 8 x half>, <vscale x 8 x half>, <vscale x 8 x half>, <vscale x 8 x half> } [[TMP0]], 0
+// CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 32 x half> @llvm.vector.insert.nxv32f16.nxv8f16(<vscale x 32 x half> poison, <vscale x 8 x half> [[TMP1]], i64 0)
+// CHECK-NEXT:    [[TMP3:%.*]] = extractvalue { <vscale x 8 x half>, <vscale x 8 x half>, <vscale x 8 x half>, <vscale x 8 x half> } [[TMP0]], 1
+// CHECK-NEXT:    [[TMP4:%.*]] = tail call <vscale x 32 x half> @llvm.vector.insert.nxv32f16.nxv8f16(<vscale x 32 x half> [[TMP2]], <vscale x 8 x half> [[TMP3]], i64 8)
+// CHECK-NEXT:    [[TMP5:%.*]] = extractvalue { <vscale x 8 x half>, <vscale x 8 x half>, <vscale x 8 x half>, <vscale x 8 x half> } [[TMP0]], 2
+// CHECK-NEXT:    [[TMP6:%.*]] = tail call <vscale x 32 x half> @llvm.vector.insert.nxv32f16.nxv8f16(<vscale x 32 x half> [[TMP4]], <vscale x 8 x half> [[TMP5]], i64 16)
+// CHECK-NEXT:    [[TMP7:%.*]] = extractvalue { <vscale x 8 x half>, <vscale x 8 x half>, <vscale x 8 x half>, <vscale x 8 x half> } [[TMP0]], 3
+// CHECK-NEXT:    [[TMP8:%.*]] = tail call <vscale x 32 x half> @llvm.vector.insert.nxv32f16.nxv8f16(<vscale x 32 x half> [[TMP6]], <vscale x 8 x half> [[TMP7]], i64 24)
+// CHECK-NEXT:    ret <vscale x 32 x half> [[TMP8]]
+//
+// CPP-CHECK-LABEL: @_Z26test_svread_za16_f16_vg1x4j(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call { <vscale x 8 x half>, <vscale x 8 x half>, <vscale x 8 x half>, <vscale x 8 x half> } @llvm.aarch64.sme.read.vg1x4.nxv8f16(i32 [[BASE:%.*]])
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = extractvalue { <vscale x 8 x half>, <vscale x 8 x half>, <vscale x 8 x half>, <vscale x 8 x half> } [[TMP0]], 0
+// CPP-CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 32 x half> @llvm.vector.insert.nxv32f16.nxv8f16(<vscale x 32 x half> poison, <vscale x 8 x half> [[TMP1]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP3:%.*]] = extractvalue { <vscale x 8 x half>, <vscale x 8 x half>, <vscale x 8 x half>, <vscale x 8 x half> } [[TMP0]], 1
+// CPP-CHECK-NEXT:    [[TMP4:%.*]] = tail call <vscale x 32 x half> @llvm.vector.insert.nxv32f16.nxv8f16(<vscale x 32 x half> [[TMP2]], <vscale x 8 x half> [[TMP3]], i64 8)
+// CPP-CHECK-NEXT:    [[TMP5:%.*]] = extractvalue { <vscale x 8 x half>, <vscale x 8 x half>, <vscale x 8 x half>, <vscale x 8 x half> } [[TMP0]], 2
+// CPP-CHECK-NEXT:    [[TMP6:%.*]] = tail call <vscale x 32 x half> @llvm.vector.insert.nxv32f16.nxv8f16(<vscale x 32 x half> [[TMP4]], <vscale x 8 x half> [[TMP5]], i64 16)
+// CPP-CHECK-NEXT:    [[TMP7:%.*]] = extractvalue { <vscale x 8 x half>, <vscale x 8 x half>, <vscale x 8 x half>, <vscale x 8 x half> } [[TMP0]], 3
+// CPP-CHECK-NEXT:    [[TMP8:%.*]] = tail call <vscale x 32 x half> @llvm.vector.insert.nxv32f16.nxv8f16(<vscale x 32 x half> [[TMP6]], <vscale x 8 x half> [[TMP7]], i64 24)
+// CPP-CHECK-NEXT:    ret <vscale x 32 x half> [[TMP8]]
+//
+svfloat16x4_t test_svread_za16_f16_vg1x4(uint32_t base) __arm_streaming __arm_shared_za __arm_preserves_za {
+  return svread_za16_f16_vg1x4(base);
+}
+
+// CHECK-LABEL: @test_svread_za32_s32_vg1x4(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call { <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32> } @llvm.aarch64.sme.read.vg1x4.nxv4i32(i32 [[BASE:%.*]])
+// CHECK-NEXT:    [[TMP1:%.*]] = extractvalue { <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32> } [[TMP0]], 0
+// CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 16 x i32> @llvm.vector.insert.nxv16i32.nxv4i32(<vscale x 16 x i32> poison, <vscale x 4 x i32> [[TMP1]], i64 0)
+// CHECK-NEXT:    [[TMP3:%.*]] = extractvalue { <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32> } [[TMP0]], 1
+// CHECK-NEXT:    [[TMP4:%.*]] = tail call <vscale x 16 x i32> @llvm.vector.insert.nxv16i32.nxv4i32(<vscale x 16 x i32> [[TMP2]], <vscale x 4 x i32> [[TMP3]], i64 4)
+// CHECK-NEXT:    [[TMP5:%.*]] = extractvalue { <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32> } [[TMP0]], 2
+// CHECK-NEXT:    [[TMP6:%.*]] = tail call <vscale x 16 x i32> @llvm.vector.insert.nxv16i32.nxv4i32(<vscale x 16 x i32> [[TMP4]], <vscale x 4 x i32> [[TMP5]], i64 8)
+// CHECK-NEXT:    [[TMP7:%.*]] = extractvalue { <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32> } [[TMP0]], 3
+// CHECK-NEXT:    [[TMP8:%.*]] = tail call <vscale x 16 x i32> @llvm.vector.insert.nxv16i32.nxv4i32(<vscale x 16 x i32> [[TMP6]], <vscale x 4 x i32> [[TMP7]], i64 12)
+// CHECK-NEXT:    ret <vscale x 16 x i32> [[TMP8]]
+//
+// CPP-CHECK-LABEL: @_Z26test_svread_za32_s32_vg1x4j(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call { <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32> } @llvm.aarch64.sme.read.vg1x4.nxv4i32(i32 [[BASE:%.*]])
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = extractvalue { <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32> } [[TMP0]], 0
+// CPP-CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 16 x i32> @llvm.vector.insert.nxv16i32.nxv4i32(<vscale x 16 x i32> poison, <vscale x 4 x i32> [[TMP1]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP3:%.*]] = extractvalue { <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32> } [[TMP0]], 1
+// CPP-CHECK-NEXT:    [[TMP4:%.*]] = tail call <vscale x 16 x i32> @llvm.vector.insert.nxv16i32.nxv4i32(<vscale x 16 x i32> [[TMP2]], <vscale x 4 x i32> [[TMP3]], i64 4)
+// CPP-CHECK-NEXT:    [[TMP5:%.*]] = extractvalue { <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32> } [[TMP0]], 2
+// CPP-CHECK-NEXT:    [[TMP6:%.*]] = tail call <vscale x 16 x i32> @llvm.vector.insert.nxv16i32.nxv4i32(<vscale x 16 x i32> [[TMP4]], <vscale x 4 x i32> [[TMP5]], i64 8)
+// CPP-CHECK-NEXT:    [[TMP7:%.*]] = extractvalue { <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32> } [[TMP0]], 3
+// CPP-CHECK-NEXT:    [[TMP8:%.*]] = tail call <vscale x 16 x i32> @llvm.vector.insert.nxv16i32.nxv4i32(<vscale x 16 x i32> [[TMP6]], <vscale x 4 x i32> [[TMP7]], i64 12)
+// CPP-CHECK-NEXT:    ret <vscale x 16 x i32> [[TMP8]]
+//
+svint32x4_t test_svread_za32_s32_vg1x4(uint32_t base) __arm_streaming __arm_shared_za __arm_preserves_za {
+  return svread_za32_s32_vg1x4(base);
+}
+
+// CHECK-LABEL: @test_svread_za32_u32_vg1x4(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call { <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32> } @llvm.aarch64.sme.read.vg1x4.nxv4i32(i32 [[BASE:%.*]])
+// CHECK-NEXT:    [[TMP1:%.*]] = extractvalue { <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32> } [[TMP0]], 0
+// CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 16 x i32> @llvm.vector.insert.nxv16i32.nxv4i32(<vscale x 16 x i32> poison, <vscale x 4 x i32> [[TMP1]], i64 0)
+// CHECK-NEXT:    [[TMP3:%.*]] = extractvalue { <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32> } [[TMP0]], 1
+// CHECK-NEXT:    [[TMP4:%.*]] = tail call <vscale x 16 x i32> @llvm.vector.insert.nxv16i32.nxv4i32(<vscale x 16 x i32> [[TMP2]], <vscale x 4 x i32> [[TMP3]], i64 4)
+// CHECK-NEXT:    [[TMP5:%.*]] = extractvalue { <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32> } [[TMP0]], 2
+// CHECK-NEXT:    [[TMP6:%.*]] = tail call <vscale x 16 x i32> @llvm.vector.insert.nxv16i32.nxv4i32(<vscale x 16 x i32> [[TMP4]], <vscale x 4 x i32> [[TMP5]], i64 8)
+// CHECK-NEXT:    [[TMP7:%.*]] = extractvalue { <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32> } [[TMP0]], 3
+// CHECK-NEXT:    [[TMP8:%.*]] = tail call <vscale x 16 x i32> @llvm.vector.insert.nxv16i32.nxv4i32(<vscale x 16 x i32> [[TMP6]], <vscale x 4 x i32> [[TMP7]], i64 12)
+// CHECK-NEXT:    ret <vscale x 16 x i32> [[TMP8]]
+//
+// CPP-CHECK-LABEL: @_Z26test_svread_za32_u32_vg1x4j(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call { <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32> } @llvm.aarch64.sme.read.vg1x4.nxv4i32(i32 [[BASE:%.*]])
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = extractvalue { <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32> } [[TMP0]], 0
+// CPP-CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 16 x i32> @llvm.vector.insert.nxv16i32.nxv4i32(<vscale x 16 x i32> poison, <vscale x 4 x i32> [[TMP1]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP3:%.*]] = extractvalue { <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32> } [[TMP0]], 1
+// CPP-CHECK-NEXT:    [[TMP4:%.*]] = tail call <vscale x 16 x i32> @llvm.vector.insert.nxv16i32.nxv4i32(<vscale x 16 x i32> [[TMP2]], <vscale x 4 x i32> [[TMP3]], i64 4)
+// CPP-CHECK-NEXT:    [[TMP5:%.*]] = extractvalue { <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32> } [[TMP0]], 2
+// CPP-CHECK-NEXT:    [[TMP6:%.*]] = tail call <vscale x 16 x i32> @llvm.vector.insert.nxv16i32.nxv4i32(<vscale x 16 x i32> [[TMP4]], <vscale x 4 x i32> [[TMP5]], i64 8)
+// CPP-CHECK-NEXT:    [[TMP7:%.*]] = extractvalue { <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32> } [[TMP0]], 3
+// CPP-CHECK-NEXT:    [[TMP8:%.*]] = tail call <vscale x 16 x i32> @llvm.vector.insert.nxv16i32.nxv4i32(<vscale x 16 x i32> [[TMP6]], <vscale x 4 x i32> [[TMP7]], i64 12)
+// CPP-CHECK-NEXT:    ret <vscale x 16 x i32> [[TMP8]]
+//
+svuint32x4_t test_svread_za32_u32_vg1x4(uint32_t base) __arm_streaming __arm_shared_za __arm_preserves_za {
+  return svread_za32_u32_vg1x4(base);
+}
+
+// CHECK-LABEL: @test_svread_za32_f32_vg1x4(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call { <vscale x 4 x float>, <vscale x 4 x float>, <vscale x 4 x float>, <vscale x 4 x float> } @llvm.aarch64.sme.read.vg1x4.nxv4f32(i32 [[BASE:%.*]])
+// CHECK-NEXT:    [[TMP1:%.*]] = extractvalue { <vscale x 4 x float>, <vscale x 4 x float>, <vscale x 4 x float>, <vscale x 4 x float> } [[TMP0]], 0
+// CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 16 x float> @llvm.vector.insert.nxv16f32.nxv4f32(<vscale x 16 x float> poison, <vscale x 4 x float> [[TMP1]], i64 0)
+// CHECK-NEXT:    [[TMP3:%.*]] = extractvalue { <vscale x 4 x float>, <vscale x 4 x float>, <vscale x 4 x float>, <vscale x 4 x float> } [[TMP0]], 1
+// CHECK-NEXT:    [[TMP4:%.*]] = tail call <vscale x 16 x float> @llvm.vector.insert.nxv16f32.nxv4f32(<vscale x 16 x float> [[TMP2]], <vscale x 4 x float> [[TMP3]], i64 4)
+// CHECK-NEXT:    [[TMP5:%.*]] = extractvalue { <vscale x 4 x float>, <vscale x 4 x float>, <vscale x 4 x float>, <vscale x 4 x float> } [[TMP0]], 2
+// CHECK-NEXT:    [[TMP6:%.*]] = tail call <vscale x 16 x float> @llvm.vector.insert.nxv16f32.nxv4f32(<vscale x 16 x float> [[TMP4]], <vscale x 4 x float> [[TMP5]], i64 8)
+// CHECK-NEXT:    [[TMP7:%.*]] = extractvalue { <vscale x 4 x float>, <vscale x 4 x float>, <vscale x 4 x float>, <vscale x 4 x float> } [[TMP0]], 3
+// CHECK-NEXT:    [[TMP8:%.*]] = tail call <vscale x 16 x float> @llvm.vector.insert.nxv16f32.nxv4f32(<vscale x 16 x float> [[TMP6]], <vscale x 4 x float> [[TMP7]], i64 12)
+// CHECK-NEXT:    ret <vscale x 16 x float> [[TMP8]]
+//
+// CPP-CHECK-LABEL: @_Z26test_svread_za32_f32_vg1x4j(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call { <vscale x 4 x float>, <vscale x 4 x float>, <vscale x 4 x float>, <vscale x 4 x float> } @llvm.aarch64.sme.read.vg1x4.nxv4f32(i32 [[BASE:%.*]])
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = extractvalue { <vscale x 4 x float>, <vscale x 4 x float>, <vscale x 4 x float>, <vscale x 4 x float> } [[TMP0]], 0
+// CPP-CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 16 x float> @llvm.vector.insert.nxv16f32.nxv4f32(<vscale x 16 x float> poison, <vscale x 4 x float> [[TMP1]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP3:%.*]] = extractvalue { <vscale x 4 x float>, <vscale x 4 x float>, <vscale x 4 x float>, <vscale x 4 x float> } [[TMP0]], 1
+// CPP-CHECK-NEXT:    [[TMP4:%.*]] = tail call <vscale x 16 x float> @llvm.vector.insert.nxv16f32.nxv4f32(<vscale x 16 x float> [[TMP2]], <vscale x 4 x float> [[TMP3]], i64 4)
+// CPP-CHECK-NEXT:    [[TMP5:%.*]] = extractvalue { <vscale x 4 x float>, <vscale x 4 x float>, <vscale x 4 x float>, <vscale x 4 x float> } [[TMP0]], 2
+// CPP-CHECK-NEXT:    [[TMP6:%.*]] = tail call <vscale x 16 x float> @llvm.vector.insert.nxv16f32.nxv4f32(<vscale x 16 x float> [[TMP4]], <vscale x 4 x float> [[TMP5]], i64 8)
+// CPP-CHECK-NEXT:    [[TMP7:%.*]] = extractvalue { <vscale x 4 x float>, <vscale x 4 x float>, <vscale x 4 x float>, <vscale x 4 x float> } [[TMP0]], 3
+// CPP-CHECK-NEXT:    [[TMP8:%.*]] = tail call <vscale x 16 x float> @llvm.vector.insert.nxv16f32.nxv4f32(<vscale x 16 x float> [[TMP6]], <vscale x 4 x float> [[TMP7]], i64 12)
+// CPP-CHECK-NEXT:    ret <vscale x 16 x float> [[TMP8]]
+//
+svfloat32x4_t test_svread_za32_f32_vg1x4(uint32_t base) __arm_streaming __arm_shared_za __arm_preserves_za {
+  return svread_za32_f32_vg1x4(base);
+}
+
 // CHECK-LABEL: @test_svread_za64_u64_vg1x4(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = tail call { <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64> } @llvm.aarch64.sme.read.vg1x4.nxv2i64(i32 [[BASE:%.*]])
@@ -1419,4 +1887,3 @@ svfloat64x4_t test_svread_za64_f64_vg1x4(uint32_t base) __arm_streaming __arm_sh
 svint64x4_t test_svread_za64_s64_vg1x4(uint32_t base) __arm_streaming __arm_shared_za __arm_preserves_za {
   return svread_za64_s64_vg1x4(base);
 }
-
diff --git a/clang/test/CodeGen/aarch64-sme2-intrinsics/acle_sme2_write.c b/clang/test/CodeGen/aarch64-sme2-intrinsics/acle_sme2_write.c
index 99aa221834724..14b0371bce574 100644
--- a/clang/test/CodeGen/aarch64-sme2-intrinsics/acle_sme2_write.c
+++ b/clang/test/CodeGen/aarch64-sme2-intrinsics/acle_sme2_write.c
@@ -976,6 +976,168 @@ void test_svwrite_ver_za64_s64_vg4(uint32_t base, svint64x4_t val) __arm_streami
   SVE_ACLE_FUNC(svwrite_ver_za64,_s64,_vg4,)(7, base, val);
 }
 
+// CHECK-LABEL: @test_svwrite_za8_s8_vg1x2(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv32i8(<vscale x 32 x i8> [[VAL:%.*]], i64 0)
+// CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv32i8(<vscale x 32 x i8> [[VAL]], i64 16)
+// CHECK-NEXT:    tail call void @llvm.aarch64.sme.write.vg1x2.nxv16i8(i32 [[BASE:%.*]], <vscale x 16 x i8> [[TMP0]], <vscale x 16 x i8> [[TMP1]])
+// CHECK-NEXT:    ret void
+//
+// CPP-CHECK-LABEL: @_Z25test_svwrite_za8_s8_vg1x2j10svint8x2_t(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv32i8(<vscale x 32 x i8> [[VAL:%.*]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv32i8(<vscale x 32 x i8> [[VAL]], i64 16)
+// CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.write.vg1x2.nxv16i8(i32 [[BASE:%.*]], <vscale x 16 x i8> [[TMP0]], <vscale x 16 x i8> [[TMP1]])
+// CPP-CHECK-NEXT:    ret void
+//
+void test_svwrite_za8_s8_vg1x2(uint32_t base, svint8x2_t val) __arm_streaming __arm_shared_za {
+  SVE_ACLE_FUNC(svwrite_za8,_s8,_vg1x2,)(base, val);
+}
+
+// CHECK-LABEL: @test_svwrite_za8_u8_vg1x2(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv32i8(<vscale x 32 x i8> [[VAL:%.*]], i64 0)
+// CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv32i8(<vscale x 32 x i8> [[VAL]], i64 16)
+// CHECK-NEXT:    tail call void @llvm.aarch64.sme.write.vg1x2.nxv16i8(i32 [[BASE:%.*]], <vscale x 16 x i8> [[TMP0]], <vscale x 16 x i8> [[TMP1]])
+// CHECK-NEXT:    ret void
+//
+// CPP-CHECK-LABEL: @_Z25test_svwrite_za8_u8_vg1x2j11svuint8x2_t(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv32i8(<vscale x 32 x i8> [[VAL:%.*]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv32i8(<vscale x 32 x i8> [[VAL]], i64 16)
+// CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.write.vg1x2.nxv16i8(i32 [[BASE:%.*]], <vscale x 16 x i8> [[TMP0]], <vscale x 16 x i8> [[TMP1]])
+// CPP-CHECK-NEXT:    ret void
+//
+void test_svwrite_za8_u8_vg1x2(uint32_t base, svuint8x2_t val) __arm_streaming __arm_shared_za {
+  SVE_ACLE_FUNC(svwrite_za8,_u8,_vg1x2,)(base, val);
+}
+
+// CHECK-LABEL: @test_svwrite_za16_s16_vg1x2(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv16i16(<vscale x 16 x i16> [[VAL:%.*]], i64 0)
+// CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv16i16(<vscale x 16 x i16> [[VAL]], i64 8)
+// CHECK-NEXT:    tail call void @llvm.aarch64.sme.write.vg1x2.nxv8i16(i32 [[BASE:%.*]], <vscale x 8 x i16> [[TMP0]], <vscale x 8 x i16> [[TMP1]])
+// CHECK-NEXT:    ret void
+//
+// CPP-CHECK-LABEL: @_Z27test_svwrite_za16_s16_vg1x2j11svint16x2_t(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv16i16(<vscale x 16 x i16> [[VAL:%.*]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv16i16(<vscale x 16 x i16> [[VAL]], i64 8)
+// CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.write.vg1x2.nxv8i16(i32 [[BASE:%.*]], <vscale x 8 x i16> [[TMP0]], <vscale x 8 x i16> [[TMP1]])
+// CPP-CHECK-NEXT:    ret void
+//
+void test_svwrite_za16_s16_vg1x2(uint32_t base, svint16x2_t val) __arm_streaming __arm_shared_za {
+  SVE_ACLE_FUNC(svwrite_za16,_s16,_vg1x2,)(base, val);
+}
+
+// CHECK-LABEL: @test_svwrite_za16_u16_vg1x2(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv16i16(<vscale x 16 x i16> [[VAL:%.*]], i64 0)
+// CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv16i16(<vscale x 16 x i16> [[VAL]], i64 8)
+// CHECK-NEXT:    tail call void @llvm.aarch64.sme.write.vg1x2.nxv8i16(i32 [[BASE:%.*]], <vscale x 8 x i16> [[TMP0]], <vscale x 8 x i16> [[TMP1]])
+// CHECK-NEXT:    ret void
+//
+// CPP-CHECK-LABEL: @_Z27test_svwrite_za16_u16_vg1x2j12svuint16x2_t(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv16i16(<vscale x 16 x i16> [[VAL:%.*]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv16i16(<vscale x 16 x i16> [[VAL]], i64 8)
+// CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.write.vg1x2.nxv8i16(i32 [[BASE:%.*]], <vscale x 8 x i16> [[TMP0]], <vscale x 8 x i16> [[TMP1]])
+// CPP-CHECK-NEXT:    ret void
+//
+void test_svwrite_za16_u16_vg1x2(uint32_t base, svuint16x2_t val) __arm_streaming __arm_shared_za {
+  SVE_ACLE_FUNC(svwrite_za16,_u16,_vg1x2,)(base, val);
+}
+
+// CHECK-LABEL: @test_svwrite_za16_bf16_vg1x2(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 8 x bfloat> @llvm.vector.extract.nxv8bf16.nxv16bf16(<vscale x 16 x bfloat> [[VAL:%.*]], i64 0)
+// CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 8 x bfloat> @llvm.vector.extract.nxv8bf16.nxv16bf16(<vscale x 16 x bfloat> [[VAL]], i64 8)
+// CHECK-NEXT:    tail call void @llvm.aarch64.sme.write.vg1x2.nxv8bf16(i32 [[BASE:%.*]], <vscale x 8 x bfloat> [[TMP0]], <vscale x 8 x bfloat> [[TMP1]])
+// CHECK-NEXT:    ret void
+//
+// CPP-CHECK-LABEL: @_Z28test_svwrite_za16_bf16_vg1x2j14svbfloat16x2_t(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 8 x bfloat> @llvm.vector.extract.nxv8bf16.nxv16bf16(<vscale x 16 x bfloat> [[VAL:%.*]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 8 x bfloat> @llvm.vector.extract.nxv8bf16.nxv16bf16(<vscale x 16 x bfloat> [[VAL]], i64 8)
+// CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.write.vg1x2.nxv8bf16(i32 [[BASE:%.*]], <vscale x 8 x bfloat> [[TMP0]], <vscale x 8 x bfloat> [[TMP1]])
+// CPP-CHECK-NEXT:    ret void
+//
+void test_svwrite_za16_bf16_vg1x2(uint32_t base, svbfloat16x2_t val) __arm_streaming __arm_shared_za {
+  SVE_ACLE_FUNC(svwrite_za16,_bf16,_vg1x2,)(base, val);
+}
+
+// CHECK-LABEL: @test_svwrite_za16_f16_vg1x2(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 8 x half> @llvm.vector.extract.nxv8f16.nxv16f16(<vscale x 16 x half> [[VAL:%.*]], i64 0)
+// CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 8 x half> @llvm.vector.extract.nxv8f16.nxv16f16(<vscale x 16 x half> [[VAL]], i64 8)
+// CHECK-NEXT:    tail call void @llvm.aarch64.sme.write.vg1x2.nxv8f16(i32 [[BASE:%.*]], <vscale x 8 x half> [[TMP0]], <vscale x 8 x half> [[TMP1]])
+// CHECK-NEXT:    ret void
+//
+// CPP-CHECK-LABEL: @_Z27test_svwrite_za16_f16_vg1x2j13svfloat16x2_t(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 8 x half> @llvm.vector.extract.nxv8f16.nxv16f16(<vscale x 16 x half> [[VAL:%.*]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 8 x half> @llvm.vector.extract.nxv8f16.nxv16f16(<vscale x 16 x half> [[VAL]], i64 8)
+// CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.write.vg1x2.nxv8f16(i32 [[BASE:%.*]], <vscale x 8 x half> [[TMP0]], <vscale x 8 x half> [[TMP1]])
+// CPP-CHECK-NEXT:    ret void
+//
+void test_svwrite_za16_f16_vg1x2(uint32_t base, svfloat16x2_t val) __arm_streaming __arm_shared_za {
+  SVE_ACLE_FUNC(svwrite_za16,_f16,_vg1x2,)(base, val);
+}
+
+// CHECK-LABEL: @test_svwrite_za32_s32_vg1x2(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 4 x i32> @llvm.vector.extract.nxv4i32.nxv8i32(<vscale x 8 x i32> [[VAL:%.*]], i64 0)
+// CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 4 x i32> @llvm.vector.extract.nxv4i32.nxv8i32(<vscale x 8 x i32> [[VAL]], i64 4)
+// CHECK-NEXT:    tail call void @llvm.aarch64.sme.write.vg1x2.nxv4i32(i32 [[BASE:%.*]], <vscale x 4 x i32> [[TMP0]], <vscale x 4 x i32> [[TMP1]])
+// CHECK-NEXT:    ret void
+//
+// CPP-CHECK-LABEL: @_Z27test_svwrite_za32_s32_vg1x2j11svint32x2_t(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 4 x i32> @llvm.vector.extract.nxv4i32.nxv8i32(<vscale x 8 x i32> [[VAL:%.*]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 4 x i32> @llvm.vector.extract.nxv4i32.nxv8i32(<vscale x 8 x i32> [[VAL]], i64 4)
+// CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.write.vg1x2.nxv4i32(i32 [[BASE:%.*]], <vscale x 4 x i32> [[TMP0]], <vscale x 4 x i32> [[TMP1]])
+// CPP-CHECK-NEXT:    ret void
+//
+void test_svwrite_za32_s32_vg1x2(uint32_t base, svint32x2_t val) __arm_streaming __arm_shared_za {
+  SVE_ACLE_FUNC(svwrite_za32,_s32,_vg1x2,)(base, val);
+}
+
+// CHECK-LABEL: @test_svwrite_za32_u32_vg1x2(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 4 x i32> @llvm.vector.extract.nxv4i32.nxv8i32(<vscale x 8 x i32> [[VAL:%.*]], i64 0)
+// CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 4 x i32> @llvm.vector.extract.nxv4i32.nxv8i32(<vscale x 8 x i32> [[VAL]], i64 4)
+// CHECK-NEXT:    tail call void @llvm.aarch64.sme.write.vg1x2.nxv4i32(i32 [[BASE:%.*]], <vscale x 4 x i32> [[TMP0]], <vscale x 4 x i32> [[TMP1]])
+// CHECK-NEXT:    ret void
+//
+// CPP-CHECK-LABEL: @_Z27test_svwrite_za32_u32_vg1x2j12svuint32x2_t(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 4 x i32> @llvm.vector.extract.nxv4i32.nxv8i32(<vscale x 8 x i32> [[VAL:%.*]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 4 x i32> @llvm.vector.extract.nxv4i32.nxv8i32(<vscale x 8 x i32> [[VAL]], i64 4)
+// CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.write.vg1x2.nxv4i32(i32 [[BASE:%.*]], <vscale x 4 x i32> [[TMP0]], <vscale x 4 x i32> [[TMP1]])
+// CPP-CHECK-NEXT:    ret void
+//
+void test_svwrite_za32_u32_vg1x2(uint32_t base, svuint32x2_t val) __arm_streaming __arm_shared_za {
+  SVE_ACLE_FUNC(svwrite_za32,_u32,_vg1x2,)(base, val);
+}
+
+// CHECK-LABEL: @test_svwrite_za32_f32_vg1x2(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 4 x float> @llvm.vector.extract.nxv4f32.nxv8f32(<vscale x 8 x float> [[VAL:%.*]], i64 0)
+// CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 4 x float> @llvm.vector.extract.nxv4f32.nxv8f32(<vscale x 8 x float> [[VAL]], i64 4)
+// CHECK-NEXT:    tail call void @llvm.aarch64.sme.write.vg1x2.nxv4f32(i32 [[BASE:%.*]], <vscale x 4 x float> [[TMP0]], <vscale x 4 x float> [[TMP1]])
+// CHECK-NEXT:    ret void
+//
+// CPP-CHECK-LABEL: @_Z27test_svwrite_za32_f32_vg1x2j13svfloat32x2_t(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 4 x float> @llvm.vector.extract.nxv4f32.nxv8f32(<vscale x 8 x float> [[VAL:%.*]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 4 x float> @llvm.vector.extract.nxv4f32.nxv8f32(<vscale x 8 x float> [[VAL]], i64 4)
+// CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.write.vg1x2.nxv4f32(i32 [[BASE:%.*]], <vscale x 4 x float> [[TMP0]], <vscale x 4 x float> [[TMP1]])
+// CPP-CHECK-NEXT:    ret void
+//
+void test_svwrite_za32_f32_vg1x2(uint32_t base, svfloat32x2_t val) __arm_streaming __arm_shared_za {
+  SVE_ACLE_FUNC(svwrite_za32,_f32,_vg1x2,)(base, val);
+}
+
 // CHECK-LABEL: @test_svwrite_za64_u64_vg1x2(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 2 x i64> @llvm.vector.extract.nxv2i64.nxv4i64(<vscale x 4 x i64> [[VAL:%.*]], i64 0)
@@ -1030,6 +1192,204 @@ void test_svwrite_za64_s64_vg1x2(uint32_t base, svint64x2_t val) __arm_streaming
   SVE_ACLE_FUNC(svwrite_za64,_s64,_vg1x2,)(base, val);
 }
 
+// CHECK-LABEL: @test_svwrite_za8_s8_vg1x4(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[VAL:%.*]], i64 0)
+// CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[VAL]], i64 16)
+// CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[VAL]], i64 32)
+// CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[VAL]], i64 48)
+// CHECK-NEXT:    tail call void @llvm.aarch64.sme.write.vg1x4.nxv16i8(i32 [[BASE:%.*]], <vscale x 16 x i8> [[TMP0]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP2]], <vscale x 16 x i8> [[TMP3]])
+// CHECK-NEXT:    ret void
+//
+// CPP-CHECK-LABEL: @_Z25test_svwrite_za8_s8_vg1x4j10svint8x4_t(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[VAL:%.*]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[VAL]], i64 16)
+// CPP-CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[VAL]], i64 32)
+// CPP-CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[VAL]], i64 48)
+// CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.write.vg1x4.nxv16i8(i32 [[BASE:%.*]], <vscale x 16 x i8> [[TMP0]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP2]], <vscale x 16 x i8> [[TMP3]])
+// CPP-CHECK-NEXT:    ret void
+//
+void test_svwrite_za8_s8_vg1x4(uint32_t base, svint8x4_t val) __arm_streaming __arm_shared_za {
+  SVE_ACLE_FUNC(svwrite_za8,_s8,_vg1x4,)(base, val);
+}
+
+// CHECK-LABEL: @test_svwrite_za8_u8_vg1x4(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[VAL:%.*]], i64 0)
+// CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[VAL]], i64 16)
+// CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[VAL]], i64 32)
+// CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[VAL]], i64 48)
+// CHECK-NEXT:    tail call void @llvm.aarch64.sme.write.vg1x4.nxv16i8(i32 [[BASE:%.*]], <vscale x 16 x i8> [[TMP0]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP2]], <vscale x 16 x i8> [[TMP3]])
+// CHECK-NEXT:    ret void
+//
+// CPP-CHECK-LABEL: @_Z25test_svwrite_za8_u8_vg1x4j11svuint8x4_t(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[VAL:%.*]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[VAL]], i64 16)
+// CPP-CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[VAL]], i64 32)
+// CPP-CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv64i8(<vscale x 64 x i8> [[VAL]], i64 48)
+// CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.write.vg1x4.nxv16i8(i32 [[BASE:%.*]], <vscale x 16 x i8> [[TMP0]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP2]], <vscale x 16 x i8> [[TMP3]])
+// CPP-CHECK-NEXT:    ret void
+//
+void test_svwrite_za8_u8_vg1x4(uint32_t base, svuint8x4_t val) __arm_streaming __arm_shared_za {
+  SVE_ACLE_FUNC(svwrite_za8,_u8,_vg1x4,)(base, val);
+}
+
+// CHECK-LABEL: @test_svwrite_za16_s16_vg1x4(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[VAL:%.*]], i64 0)
+// CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[VAL]], i64 8)
+// CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[VAL]], i64 16)
+// CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[VAL]], i64 24)
+// CHECK-NEXT:    tail call void @llvm.aarch64.sme.write.vg1x4.nxv8i16(i32 [[BASE:%.*]], <vscale x 8 x i16> [[TMP0]], <vscale x 8 x i16> [[TMP1]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP3]])
+// CHECK-NEXT:    ret void
+//
+// CPP-CHECK-LABEL: @_Z27test_svwrite_za16_s16_vg1x4j11svint16x4_t(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[VAL:%.*]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[VAL]], i64 8)
+// CPP-CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[VAL]], i64 16)
+// CPP-CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[VAL]], i64 24)
+// CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.write.vg1x4.nxv8i16(i32 [[BASE:%.*]], <vscale x 8 x i16> [[TMP0]], <vscale x 8 x i16> [[TMP1]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP3]])
+// CPP-CHECK-NEXT:    ret void
+//
+void test_svwrite_za16_s16_vg1x4(uint32_t base, svint16x4_t val) __arm_streaming __arm_shared_za {
+  SVE_ACLE_FUNC(svwrite_za16,_s16,_vg1x4,)(base, val);
+}
+
+// CHECK-LABEL: @test_svwrite_za16_u16_vg1x4(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[VAL:%.*]], i64 0)
+// CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[VAL]], i64 8)
+// CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[VAL]], i64 16)
+// CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[VAL]], i64 24)
+// CHECK-NEXT:    tail call void @llvm.aarch64.sme.write.vg1x4.nxv8i16(i32 [[BASE:%.*]], <vscale x 8 x i16> [[TMP0]], <vscale x 8 x i16> [[TMP1]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP3]])
+// CHECK-NEXT:    ret void
+//
+// CPP-CHECK-LABEL: @_Z27test_svwrite_za16_u16_vg1x4j12svuint16x4_t(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[VAL:%.*]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[VAL]], i64 8)
+// CPP-CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[VAL]], i64 16)
+// CPP-CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv32i16(<vscale x 32 x i16> [[VAL]], i64 24)
+// CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.write.vg1x4.nxv8i16(i32 [[BASE:%.*]], <vscale x 8 x i16> [[TMP0]], <vscale x 8 x i16> [[TMP1]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP3]])
+// CPP-CHECK-NEXT:    ret void
+//
+void test_svwrite_za16_u16_vg1x4(uint32_t base, svuint16x4_t val) __arm_streaming __arm_shared_za {
+  SVE_ACLE_FUNC(svwrite_za16,_u16,_vg1x4,)(base, val);
+}
+
+// CHECK-LABEL: @test_svwrite_za16_bf16_vg1x4(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 8 x bfloat> @llvm.vector.extract.nxv8bf16.nxv32bf16(<vscale x 32 x bfloat> [[VAL:%.*]], i64 0)
+// CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 8 x bfloat> @llvm.vector.extract.nxv8bf16.nxv32bf16(<vscale x 32 x bfloat> [[VAL]], i64 8)
+// CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 8 x bfloat> @llvm.vector.extract.nxv8bf16.nxv32bf16(<vscale x 32 x bfloat> [[VAL]], i64 16)
+// CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 8 x bfloat> @llvm.vector.extract.nxv8bf16.nxv32bf16(<vscale x 32 x bfloat> [[VAL]], i64 24)
+// CHECK-NEXT:    tail call void @llvm.aarch64.sme.write.vg1x4.nxv8bf16(i32 [[BASE:%.*]], <vscale x 8 x bfloat> [[TMP0]], <vscale x 8 x bfloat> [[TMP1]], <vscale x 8 x bfloat> [[TMP2]], <vscale x 8 x bfloat> [[TMP3]])
+// CHECK-NEXT:    ret void
+//
+// CPP-CHECK-LABEL: @_Z28test_svwrite_za16_bf16_vg1x4j14svbfloat16x4_t(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 8 x bfloat> @llvm.vector.extract.nxv8bf16.nxv32bf16(<vscale x 32 x bfloat> [[VAL:%.*]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 8 x bfloat> @llvm.vector.extract.nxv8bf16.nxv32bf16(<vscale x 32 x bfloat> [[VAL]], i64 8)
+// CPP-CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 8 x bfloat> @llvm.vector.extract.nxv8bf16.nxv32bf16(<vscale x 32 x bfloat> [[VAL]], i64 16)
+// CPP-CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 8 x bfloat> @llvm.vector.extract.nxv8bf16.nxv32bf16(<vscale x 32 x bfloat> [[VAL]], i64 24)
+// CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.write.vg1x4.nxv8bf16(i32 [[BASE:%.*]], <vscale x 8 x bfloat> [[TMP0]], <vscale x 8 x bfloat> [[TMP1]], <vscale x 8 x bfloat> [[TMP2]], <vscale x 8 x bfloat> [[TMP3]])
+// CPP-CHECK-NEXT:    ret void
+//
+void test_svwrite_za16_bf16_vg1x4(uint32_t base, svbfloat16x4_t val) __arm_streaming __arm_shared_za {
+  SVE_ACLE_FUNC(svwrite_za16,_bf16,_vg1x4,)(base, val);
+}
+
+// CHECK-LABEL: @test_svwrite_za16_f16_vg1x4(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 8 x half> @llvm.vector.extract.nxv8f16.nxv32f16(<vscale x 32 x half> [[VAL:%.*]], i64 0)
+// CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 8 x half> @llvm.vector.extract.nxv8f16.nxv32f16(<vscale x 32 x half> [[VAL]], i64 8)
+// CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 8 x half> @llvm.vector.extract.nxv8f16.nxv32f16(<vscale x 32 x half> [[VAL]], i64 16)
+// CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 8 x half> @llvm.vector.extract.nxv8f16.nxv32f16(<vscale x 32 x half> [[VAL]], i64 24)
+// CHECK-NEXT:    tail call void @llvm.aarch64.sme.write.vg1x4.nxv8f16(i32 [[BASE:%.*]], <vscale x 8 x half> [[TMP0]], <vscale x 8 x half> [[TMP1]], <vscale x 8 x half> [[TMP2]], <vscale x 8 x half> [[TMP3]])
+// CHECK-NEXT:    ret void
+//
+// CPP-CHECK-LABEL: @_Z27test_svwrite_za16_f16_vg1x4j13svfloat16x4_t(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 8 x half> @llvm.vector.extract.nxv8f16.nxv32f16(<vscale x 32 x half> [[VAL:%.*]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 8 x half> @llvm.vector.extract.nxv8f16.nxv32f16(<vscale x 32 x half> [[VAL]], i64 8)
+// CPP-CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 8 x half> @llvm.vector.extract.nxv8f16.nxv32f16(<vscale x 32 x half> [[VAL]], i64 16)
+// CPP-CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 8 x half> @llvm.vector.extract.nxv8f16.nxv32f16(<vscale x 32 x half> [[VAL]], i64 24)
+// CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.write.vg1x4.nxv8f16(i32 [[BASE:%.*]], <vscale x 8 x half> [[TMP0]], <vscale x 8 x half> [[TMP1]], <vscale x 8 x half> [[TMP2]], <vscale x 8 x half> [[TMP3]])
+// CPP-CHECK-NEXT:    ret void
+//
+void test_svwrite_za16_f16_vg1x4(uint32_t base, svfloat16x4_t val) __arm_streaming __arm_shared_za {
+  SVE_ACLE_FUNC(svwrite_za16,_f16,_vg1x4,)(base, val);
+}
+
+// CHECK-LABEL: @test_svwrite_za32_s32_vg1x4(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 4 x i32> @llvm.vector.extract.nxv4i32.nxv16i32(<vscale x 16 x i32> [[VAL:%.*]], i64 0)
+// CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 4 x i32> @llvm.vector.extract.nxv4i32.nxv16i32(<vscale x 16 x i32> [[VAL]], i64 4)
+// CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 4 x i32> @llvm.vector.extract.nxv4i32.nxv16i32(<vscale x 16 x i32> [[VAL]], i64 8)
+// CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 4 x i32> @llvm.vector.extract.nxv4i32.nxv16i32(<vscale x 16 x i32> [[VAL]], i64 12)
+// CHECK-NEXT:    tail call void @llvm.aarch64.sme.write.vg1x4.nxv4i32(i32 [[BASE:%.*]], <vscale x 4 x i32> [[TMP0]], <vscale x 4 x i32> [[TMP1]], <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> [[TMP3]])
+// CHECK-NEXT:    ret void
+//
+// CPP-CHECK-LABEL: @_Z27test_svwrite_za32_s32_vg1x4j11svint32x4_t(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 4 x i32> @llvm.vector.extract.nxv4i32.nxv16i32(<vscale x 16 x i32> [[VAL:%.*]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 4 x i32> @llvm.vector.extract.nxv4i32.nxv16i32(<vscale x 16 x i32> [[VAL]], i64 4)
+// CPP-CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 4 x i32> @llvm.vector.extract.nxv4i32.nxv16i32(<vscale x 16 x i32> [[VAL]], i64 8)
+// CPP-CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 4 x i32> @llvm.vector.extract.nxv4i32.nxv16i32(<vscale x 16 x i32> [[VAL]], i64 12)
+// CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.write.vg1x4.nxv4i32(i32 [[BASE:%.*]], <vscale x 4 x i32> [[TMP0]], <vscale x 4 x i32> [[TMP1]], <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> [[TMP3]])
+// CPP-CHECK-NEXT:    ret void
+//
+void test_svwrite_za32_s32_vg1x4(uint32_t base, svint32x4_t val) __arm_streaming __arm_shared_za {
+  SVE_ACLE_FUNC(svwrite_za32,_s32,_vg1x4,)(base, val);
+}
+
+// CHECK-LABEL: @test_svwrite_za32_u32_vg1x4(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 4 x i32> @llvm.vector.extract.nxv4i32.nxv16i32(<vscale x 16 x i32> [[VAL:%.*]], i64 0)
+// CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 4 x i32> @llvm.vector.extract.nxv4i32.nxv16i32(<vscale x 16 x i32> [[VAL]], i64 4)
+// CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 4 x i32> @llvm.vector.extract.nxv4i32.nxv16i32(<vscale x 16 x i32> [[VAL]], i64 8)
+// CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 4 x i32> @llvm.vector.extract.nxv4i32.nxv16i32(<vscale x 16 x i32> [[VAL]], i64 12)
+// CHECK-NEXT:    tail call void @llvm.aarch64.sme.write.vg1x4.nxv4i32(i32 [[BASE:%.*]], <vscale x 4 x i32> [[TMP0]], <vscale x 4 x i32> [[TMP1]], <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> [[TMP3]])
+// CHECK-NEXT:    ret void
+//
+// CPP-CHECK-LABEL: @_Z27test_svwrite_za32_u32_vg1x4j12svuint32x4_t(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 4 x i32> @llvm.vector.extract.nxv4i32.nxv16i32(<vscale x 16 x i32> [[VAL:%.*]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 4 x i32> @llvm.vector.extract.nxv4i32.nxv16i32(<vscale x 16 x i32> [[VAL]], i64 4)
+// CPP-CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 4 x i32> @llvm.vector.extract.nxv4i32.nxv16i32(<vscale x 16 x i32> [[VAL]], i64 8)
+// CPP-CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 4 x i32> @llvm.vector.extract.nxv4i32.nxv16i32(<vscale x 16 x i32> [[VAL]], i64 12)
+// CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.write.vg1x4.nxv4i32(i32 [[BASE:%.*]], <vscale x 4 x i32> [[TMP0]], <vscale x 4 x i32> [[TMP1]], <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> [[TMP3]])
+// CPP-CHECK-NEXT:    ret void
+//
+void test_svwrite_za32_u32_vg1x4(uint32_t base, svuint32x4_t val) __arm_streaming __arm_shared_za {
+  SVE_ACLE_FUNC(svwrite_za32,_u32,_vg1x4,)(base, val);
+}
+
+// CHECK-LABEL: @test_svwrite_za32_f32_vg1x4(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 4 x float> @llvm.vector.extract.nxv4f32.nxv16f32(<vscale x 16 x float> [[VAL:%.*]], i64 0)
+// CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 4 x float> @llvm.vector.extract.nxv4f32.nxv16f32(<vscale x 16 x float> [[VAL]], i64 4)
+// CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 4 x float> @llvm.vector.extract.nxv4f32.nxv16f32(<vscale x 16 x float> [[VAL]], i64 8)
+// CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 4 x float> @llvm.vector.extract.nxv4f32.nxv16f32(<vscale x 16 x float> [[VAL]], i64 12)
+// CHECK-NEXT:    tail call void @llvm.aarch64.sme.write.vg1x4.nxv4f32(i32 [[BASE:%.*]], <vscale x 4 x float> [[TMP0]], <vscale x 4 x float> [[TMP1]], <vscale x 4 x float> [[TMP2]], <vscale x 4 x float> [[TMP3]])
+// CHECK-NEXT:    ret void
+//
+// CPP-CHECK-LABEL: @_Z27test_svwrite_za32_f32_vg1x4j13svfloat32x4_t(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 4 x float> @llvm.vector.extract.nxv4f32.nxv16f32(<vscale x 16 x float> [[VAL:%.*]], i64 0)
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = tail call <vscale x 4 x float> @llvm.vector.extract.nxv4f32.nxv16f32(<vscale x 16 x float> [[VAL]], i64 4)
+// CPP-CHECK-NEXT:    [[TMP2:%.*]] = tail call <vscale x 4 x float> @llvm.vector.extract.nxv4f32.nxv16f32(<vscale x 16 x float> [[VAL]], i64 8)
+// CPP-CHECK-NEXT:    [[TMP3:%.*]] = tail call <vscale x 4 x float> @llvm.vector.extract.nxv4f32.nxv16f32(<vscale x 16 x float> [[VAL]], i64 12)
+// CPP-CHECK-NEXT:    tail call void @llvm.aarch64.sme.write.vg1x4.nxv4f32(i32 [[BASE:%.*]], <vscale x 4 x float> [[TMP0]], <vscale x 4 x float> [[TMP1]], <vscale x 4 x float> [[TMP2]], <vscale x 4 x float> [[TMP3]])
+// CPP-CHECK-NEXT:    ret void
+//
+void test_svwrite_za32_f32_vg1x4(uint32_t base, svfloat32x4_t val) __arm_streaming __arm_shared_za {
+  SVE_ACLE_FUNC(svwrite_za32,_f32,_vg1x4,)(base, val);
+}
+
 // CHECK-LABEL: @test_svwrite_za64_u64_vg1x4(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = tail call <vscale x 2 x i64> @llvm.vector.extract.nxv2i64.nxv8i64(<vscale x 8 x i64> [[VAL:%.*]], i64 0)
diff --git a/llvm/lib/Target/AArch64/SMEInstrFormats.td b/llvm/lib/Target/AArch64/SMEInstrFormats.td
index 408b897070af0..46353c60fe851 100644
--- a/llvm/lib/Target/AArch64/SMEInstrFormats.td
+++ b/llvm/lib/Target/AArch64/SMEInstrFormats.td
@@ -3686,8 +3686,14 @@ multiclass sme2_mova_vec_to_array_vg2_multi<string mnemonic, SDPatternOperator i
 
   def NAME # _PSEUDO : sme2_move_to_za_pseudo<NAME, sme_elm_idx0_7, ZZ_d_mul_r, SMEMatrixArray>;
 
-  def : SME2_ZA_VG1x2_Multi_Pat<NAME, intrinsic, nxv2i64, sme_elm_idx0_7, tileslice16>;
-  def : SME2_ZA_VG1x2_Multi_Pat<NAME, intrinsic, nxv2f64, sme_elm_idx0_7, tileslice16>;
+  def : SME2_ZA_VG1x2_Multi_Pat<NAME, intrinsic, nxv16i8,  sme_elm_idx0_7, tileslice16>;
+  def : SME2_ZA_VG1x2_Multi_Pat<NAME, intrinsic, nxv8i16,  sme_elm_idx0_7, tileslice16>;
+  def : SME2_ZA_VG1x2_Multi_Pat<NAME, intrinsic, nxv8f16,  sme_elm_idx0_7, tileslice16>;
+  def : SME2_ZA_VG1x2_Multi_Pat<NAME, intrinsic, nxv8bf16, sme_elm_idx0_7, tileslice16>;
+  def : SME2_ZA_VG1x2_Multi_Pat<NAME, intrinsic, nxv4i32,  sme_elm_idx0_7, tileslice16>;
+  def : SME2_ZA_VG1x2_Multi_Pat<NAME, intrinsic, nxv4f32,  sme_elm_idx0_7, tileslice16>;
+  def : SME2_ZA_VG1x2_Multi_Pat<NAME, intrinsic, nxv2i64,  sme_elm_idx0_7, tileslice16>;
+  def : SME2_ZA_VG1x2_Multi_Pat<NAME, intrinsic, nxv2f64,  sme_elm_idx0_7, tileslice16>;
 
   defm : sme2_mova_vec_to_tile_or_array_aliases<0, !cast<Instruction>(NAME),
                                                 MatrixOp8,
@@ -3779,8 +3785,14 @@ multiclass sme2_mova_vec_to_array_vg4_multi<string mnemonic, SDPatternOperator i
 
   def NAME # _PSEUDO : sme2_move_to_za_pseudo<NAME, sme_elm_idx0_7, ZZZZ_d_mul_r, SMEMatrixArray>;
 
-  def : SME2_ZA_VG1x4_Multi_Pat<NAME, intrinsic, nxv2i64, sme_elm_idx0_7, tileslice16>;
-  def : SME2_ZA_VG1x4_Multi_Pat<NAME, intrinsic, nxv2f64, sme_elm_idx0_7, tileslice16>;
+  def : SME2_ZA_VG1x4_Multi_Pat<NAME, intrinsic, nxv16i8,  sme_elm_idx0_7, tileslice16>;
+  def : SME2_ZA_VG1x4_Multi_Pat<NAME, intrinsic, nxv8i16,  sme_elm_idx0_7, tileslice16>;
+  def : SME2_ZA_VG1x4_Multi_Pat<NAME, intrinsic, nxv8f16,  sme_elm_idx0_7, tileslice16>;
+  def : SME2_ZA_VG1x4_Multi_Pat<NAME, intrinsic, nxv8bf16, sme_elm_idx0_7, tileslice16>;
+  def : SME2_ZA_VG1x4_Multi_Pat<NAME, intrinsic, nxv4i32,  sme_elm_idx0_7, tileslice16>;
+  def : SME2_ZA_VG1x4_Multi_Pat<NAME, intrinsic, nxv4f32,  sme_elm_idx0_7, tileslice16>;
+  def : SME2_ZA_VG1x4_Multi_Pat<NAME, intrinsic, nxv2i64,  sme_elm_idx0_7, tileslice16>;
+  def : SME2_ZA_VG1x4_Multi_Pat<NAME, intrinsic, nxv2f64,  sme_elm_idx0_7, tileslice16>;
 
   defm : sme2_mova_vec_to_tile_or_array_aliases<0, !cast<Instruction>(NAME),
                                                 MatrixOp8,
diff --git a/llvm/test/CodeGen/AArch64/sme2-intrinsics-extract-mova.ll b/llvm/test/CodeGen/AArch64/sme2-intrinsics-extract-mova.ll
index ad1c524d5610e..1e1895fc54c2a 100644
--- a/llvm/test/CodeGen/AArch64/sme2-intrinsics-extract-mova.ll
+++ b/llvm/test/CodeGen/AArch64/sme2-intrinsics-extract-mova.ll
@@ -399,6 +399,84 @@ define { <vscale x 2 x double>, <vscale x 2 x double>, <vscale x 2 x double>, <v
 
 ; Move Multi-Vector From ZA (Read) x2
 
+define { <vscale x 16 x i8>, <vscale x 16 x i8> } @za_read_vg1x2_b(i32 %slice) {
+; CHECK-LABEL: za_read_vg1x2_b:
+; CHECK:       // %bb.0:
+; CHECK-NEXT:    mov w8, w0
+; CHECK-NEXT:    mov { z0.d, z1.d }, za.d[w8, 0, vgx2]
+; CHECK-NEXT:    mov { z0.d, z1.d }, za.d[w8, 7, vgx2]
+; CHECK-NEXT:    ret
+  %res = call { <vscale x 16 x i8>, <vscale x 16 x i8> } @llvm.aarch64.sme.read.vg1x2.nxv16i8(i32 %slice)
+  %slice.7 = add i32 %slice, 7
+  %res2 = call { <vscale x 16 x i8>, <vscale x 16 x i8> } @llvm.aarch64.sme.read.vg1x2.nxv16i8(i32 %slice.7)
+  ret { <vscale x 16 x i8>, <vscale x 16 x i8> } %res2
+}
+
+define { <vscale x 8 x i16>, <vscale x 8 x i16> } @za_read_vg1x2_h(i32 %slice) {
+; CHECK-LABEL: za_read_vg1x2_h:
+; CHECK:       // %bb.0:
+; CHECK-NEXT:    mov w8, w0
+; CHECK-NEXT:    mov { z0.d, z1.d }, za.d[w8, 0, vgx2]
+; CHECK-NEXT:    mov { z0.d, z1.d }, za.d[w8, 7, vgx2]
+; CHECK-NEXT:    ret
+  %res = call { <vscale x 8 x i16>, <vscale x 8 x i16> } @llvm.aarch64.sme.read.vg1x2.nxv8i16(i32 %slice)
+  %slice.7 = add i32 %slice, 7
+  %res2 = call { <vscale x 8 x i16>, <vscale x 8 x i16> } @llvm.aarch64.sme.read.vg1x2.nxv8i16(i32 %slice.7)
+  ret { <vscale x 8 x i16>, <vscale x 8 x i16> } %res2
+}
+
+define { <vscale x 8 x half>, <vscale x 8 x half> } @za_read_vg1x2_f16(i32 %slice) {
+; CHECK-LABEL: za_read_vg1x2_f16:
+; CHECK:       // %bb.0:
+; CHECK-NEXT:    mov w8, w0
+; CHECK-NEXT:    mov { z0.d, z1.d }, za.d[w8, 0, vgx2]
+; CHECK-NEXT:    mov { z0.d, z1.d }, za.d[w8, 7, vgx2]
+; CHECK-NEXT:    ret
+  %res = call { <vscale x 8 x half>, <vscale x 8 x half> } @llvm.aarch64.sme.read.vg1x2.nxv8f16(i32 %slice)
+  %slice.7 = add i32 %slice, 7
+  %res2 = call { <vscale x 8 x half>, <vscale x 8 x half> } @llvm.aarch64.sme.read.vg1x2.nxv8f16(i32 %slice.7)
+  ret { <vscale x 8 x half>, <vscale x 8 x half> } %res2
+}
+
+define { <vscale x 8 x bfloat>, <vscale x 8 x bfloat> } @za_read_vg1x2_bf16(i32 %slice) {
+; CHECK-LABEL: za_read_vg1x2_bf16:
+; CHECK:       // %bb.0:
+; CHECK-NEXT:    mov w8, w0
+; CHECK-NEXT:    mov { z0.d, z1.d }, za.d[w8, 0, vgx2]
+; CHECK-NEXT:    mov { z0.d, z1.d }, za.d[w8, 7, vgx2]
+; CHECK-NEXT:    ret
+  %res = call { <vscale x 8 x bfloat>, <vscale x 8 x bfloat> } @llvm.aarch64.sme.read.vg1x2.nxv8bf16(i32 %slice)
+  %slice.7 = add i32 %slice, 7
+  %res2 = call { <vscale x 8 x bfloat>, <vscale x 8 x bfloat> } @llvm.aarch64.sme.read.vg1x2.nxv8bf16(i32 %slice.7)
+  ret { <vscale x 8 x bfloat>, <vscale x 8 x bfloat> } %res2
+}
+
+define { <vscale x 4 x i32>, <vscale x 4 x i32> } @za_read_vg1x2_s(i32 %slice) {
+; CHECK-LABEL: za_read_vg1x2_s:
+; CHECK:       // %bb.0:
+; CHECK-NEXT:    mov w8, w0
+; CHECK-NEXT:    mov { z0.d, z1.d }, za.d[w8, 0, vgx2]
+; CHECK-NEXT:    mov { z0.d, z1.d }, za.d[w8, 7, vgx2]
+; CHECK-NEXT:    ret
+  %res = call { <vscale x 4 x i32>, <vscale x 4 x i32> } @llvm.aarch64.sme.read.vg1x2.nxv4i32(i32 %slice)
+  %slice.7 = add i32 %slice, 7
+  %res2 = call { <vscale x 4 x i32>, <vscale x 4 x i32> } @llvm.aarch64.sme.read.vg1x2.nxv4i32(i32 %slice.7)
+  ret { <vscale x 4 x i32>, <vscale x 4 x i32> } %res2
+}
+
+define { <vscale x 4 x float>, <vscale x 4 x float> } @za_read_vg1x2_f32(i32 %slice) {
+; CHECK-LABEL: za_read_vg1x2_f32:
+; CHECK:       // %bb.0:
+; CHECK-NEXT:    mov w8, w0
+; CHECK-NEXT:    mov { z0.d, z1.d }, za.d[w8, 0, vgx2]
+; CHECK-NEXT:    mov { z0.d, z1.d }, za.d[w8, 7, vgx2]
+; CHECK-NEXT:    ret
+  %res = call { <vscale x 4 x float>, <vscale x 4 x float> } @llvm.aarch64.sme.read.vg1x2.nxv4f32(i32 %slice)
+  %slice.7 = add i32 %slice, 7
+  %res2 = call { <vscale x 4 x float>, <vscale x 4 x float> } @llvm.aarch64.sme.read.vg1x2.nxv4f32(i32 %slice.7)
+  ret { <vscale x 4 x float>, <vscale x 4 x float> } %res2
+}
+
 define { <vscale x 2 x i64>, <vscale x 2 x i64> } @za_read_vg1x2_d(i32 %slice) {
 ; CHECK-LABEL: za_read_vg1x2_d:
 ; CHECK:       // %bb.0:
@@ -427,6 +505,84 @@ define { <vscale x 2 x double>, <vscale x 2 x double> } @za_read_vg1x2_f64(i32 %
 
 ; Move Multi-Vector From ZA (Read) x4
 
+define { <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8> } @za_read_vg1x4_b(i32 %slice) {
+; CHECK-LABEL: za_read_vg1x4_b:
+; CHECK:       // %bb.0:
+; CHECK-NEXT:    mov w8, w0
+; CHECK-NEXT:    mov { z0.d - z3.d }, za.d[w8, 0, vgx4]
+; CHECK-NEXT:    mov { z0.d - z3.d }, za.d[w8, 7, vgx4]
+; CHECK-NEXT:    ret
+  %res = call { <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8> } @llvm.aarch64.sme.read.vg1x4.nxv16i8(i32 %slice)
+  %slice.7 = add i32 %slice, 7
+  %res2 = call { <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8> } @llvm.aarch64.sme.read.vg1x4.nxv16i8(i32 %slice.7)
+  ret { <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8> } %res2
+}
+
+define { <vscale x 8 x i16>, <vscale x 8 x i16>, <vscale x 8 x i16>, <vscale x 8 x i16> } @za_read_vg1x4_h(i32 %slice) {
+; CHECK-LABEL: za_read_vg1x4_h:
+; CHECK:       // %bb.0:
+; CHECK-NEXT:    mov w8, w0
+; CHECK-NEXT:    mov { z0.d - z3.d }, za.d[w8, 0, vgx4]
+; CHECK-NEXT:    mov { z0.d - z3.d }, za.d[w8, 7, vgx4]
+; CHECK-NEXT:    ret
+  %res = call { <vscale x 8 x i16>, <vscale x 8 x i16>, <vscale x 8 x i16>, <vscale x 8 x i16> } @llvm.aarch64.sme.read.vg1x4.nxv8i16(i32 %slice)
+  %slice.7 = add i32 %slice, 7
+  %res2 = call { <vscale x 8 x i16>, <vscale x 8 x i16>, <vscale x 8 x i16>, <vscale x 8 x i16> } @llvm.aarch64.sme.read.vg1x4.nxv8i16(i32 %slice.7)
+  ret { <vscale x 8 x i16>, <vscale x 8 x i16>, <vscale x 8 x i16>, <vscale x 8 x i16> } %res2
+}
+
+define { <vscale x 8 x half>, <vscale x 8 x half>, <vscale x 8 x half>, <vscale x 8 x half> } @za_read_vg1x4_f16(i32 %slice) {
+; CHECK-LABEL: za_read_vg1x4_f16:
+; CHECK:       // %bb.0:
+; CHECK-NEXT:    mov w8, w0
+; CHECK-NEXT:    mov { z0.d - z3.d }, za.d[w8, 0, vgx4]
+; CHECK-NEXT:    mov { z0.d - z3.d }, za.d[w8, 7, vgx4]
+; CHECK-NEXT:    ret
+  %res = call { <vscale x 8 x half>, <vscale x 8 x half>, <vscale x 8 x half>, <vscale x 8 x half> } @llvm.aarch64.sme.read.vg1x4.nxv8f16(i32 %slice)
+  %slice.7 = add i32 %slice, 7
+  %res2 = call { <vscale x 8 x half>, <vscale x 8 x half>, <vscale x 8 x half>, <vscale x 8 x half> } @llvm.aarch64.sme.read.vg1x4.nxv8f16(i32 %slice.7)
+  ret { <vscale x 8 x half>, <vscale x 8 x half>, <vscale x 8 x half>, <vscale x 8 x half> } %res2
+}
+
+define { <vscale x 8 x bfloat>, <vscale x 8 x bfloat>, <vscale x 8 x bfloat>, <vscale x 8 x bfloat> } @za_read_vg1x4_bf16(i32 %slice) {
+; CHECK-LABEL: za_read_vg1x4_bf16:
+; CHECK:       // %bb.0:
+; CHECK-NEXT:    mov w8, w0
+; CHECK-NEXT:    mov { z0.d - z3.d }, za.d[w8, 0, vgx4]
+; CHECK-NEXT:    mov { z0.d - z3.d }, za.d[w8, 7, vgx4]
+; CHECK-NEXT:    ret
+  %res = call { <vscale x 8 x bfloat>, <vscale x 8 x bfloat>, <vscale x 8 x bfloat>, <vscale x 8 x bfloat> } @llvm.aarch64.sme.read.vg1x4.nxv8bf16(i32 %slice)
+  %slice.7 = add i32 %slice, 7
+  %res2 = call { <vscale x 8 x bfloat>, <vscale x 8 x bfloat>, <vscale x 8 x bfloat>, <vscale x 8 x bfloat> } @llvm.aarch64.sme.read.vg1x4.nxv8bf16(i32 %slice.7)
+  ret { <vscale x 8 x bfloat>, <vscale x 8 x bfloat>, <vscale x 8 x bfloat>, <vscale x 8 x bfloat> } %res2
+}
+
+define { <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32> } @za_read_vg1x4_s(i32 %slice) {
+; CHECK-LABEL: za_read_vg1x4_s:
+; CHECK:       // %bb.0:
+; CHECK-NEXT:    mov w8, w0
+; CHECK-NEXT:    mov { z0.d - z3.d }, za.d[w8, 0, vgx4]
+; CHECK-NEXT:    mov { z0.d - z3.d }, za.d[w8, 7, vgx4]
+; CHECK-NEXT:    ret
+  %res = call { <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32> } @llvm.aarch64.sme.read.vg1x4.nxv4i32(i32 %slice)
+  %slice.7 = add i32 %slice, 7
+  %res2 = call { <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32> } @llvm.aarch64.sme.read.vg1x4.nxv4i32(i32 %slice.7)
+  ret { <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32> } %res2
+}
+
+define { <vscale x 4 x float>, <vscale x 4 x float>, <vscale x 4 x float>, <vscale x 4 x float> } @za_read_vg1x4_f32(i32 %slice) {
+; CHECK-LABEL: za_read_vg1x4_f32:
+; CHECK:       // %bb.0:
+; CHECK-NEXT:    mov w8, w0
+; CHECK-NEXT:    mov { z0.d - z3.d }, za.d[w8, 0, vgx4]
+; CHECK-NEXT:    mov { z0.d - z3.d }, za.d[w8, 7, vgx4]
+; CHECK-NEXT:    ret
+  %res = call { <vscale x 4 x float>, <vscale x 4 x float>, <vscale x 4 x float>, <vscale x 4 x float> } @llvm.aarch64.sme.read.vg1x4.nxv4f32(i32 %slice)
+  %slice.7 = add i32 %slice, 7
+  %res2 = call { <vscale x 4 x float>, <vscale x 4 x float>, <vscale x 4 x float>, <vscale x 4 x float> } @llvm.aarch64.sme.read.vg1x4.nxv4f32(i32 %slice.7)
+  ret { <vscale x 4 x float>, <vscale x 4 x float>, <vscale x 4 x float>, <vscale x 4 x float> } %res2
+}
+
 define { <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64> } @za_read_vg1x4_d(i32 %slice) {
 ; CHECK-LABEL: za_read_vg1x4_d:
 ; CHECK:       // %bb.0:
@@ -489,8 +645,20 @@ declare { <vscale x 4 x float>, <vscale x 4 x float>, <vscale x 4 x float>, <vsc
 declare { <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64> } @llvm.aarch64.sme.read.ver.vg4.nxv2i64(i32, i32)
 declare { <vscale x 2 x double>, <vscale x 2 x double>, <vscale x 2 x double>, <vscale x 2 x double> } @llvm.aarch64.sme.read.ver.vg4.nxv2f64(i32, i32)
 
+declare { <vscale x 16 x i8>, <vscale x 16 x i8> } @llvm.aarch64.sme.read.vg1x2.nxv16i8(i32)
+declare { <vscale x 8 x i16>, <vscale x 8 x i16> } @llvm.aarch64.sme.read.vg1x2.nxv8i16(i32)
+declare { <vscale x 4 x i32>, <vscale x 4 x i32> } @llvm.aarch64.sme.read.vg1x2.nxv4i32(i32)
 declare { <vscale x 2 x i64>, <vscale x 2 x i64> } @llvm.aarch64.sme.read.vg1x2.nxv2i64(i32)
+declare { <vscale x 8 x half>, <vscale x 8 x half> } @llvm.aarch64.sme.read.vg1x2.nxv8f16(i32)
+declare { <vscale x 8 x bfloat>, <vscale x 8 x bfloat> } @llvm.aarch64.sme.read.vg1x2.nxv8bf16(i32)
+declare { <vscale x 4 x float>, <vscale x 4 x float> } @llvm.aarch64.sme.read.vg1x2.nxv4f32(i32)
 declare { <vscale x 2 x double>, <vscale x 2 x double> } @llvm.aarch64.sme.read.vg1x2.nxv2f64(i32)
 
+declare { <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8> } @llvm.aarch64.sme.read.vg1x4.nxv16i8(i32)
+declare { <vscale x 8 x i16>, <vscale x 8 x i16>, <vscale x 8 x i16>, <vscale x 8 x i16> } @llvm.aarch64.sme.read.vg1x4.nxv8i16(i32)
+declare { <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32> } @llvm.aarch64.sme.read.vg1x4.nxv4i32(i32)
 declare { <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64> } @llvm.aarch64.sme.read.vg1x4.nxv2i64(i32)
+declare { <vscale x 8 x half>, <vscale x 8 x half>, <vscale x 8 x half>, <vscale x 8 x half> } @llvm.aarch64.sme.read.vg1x4.nxv8f16(i32)
+declare { <vscale x 8 x bfloat>, <vscale x 8 x bfloat>, <vscale x 8 x bfloat>, <vscale x 8 x bfloat> } @llvm.aarch64.sme.read.vg1x4.nxv8bf16(i32)
+declare { <vscale x 4 x float>, <vscale x 4 x float>, <vscale x 4 x float>, <vscale x 4 x float> } @llvm.aarch64.sme.read.vg1x4.nxv4f32(i32)
 declare { <vscale x 2 x double>, <vscale x 2 x double>, <vscale x 2 x double>, <vscale x 2 x double> } @llvm.aarch64.sme.read.vg1x4.nxv2f64(i32)
diff --git a/llvm/test/CodeGen/AArch64/sme2-intrinsics-insert-mova.ll b/llvm/test/CodeGen/AArch64/sme2-intrinsics-insert-mova.ll
index 2712cbc1d7efd..99de6f832a3c9 100644
--- a/llvm/test/CodeGen/AArch64/sme2-intrinsics-insert-mova.ll
+++ b/llvm/test/CodeGen/AArch64/sme2-intrinsics-insert-mova.ll
@@ -497,6 +497,96 @@ define void @za_write_vg4_vert_f64(i32 %slice, <vscale x 2 x double> %zn1, <vsca
 ; Move Multi-Vector To ZA (Write) x2
 ;
 
+define void @za_write_vg1x2_b(i32 %slice, <vscale x 16 x i8> %za1, <vscale x 16 x i8> %za2) {
+; CHECK-LABEL: za_write_vg1x2_b:
+; CHECK:       // %bb.0:
+; CHECK-NEXT:    // kill: def $z1 killed $z1 killed $z0_z1 def $z0_z1
+; CHECK-NEXT:    mov w8, w0
+; CHECK-NEXT:    // kill: def $z0 killed $z0 killed $z0_z1 def $z0_z1
+; CHECK-NEXT:    mov za.d[w8, 0, vgx2], { z0.d, z1.d }
+; CHECK-NEXT:    mov za.d[w8, 7, vgx2], { z0.d, z1.d }
+; CHECK-NEXT:    ret
+  call void @llvm.aarch64.sme.write.vg1x2.nxv16i8(i32 %slice, <vscale x 16 x i8> %za1, <vscale x 16 x i8> %za2)
+  %slice.7 = add i32 %slice, 7
+  call void @llvm.aarch64.sme.write.vg1x2.nxv16i8(i32 %slice.7, <vscale x 16 x i8> %za1, <vscale x 16 x i8> %za2)
+  ret void
+}
+
+define void @za_write_vg1x2_h(i32 %slice, <vscale x 8 x i16> %za1, <vscale x 8 x i16> %za2) {
+; CHECK-LABEL: za_write_vg1x2_h:
+; CHECK:       // %bb.0:
+; CHECK-NEXT:    // kill: def $z1 killed $z1 killed $z0_z1 def $z0_z1
+; CHECK-NEXT:    mov w8, w0
+; CHECK-NEXT:    // kill: def $z0 killed $z0 killed $z0_z1 def $z0_z1
+; CHECK-NEXT:    mov za.d[w8, 0, vgx2], { z0.d, z1.d }
+; CHECK-NEXT:    mov za.d[w8, 7, vgx2], { z0.d, z1.d }
+; CHECK-NEXT:    ret
+  call void @llvm.aarch64.sme.write.vg1x2.nxv8i16(i32 %slice, <vscale x 8 x i16> %za1, <vscale x 8 x i16> %za2)
+  %slice.7 = add i32 %slice, 7
+  call void @llvm.aarch64.sme.write.vg1x2.nxv8i16(i32 %slice.7, <vscale x 8 x i16> %za1, <vscale x 8 x i16> %za2)
+  ret void
+}
+
+define void @za_write_vg1x2_f16(i32 %slice, <vscale x 8 x half> %za1, <vscale x 8 x half> %za2) {
+; CHECK-LABEL: za_write_vg1x2_f16:
+; CHECK:       // %bb.0:
+; CHECK-NEXT:    // kill: def $z1 killed $z1 killed $z0_z1 def $z0_z1
+; CHECK-NEXT:    mov w8, w0
+; CHECK-NEXT:    // kill: def $z0 killed $z0 killed $z0_z1 def $z0_z1
+; CHECK-NEXT:    mov za.d[w8, 0, vgx2], { z0.d, z1.d }
+; CHECK-NEXT:    mov za.d[w8, 7, vgx2], { z0.d, z1.d }
+; CHECK-NEXT:    ret
+  call void @llvm.aarch64.sme.write.vg1x2.nxv8f16(i32 %slice, <vscale x 8 x half> %za1, <vscale x 8 x half> %za2)
+  %slice.7 = add i32 %slice, 7
+  call void @llvm.aarch64.sme.write.vg1x2.nxv8f16(i32 %slice.7, <vscale x 8 x half> %za1, <vscale x 8 x half> %za2)
+  ret void
+}
+
+define void @za_write_vg1x2_bf16(i32 %slice, <vscale x 8 x bfloat> %za1, <vscale x 8 x bfloat> %za2) {
+; CHECK-LABEL: za_write_vg1x2_bf16:
+; CHECK:       // %bb.0:
+; CHECK-NEXT:    // kill: def $z1 killed $z1 killed $z0_z1 def $z0_z1
+; CHECK-NEXT:    mov w8, w0
+; CHECK-NEXT:    // kill: def $z0 killed $z0 killed $z0_z1 def $z0_z1
+; CHECK-NEXT:    mov za.d[w8, 0, vgx2], { z0.d, z1.d }
+; CHECK-NEXT:    mov za.d[w8, 7, vgx2], { z0.d, z1.d }
+; CHECK-NEXT:    ret
+  call void @llvm.aarch64.sme.write.vg1x2.nxv8bf16(i32 %slice, <vscale x 8 x bfloat> %za1, <vscale x 8 x bfloat> %za2)
+  %slice.7 = add i32 %slice, 7
+  call void @llvm.aarch64.sme.write.vg1x2.nxv8bf16(i32 %slice.7, <vscale x 8 x bfloat> %za1, <vscale x 8 x bfloat> %za2)
+  ret void
+}
+
+define void @za_write_vg1x2_s(i32 %slice, <vscale x 4 x i32> %za1, <vscale x 4 x i32> %za2) {
+; CHECK-LABEL: za_write_vg1x2_s:
+; CHECK:       // %bb.0:
+; CHECK-NEXT:    // kill: def $z1 killed $z1 killed $z0_z1 def $z0_z1
+; CHECK-NEXT:    mov w8, w0
+; CHECK-NEXT:    // kill: def $z0 killed $z0 killed $z0_z1 def $z0_z1
+; CHECK-NEXT:    mov za.d[w8, 0, vgx2], { z0.d, z1.d }
+; CHECK-NEXT:    mov za.d[w8, 7, vgx2], { z0.d, z1.d }
+; CHECK-NEXT:    ret
+  call void @llvm.aarch64.sme.write.vg1x2.nxv4i32(i32 %slice, <vscale x 4 x i32> %za1, <vscale x 4 x i32> %za2)
+  %slice.7 = add i32 %slice, 7
+  call void @llvm.aarch64.sme.write.vg1x2.nxv4i32(i32 %slice.7, <vscale x 4 x i32> %za1, <vscale x 4 x i32> %za2)
+  ret void
+}
+
+define void @za_write_vg1x2_f32(i32 %slice, <vscale x 4 x float> %za1, <vscale x 4 x float> %za2) {
+; CHECK-LABEL: za_write_vg1x2_f32:
+; CHECK:       // %bb.0:
+; CHECK-NEXT:    // kill: def $z1 killed $z1 killed $z0_z1 def $z0_z1
+; CHECK-NEXT:    mov w8, w0
+; CHECK-NEXT:    // kill: def $z0 killed $z0 killed $z0_z1 def $z0_z1
+; CHECK-NEXT:    mov za.d[w8, 0, vgx2], { z0.d, z1.d }
+; CHECK-NEXT:    mov za.d[w8, 7, vgx2], { z0.d, z1.d }
+; CHECK-NEXT:    ret
+  call void @llvm.aarch64.sme.write.vg1x2.nxv4f32(i32 %slice, <vscale x 4 x float> %za1, <vscale x 4 x float> %za2)
+  %slice.7 = add i32 %slice, 7
+  call void @llvm.aarch64.sme.write.vg1x2.nxv4f32(i32 %slice.7, <vscale x 4 x float> %za1, <vscale x 4 x float> %za2)
+  ret void
+}
+
 define void @za_write_vg1x2_d(i32 %slice, <vscale x 2 x i64> %za1, <vscale x 2 x i64> %za2) {
 ; CHECK-LABEL: za_write_vg1x2_d:
 ; CHECK:       // %bb.0:
@@ -531,6 +621,108 @@ define void @za_write_vg1x2_f64(i32 %slice, <vscale x 2 x double> %za1, <vscale
 ; Move Multi-Vector To ZA (Write) x4
 ;
 
+define void @za_write_vg1x4_b(i32 %slice, <vscale x 16 x i8> %za1, <vscale x 16 x i8> %za2, <vscale x 16 x i8> %za3, <vscale x 16 x i8> %za4) {
+; CHECK-LABEL: za_write_vg1x4_b:
+; CHECK:       // %bb.0:
+; CHECK-NEXT:    // kill: def $z3 killed $z3 killed $z0_z1_z2_z3 def $z0_z1_z2_z3
+; CHECK-NEXT:    mov w8, w0
+; CHECK-NEXT:    // kill: def $z2 killed $z2 killed $z0_z1_z2_z3 def $z0_z1_z2_z3
+; CHECK-NEXT:    // kill: def $z1 killed $z1 killed $z0_z1_z2_z3 def $z0_z1_z2_z3
+; CHECK-NEXT:    // kill: def $z0 killed $z0 killed $z0_z1_z2_z3 def $z0_z1_z2_z3
+; CHECK-NEXT:    mov za.d[w8, 0, vgx4], { z0.d - z3.d }
+; CHECK-NEXT:    mov za.d[w8, 7, vgx4], { z0.d - z3.d }
+; CHECK-NEXT:    ret
+  call void @llvm.aarch64.sme.write.vg1x4.nxv16i8(i32 %slice, <vscale x 16 x i8> %za1, <vscale x 16 x i8> %za2, <vscale x 16 x i8> %za3, <vscale x 16 x i8> %za4)
+  %slice.7 = add i32 %slice, 7
+  call void @llvm.aarch64.sme.write.vg1x4.nxv16i8(i32 %slice.7, <vscale x 16 x i8> %za1, <vscale x 16 x i8> %za2, <vscale x 16 x i8> %za3, <vscale x 16 x i8> %za4)
+  ret void
+}
+
+define void @za_write_vg1x4_h(i32 %slice, <vscale x 8 x i16> %za1, <vscale x 8 x i16> %za2, <vscale x 8 x i16> %za3, <vscale x 8 x i16> %za4) {
+; CHECK-LABEL: za_write_vg1x4_h:
+; CHECK:       // %bb.0:
+; CHECK-NEXT:    // kill: def $z3 killed $z3 killed $z0_z1_z2_z3 def $z0_z1_z2_z3
+; CHECK-NEXT:    mov w8, w0
+; CHECK-NEXT:    // kill: def $z2 killed $z2 killed $z0_z1_z2_z3 def $z0_z1_z2_z3
+; CHECK-NEXT:    // kill: def $z1 killed $z1 killed $z0_z1_z2_z3 def $z0_z1_z2_z3
+; CHECK-NEXT:    // kill: def $z0 killed $z0 killed $z0_z1_z2_z3 def $z0_z1_z2_z3
+; CHECK-NEXT:    mov za.d[w8, 0, vgx4], { z0.d - z3.d }
+; CHECK-NEXT:    mov za.d[w8, 7, vgx4], { z0.d - z3.d }
+; CHECK-NEXT:    ret
+  call void @llvm.aarch64.sme.write.vg1x4.nxv8i16(i32 %slice, <vscale x 8 x i16> %za1, <vscale x 8 x i16> %za2, <vscale x 8 x i16> %za3, <vscale x 8 x i16> %za4)
+  %slice.7 = add i32 %slice, 7
+  call void @llvm.aarch64.sme.write.vg1x4.nxv8i16(i32 %slice.7, <vscale x 8 x i16> %za1, <vscale x 8 x i16> %za2, <vscale x 8 x i16> %za3, <vscale x 8 x i16> %za4)
+  ret void
+}
+
+define void @za_write_vg1x4_f16(i32 %slice, <vscale x 8 x half> %za1, <vscale x 8 x half> %za2, <vscale x 8 x half> %za3, <vscale x 8 x half> %za4) {
+; CHECK-LABEL: za_write_vg1x4_f16:
+; CHECK:       // %bb.0:
+; CHECK-NEXT:    // kill: def $z3 killed $z3 killed $z0_z1_z2_z3 def $z0_z1_z2_z3
+; CHECK-NEXT:    mov w8, w0
+; CHECK-NEXT:    // kill: def $z2 killed $z2 killed $z0_z1_z2_z3 def $z0_z1_z2_z3
+; CHECK-NEXT:    // kill: def $z1 killed $z1 killed $z0_z1_z2_z3 def $z0_z1_z2_z3
+; CHECK-NEXT:    // kill: def $z0 killed $z0 killed $z0_z1_z2_z3 def $z0_z1_z2_z3
+; CHECK-NEXT:    mov za.d[w8, 0, vgx4], { z0.d - z3.d }
+; CHECK-NEXT:    mov za.d[w8, 7, vgx4], { z0.d - z3.d }
+; CHECK-NEXT:    ret
+  call void @llvm.aarch64.sme.write.vg1x4.nxv8f16(i32 %slice, <vscale x 8 x half> %za1, <vscale x 8 x half> %za2, <vscale x 8 x half> %za3, <vscale x 8 x half> %za4)
+  %slice.7 = add i32 %slice, 7
+  call void @llvm.aarch64.sme.write.vg1x4.nxv8f16(i32 %slice.7, <vscale x 8 x half> %za1, <vscale x 8 x half> %za2, <vscale x 8 x half> %za3, <vscale x 8 x half> %za4)
+  ret void
+}
+
+define void @za_write_vg1x4_bf16(i32 %slice, <vscale x 8 x bfloat> %za1, <vscale x 8 x bfloat> %za2, <vscale x 8 x bfloat> %za3, <vscale x 8 x bfloat> %za4) {
+; CHECK-LABEL: za_write_vg1x4_bf16:
+; CHECK:       // %bb.0:
+; CHECK-NEXT:    // kill: def $z3 killed $z3 killed $z0_z1_z2_z3 def $z0_z1_z2_z3
+; CHECK-NEXT:    mov w8, w0
+; CHECK-NEXT:    // kill: def $z2 killed $z2 killed $z0_z1_z2_z3 def $z0_z1_z2_z3
+; CHECK-NEXT:    // kill: def $z1 killed $z1 killed $z0_z1_z2_z3 def $z0_z1_z2_z3
+; CHECK-NEXT:    // kill: def $z0 killed $z0 killed $z0_z1_z2_z3 def $z0_z1_z2_z3
+; CHECK-NEXT:    mov za.d[w8, 0, vgx4], { z0.d - z3.d }
+; CHECK-NEXT:    mov za.d[w8, 7, vgx4], { z0.d - z3.d }
+; CHECK-NEXT:    ret
+  call void @llvm.aarch64.sme.write.vg1x4.nxv8bf16(i32 %slice, <vscale x 8 x bfloat> %za1, <vscale x 8 x bfloat> %za2, <vscale x 8 x bfloat> %za3, <vscale x 8 x bfloat> %za4)
+  %slice.7 = add i32 %slice, 7
+  call void @llvm.aarch64.sme.write.vg1x4.nxv8bf16(i32 %slice.7, <vscale x 8 x bfloat> %za1, <vscale x 8 x bfloat> %za2, <vscale x 8 x bfloat> %za3, <vscale x 8 x bfloat> %za4)
+  ret void
+}
+
+define void @za_write_vg1x4_s(i32 %slice, <vscale x 4 x i32> %za1, <vscale x 4 x i32> %za2, <vscale x 4 x i32> %za3, <vscale x 4 x i32> %za4) {
+; CHECK-LABEL: za_write_vg1x4_s:
+; CHECK:       // %bb.0:
+; CHECK-NEXT:    // kill: def $z3 killed $z3 killed $z0_z1_z2_z3 def $z0_z1_z2_z3
+; CHECK-NEXT:    mov w8, w0
+; CHECK-NEXT:    // kill: def $z2 killed $z2 killed $z0_z1_z2_z3 def $z0_z1_z2_z3
+; CHECK-NEXT:    // kill: def $z1 killed $z1 killed $z0_z1_z2_z3 def $z0_z1_z2_z3
+; CHECK-NEXT:    // kill: def $z0 killed $z0 killed $z0_z1_z2_z3 def $z0_z1_z2_z3
+; CHECK-NEXT:    mov za.d[w8, 0, vgx4], { z0.d - z3.d }
+; CHECK-NEXT:    mov za.d[w8, 7, vgx4], { z0.d - z3.d }
+; CHECK-NEXT:    ret
+  call void @llvm.aarch64.sme.write.vg1x4.nxv4i32(i32 %slice, <vscale x 4 x i32> %za1, <vscale x 4 x i32> %za2, <vscale x 4 x i32> %za3, <vscale x 4 x i32> %za4)
+  %slice.7 = add i32 %slice, 7
+  call void @llvm.aarch64.sme.write.vg1x4.nxv4i32(i32 %slice.7, <vscale x 4 x i32> %za1, <vscale x 4 x i32> %za2, <vscale x 4 x i32> %za3, <vscale x 4 x i32> %za4)
+  ret void
+}
+
+define void @za_write_vg1x4_f32(i32 %slice, <vscale x 4 x float> %za1, <vscale x 4 x float> %za2, <vscale x 4 x float> %za3, <vscale x 4 x float> %za4) {
+; CHECK-LABEL: za_write_vg1x4_f32:
+; CHECK:       // %bb.0:
+; CHECK-NEXT:    // kill: def $z3 killed $z3 killed $z0_z1_z2_z3 def $z0_z1_z2_z3
+; CHECK-NEXT:    mov w8, w0
+; CHECK-NEXT:    // kill: def $z2 killed $z2 killed $z0_z1_z2_z3 def $z0_z1_z2_z3
+; CHECK-NEXT:    // kill: def $z1 killed $z1 killed $z0_z1_z2_z3 def $z0_z1_z2_z3
+; CHECK-NEXT:    // kill: def $z0 killed $z0 killed $z0_z1_z2_z3 def $z0_z1_z2_z3
+; CHECK-NEXT:    mov za.d[w8, 0, vgx4], { z0.d - z3.d }
+; CHECK-NEXT:    mov za.d[w8, 7, vgx4], { z0.d - z3.d }
+; CHECK-NEXT:    ret
+  call void @llvm.aarch64.sme.write.vg1x4.nxv4f32(i32 %slice, <vscale x 4 x float> %za1, <vscale x 4 x float> %za2, <vscale x 4 x float> %za3, <vscale x 4 x float> %za4)
+  %slice.7 = add i32 %slice, 7
+  call void @llvm.aarch64.sme.write.vg1x4.nxv4f32(i32 %slice.7, <vscale x 4 x float> %za1, <vscale x 4 x float> %za2, <vscale x 4 x float> %za3, <vscale x 4 x float> %za4)
+  ret void
+}
+
 define void @za_write_vg1x4_d(i32 %slice, <vscale x 2 x i64> %za1, <vscale x 2 x i64> %za2, <vscale x 2 x i64> %za3, <vscale x 2 x i64> %za4) {
 ; CHECK-LABEL: za_write_vg1x4_d:
 ; CHECK:       // %bb.0:
@@ -601,8 +793,20 @@ declare void @llvm.aarch64.sme.write.ver.vg4.nxv4f32(i32, i32, <vscale x 4 x flo
 declare void @llvm.aarch64.sme.write.ver.vg4.nxv2i64(i32, i32, <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64>)
 declare void @llvm.aarch64.sme.write.ver.vg4.nxv2f64(i32, i32, <vscale x 2 x double>, <vscale x 2 x double>, <vscale x 2 x double>, <vscale x 2 x double>)
 
+declare void @llvm.aarch64.sme.write.vg1x2.nxv16i8(i32, <vscale x 16 x i8>, <vscale x 16 x i8>)
+declare void @llvm.aarch64.sme.write.vg1x2.nxv8i16(i32, <vscale x 8 x i16>, <vscale x 8 x i16>)
+declare void @llvm.aarch64.sme.write.vg1x2.nxv4i32(i32, <vscale x 4 x i32>, <vscale x 4 x i32>)
 declare void @llvm.aarch64.sme.write.vg1x2.nxv2i64(i32, <vscale x 2 x i64>, <vscale x 2 x i64>)
+declare void @llvm.aarch64.sme.write.vg1x2.nxv8f16(i32, <vscale x 8 x half>, <vscale x 8 x half>)
+declare void @llvm.aarch64.sme.write.vg1x2.nxv8bf16(i32, <vscale x 8 x bfloat>, <vscale x 8 x bfloat>)
+declare void @llvm.aarch64.sme.write.vg1x2.nxv4f32(i32, <vscale x 4 x float>, <vscale x 4 x float>)
 declare void @llvm.aarch64.sme.write.vg1x2.nxv2f64(i32, <vscale x 2 x double>, <vscale x 2 x double>)
 
+declare void @llvm.aarch64.sme.write.vg1x4.nxv16i8(i32, <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8>, <vscale x 16 x i8>)
+declare void @llvm.aarch64.sme.write.vg1x4.nxv8i16(i32, <vscale x 8 x i16>, <vscale x 8 x i16>, <vscale x 8 x i16>, <vscale x 8 x i16>)
+declare void @llvm.aarch64.sme.write.vg1x4.nxv4i32(i32, <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32>, <vscale x 4 x i32>)
 declare void @llvm.aarch64.sme.write.vg1x4.nxv2i64(i32, <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64>, <vscale x 2 x i64>)
+declare void @llvm.aarch64.sme.write.vg1x4.nxv8f16(i32, <vscale x 8 x half>, <vscale x 8 x half>, <vscale x 8 x half>, <vscale x 8 x half>)
+declare void @llvm.aarch64.sme.write.vg1x4.nxv8bf16(i32, <vscale x 8 x bfloat>, <vscale x 8 x bfloat>, <vscale x 8 x bfloat>, <vscale x 8 x bfloat>)
+declare void @llvm.aarch64.sme.write.vg1x4.nxv4f32(i32, <vscale x 4 x float>, <vscale x 4 x float>, <vscale x 4 x float>, <vscale x 4 x float>)
 declare void @llvm.aarch64.sme.write.vg1x4.nxv2f64(i32, <vscale x 2 x double>, <vscale x 2 x double>, <vscale x 2 x double>, <vscale x 2 x double>)

From dfe59ed4ce77fbc563feba1bda3b05e90ae83cdb Mon Sep 17 00:00:00 2001
From: Kerry McLaughlin <kerry.mclaughlin@arm.com>
Date: Tue, 5 Dec 2023 16:34:04 +0000
Subject: [PATCH 4/5] - Remove unused SVE_ACLE_FUNC & SVE_OVERLOADED_FORMS run
 lines from acle_sme2_read.c

---
 .../CodeGen/aarch64-sme2-intrinsics/acle_sme2_read.c     | 9 ---------
 1 file changed, 9 deletions(-)

diff --git a/clang/test/CodeGen/aarch64-sme2-intrinsics/acle_sme2_read.c b/clang/test/CodeGen/aarch64-sme2-intrinsics/acle_sme2_read.c
index 63abc9acfdf9c..583a7fc815472 100644
--- a/clang/test/CodeGen/aarch64-sme2-intrinsics/acle_sme2_read.c
+++ b/clang/test/CodeGen/aarch64-sme2-intrinsics/acle_sme2_read.c
@@ -3,19 +3,10 @@
 
 // RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sme2 -target-feature +sve -S -disable-O0-optnone -Werror -Wall -emit-llvm -o - %s | opt -S -p mem2reg,instcombine,tailcallelim | FileCheck %s
 // RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sme2 -target-feature +sve -S -disable-O0-optnone -Werror -Wall -emit-llvm -o - -x c++ %s | opt -S -p mem2reg,instcombine,tailcallelim | FileCheck %s -check-prefix=CPP-CHECK
-// RUN: %clang_cc1 -DSVE_OVERLOADED_FORMS -triple aarch64-none-linux-gnu -target-feature +sme2 -target-feature +sve -S -disable-O0-optnone -Werror -Wall -emit-llvm -o - %s | opt -S -p mem2reg,instcombine,tailcallelim | FileCheck %s
-// RUN: %clang_cc1 -DSVE_OVERLOADED_FORMS -triple aarch64-none-linux-gnu -target-feature +sme2 -target-feature +sve -S -disable-O0-optnone -Werror -Wall -emit-llvm -o - -x c++ %s | opt -S -p mem2reg,instcombine,tailcallelim | FileCheck %s -check-prefix=CPP-CHECK
 // RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sme2 -target-feature +sve -S -disable-O0-optnone -Werror -Wall -o /dev/null %s
 
 #include <arm_sme_draft_spec_subject_to_change.h>
 
-#ifdef SVE_OVERLOADED_FORMS
-// A simple used,unused... macro, long enough to represent any SVE builtin.
-#define SVE_ACLE_FUNC(A1,A2_UNUSED,A3,A4_UNUSED) A1##A3
-#else
-#define SVE_ACLE_FUNC(A1,A2,A3,A4) A1##A2##A3##A4
-#endif
-
 // CHECK-LABEL: @test_svread_ver_za8_u8_vg2(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = tail call { <vscale x 16 x i8>, <vscale x 16 x i8> } @llvm.aarch64.sme.read.ver.vg2.nxv16i8(i32 0, i32 [[BASE:%.*]])

From b4d97a01bdfd6fa8a8a6d8180c0d6389d5538869 Mon Sep 17 00:00:00 2001
From: Kerry McLaughlin <kerry.mclaughlin@arm.com>
Date: Fri, 15 Dec 2023 13:53:09 +0000
Subject: [PATCH 5/5] - Moved more of the builtin name into the read & write
 multiclasses

---
 clang/include/clang/Basic/arm_sme.td | 40 ++++++++++++++--------------
 1 file changed, 20 insertions(+), 20 deletions(-)

diff --git a/clang/include/clang/Basic/arm_sme.td b/clang/include/clang/Basic/arm_sme.td
index a46e2bdb34c6f..ecaa59a898829 100644
--- a/clang/include/clang/Basic/arm_sme.td
+++ b/clang/include/clang/Basic/arm_sme.td
@@ -306,35 +306,35 @@ defm SVSUB : ZAAddSub<"sub">;
 //
 
 multiclass ZAWrite_VG<string n, string t, string i, list<ImmCheck> checks> {
-  def NAME # _VG2_H : Inst<"svwrite_hor_" # n # "_vg2",   "vim2", t, MergeNone, i # "_hor_vg2", [IsSharedZA, IsStreaming], checks>;
-  def NAME # _VG2_V : Inst<"svwrite_ver_" # n # "_vg2",   "vim2", t, MergeNone, i # "_ver_vg2", [IsSharedZA, IsStreaming], checks>;
-  def NAME # _VG4_H : Inst<"svwrite_hor_" # n # "_vg4",   "vim4", t, MergeNone, i # "_hor_vg4", [IsSharedZA, IsStreaming], checks>;
-  def NAME # _VG4_V : Inst<"svwrite_ver_" # n # "_vg4",   "vim4", t, MergeNone, i # "_ver_vg4", [IsSharedZA, IsStreaming], checks>;
-  def NAME # _VG1x2 : Inst<"svwrite_"     # n # "_vg1x2", "vm2",  t, MergeNone, i # "_vg1x2",   [IsSharedZA, IsStreaming], []>;
-  def NAME # _VG1x4 : Inst<"svwrite_"     # n # "_vg1x4", "vm4",  t, MergeNone, i # "_vg1x4",   [IsSharedZA, IsStreaming], []>;
+  def NAME # _VG2_H : Inst<"svwrite_hor_" # n # "[_{d}]_vg2",   "vim2", t, MergeNone, i # "_hor_vg2", [IsSharedZA, IsStreaming], checks>;
+  def NAME # _VG2_V : Inst<"svwrite_ver_" # n # "[_{d}]_vg2",   "vim2", t, MergeNone, i # "_ver_vg2", [IsSharedZA, IsStreaming], checks>;
+  def NAME # _VG4_H : Inst<"svwrite_hor_" # n # "[_{d}]_vg4",   "vim4", t, MergeNone, i # "_hor_vg4", [IsSharedZA, IsStreaming], checks>;
+  def NAME # _VG4_V : Inst<"svwrite_ver_" # n # "[_{d}]_vg4",   "vim4", t, MergeNone, i # "_ver_vg4", [IsSharedZA, IsStreaming], checks>;
+  def NAME # _VG1x2 : Inst<"svwrite_"     # n # "[_{d}]_vg1x2", "vm2",  t, MergeNone, i # "_vg1x2",   [IsSharedZA, IsStreaming], []>;
+  def NAME # _VG1x4 : Inst<"svwrite_"     # n # "[_{d}]_vg1x4", "vm4",  t, MergeNone, i # "_vg1x4",   [IsSharedZA, IsStreaming], []>;
 }
 
 let TargetGuard = "sme2" in {
-  defm SVWRITE_ZA8  : ZAWrite_VG<"za8[_{d}]",  "cUc",   "aarch64_sme_write", [ImmCheck<0, ImmCheck0_0>]>;
-  defm SVWRITE_ZA16 : ZAWrite_VG<"za16[_{d}]", "sUshb", "aarch64_sme_write", [ImmCheck<0, ImmCheck0_1>]>;
-  defm SVWRITE_ZA32 : ZAWrite_VG<"za32[_{d}]", "iUif",  "aarch64_sme_write", [ImmCheck<0, ImmCheck0_3>]>;
-  defm SVWRITE_ZA64 : ZAWrite_VG<"za64[_{d}]", "lUld",  "aarch64_sme_write", [ImmCheck<0, ImmCheck0_7>]>;
+  defm SVWRITE_ZA8  : ZAWrite_VG<"za8",  "cUc",   "aarch64_sme_write", [ImmCheck<0, ImmCheck0_0>]>;
+  defm SVWRITE_ZA16 : ZAWrite_VG<"za16", "sUshb", "aarch64_sme_write", [ImmCheck<0, ImmCheck0_1>]>;
+  defm SVWRITE_ZA32 : ZAWrite_VG<"za32", "iUif",  "aarch64_sme_write", [ImmCheck<0, ImmCheck0_3>]>;
+  defm SVWRITE_ZA64 : ZAWrite_VG<"za64", "lUld",  "aarch64_sme_write", [ImmCheck<0, ImmCheck0_7>]>;
 }
 
 multiclass ZARead_VG<string n, string t, string i, list<ImmCheck> checks> {
-  def NAME # _VG2_H : Inst<"svread_hor_" # n # "_vg2",   "2im", t, MergeNone, i # "_hor_vg2", [IsSharedZA, IsPreservesZA, IsStreaming], checks>;
-  def NAME # _VG2_V : Inst<"svread_ver_" # n # "_vg2",   "2im", t, MergeNone, i # "_ver_vg2", [IsSharedZA, IsPreservesZA, IsStreaming], checks>;
-  def NAME # _VG4_H : Inst<"svread_hor_" # n # "_vg4",   "4im", t, MergeNone, i # "_hor_vg4", [IsSharedZA, IsPreservesZA, IsStreaming], checks>;
-  def NAME # _VG4_V : Inst<"svread_ver_" # n # "_vg4",   "4im", t, MergeNone, i # "_ver_vg4", [IsSharedZA, IsPreservesZA, IsStreaming], checks>;
-  def NAME # _VG1x2 : Inst<"svread_"     # n # "_vg1x2", "2m",  t, MergeNone, i # "_vg1x2",   [IsSharedZA, IsPreservesZA, IsStreaming], []>;
-  def NAME # _VG1x4 : Inst<"svread_"     # n # "_vg1x4", "4m",  t, MergeNone, i # "_vg1x4",   [IsSharedZA, IsPreservesZA, IsStreaming], []>;
+  def NAME # _VG2_H : Inst<"svread_hor_" # n # "_{d}_vg2",   "2im", t, MergeNone, i # "_hor_vg2", [IsSharedZA, IsPreservesZA, IsStreaming], checks>;
+  def NAME # _VG2_V : Inst<"svread_ver_" # n # "_{d}_vg2",   "2im", t, MergeNone, i # "_ver_vg2", [IsSharedZA, IsPreservesZA, IsStreaming], checks>;
+  def NAME # _VG4_H : Inst<"svread_hor_" # n # "_{d}_vg4",   "4im", t, MergeNone, i # "_hor_vg4", [IsSharedZA, IsPreservesZA, IsStreaming], checks>;
+  def NAME # _VG4_V : Inst<"svread_ver_" # n # "_{d}_vg4",   "4im", t, MergeNone, i # "_ver_vg4", [IsSharedZA, IsPreservesZA, IsStreaming], checks>;
+  def NAME # _VG1x2 : Inst<"svread_"     # n # "_{d}_vg1x2", "2m",  t, MergeNone, i # "_vg1x2",   [IsSharedZA, IsPreservesZA, IsStreaming], []>;
+  def NAME # _VG1x4 : Inst<"svread_"     # n # "_{d}_vg1x4", "4m",  t, MergeNone, i # "_vg1x4",   [IsSharedZA, IsPreservesZA, IsStreaming], []>;
 }
 
 let TargetGuard = "sme2" in {
-  defm SVREAD_ZA8  : ZARead_VG<"za8_{d}",  "cUc",   "aarch64_sme_read", [ImmCheck<0, ImmCheck0_0>]>;
-  defm SVREAD_ZA16 : ZARead_VG<"za16_{d}", "sUshb", "aarch64_sme_read", [ImmCheck<0, ImmCheck0_1>]>;
-  defm SVREAD_ZA32 : ZARead_VG<"za32_{d}", "iUif",  "aarch64_sme_read", [ImmCheck<0, ImmCheck0_3>]>;
-  defm SVREAD_ZA64 : ZARead_VG<"za64_{d}", "lUld",  "aarch64_sme_read", [ImmCheck<0, ImmCheck0_7>]>;
+  defm SVREAD_ZA8  : ZARead_VG<"za8",  "cUc",   "aarch64_sme_read", [ImmCheck<0, ImmCheck0_0>]>;
+  defm SVREAD_ZA16 : ZARead_VG<"za16", "sUshb", "aarch64_sme_read", [ImmCheck<0, ImmCheck0_1>]>;
+  defm SVREAD_ZA32 : ZARead_VG<"za32", "iUif",  "aarch64_sme_read", [ImmCheck<0, ImmCheck0_3>]>;
+  defm SVREAD_ZA64 : ZARead_VG<"za64", "lUld",  "aarch64_sme_read", [ImmCheck<0, ImmCheck0_7>]>;
 }
 
 //