Skip to content

[AArch64] Implement intrinsics for SME FP8 FMLAL/FMLALL (single) #118869

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Closed
wants to merge 4 commits into from
Closed

[AArch64] Implement intrinsics for SME FP8 FMLAL/FMLALL (single) #118869

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
4 commits
Select commit Hold shift + click to select a range
a92b7a9
[AArch64] Implement intrinsics for SME FP8 FMOPA
SpencerAbson Nov 25, 2024
e804d5f
[AArch64] Implement intrinsics for SME FP8 FMLAL/FMLALL
SpencerAbson Dec 3, 2024
6559b69
[NFC] Revert refactoring
SpencerAbson Dec 5, 2024
3081959
[AArch64] Implement intrinsics for SME FP8 FMLAL/FMLALL (single)
SpencerAbson Dec 5, 2024
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
42 changes: 42 additions & 0 deletions clang/include/clang/Basic/arm_sme.td
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -824,4 +824,46 @@ let SMETargetGuard = "sme-lutv2" in {
def SVLUTI4_ZT_X4 : SInst<"svluti4_zt_{d}_x4", "4i2.u", "cUc", MergeNone, "aarch64_sme_luti4_zt_x4", [IsStreaming, IsInZT0], [ImmCheck<0, ImmCheck0_0>]>;
}

let SMETargetGuard = "sme-f8f32" in {
def SVMOPA_FP8_ZA32 : Inst<"svmopa_za32[_mf8]_m_fpm", "viPPdd>", "m", MergeNone, "aarch64_sme_fp8_fmopa_za32",
[IsStreaming, IsInOutZA, SetsFPMR, IsOverloadNone], [ImmCheck<0, ImmCheck0_3>]>;

// FMLALL (indexed)
def SVMLA_FP8_LANE_ZA32_VG4x1 : Inst<"svmla_lane_za32[_mf8]_vg4x1_fpm", "vmddi>", "m", MergeNone, "aarch64_sme_fp8_fmlall_lane_za32_vg4x1",
[IsStreaming, IsInOutZA, SetsFPMR, IsOverloadNone], [ImmCheck<3, ImmCheck0_15>]>;
def SVMLA_FP8_LANE_ZA32_VG4x2 : Inst<"svmla_lane_za32[_mf8]_vg4x2_fpm", "vm2di>", "m", MergeNone, "aarch64_sme_fp8_fmlall_lane_za32_vg4x2",
[IsStreaming, IsInOutZA, SetsFPMR, IsOverloadNone], [ImmCheck<3, ImmCheck0_15>]>;
def SVMLA_FP8_LANE_ZA16_VG4x4 : Inst<"svmla_lane_za32[_mf8]_vg4x4_fpm", "vm4di>", "m", MergeNone, "aarch64_sme_fp8_fmlall_lane_za32_vg4x4",
[IsStreaming, IsInOutZA, SetsFPMR, IsOverloadNone], [ImmCheck<3, ImmCheck0_15>]>;

// FMLAL (mutliple and single)
def SVMLA_FP8_SINGLE_ZA32_VG4x1 : Inst<"svmla[_single]_za32[_mf8]_vg4x1_fpm", "vmdd>", "m", MergeNone, "aarch64_sme_fp8_fmlall_single_za32_vg4x1",
[IsStreaming, IsInOutZA, SetsFPMR, IsOverloadNone], []>;
def SVMLA_FP8_SINGLE_ZA32_VG4x2 : Inst<"svmla[_single]_za32[_mf8]_vg4x2_fpm", "vm2d>", "m", MergeNone, "aarch64_sme_fp8_fmlall_single_za32_vg4x2",
[IsStreaming, IsInOutZA, SetsFPMR, IsOverloadNone], []>;
def SVMLA_FP8_SINGLE_ZA32_VG4x4 : Inst<"svmla[_single]_za32[_mf8]_vg4x4_fpm", "vm4d>", "m", MergeNone, "aarch64_sme_fp8_fmlall_single_za32_vg4x4",
[IsStreaming, IsInOutZA, SetsFPMR, IsOverloadNone], []>;
}

let SMETargetGuard = "sme-f8f16" in {
def SVMOPA_FP8_ZA16 : Inst<"svmopa_za16[_mf8]_m_fpm", "viPPdd>", "m", MergeNone, "aarch64_sme_fp8_fmopa_za16",
[IsStreaming, IsInOutZA, SetsFPMR, IsOverloadNone], [ImmCheck<0, ImmCheck0_1>]>;

// FMLAL (indexed)
def SVMLA_FP8_LANE_ZA16_VG2x1 : Inst<"svmla_lane_za16[_mf8]_vg2x1_fpm", "vmddi>", "m", MergeNone, "aarch64_sme_fp8_fmlal_lane_za16_vg2x1",
[IsStreaming, IsInOutZA, SetsFPMR, IsOverloadNone], [ImmCheck<3, ImmCheck0_15>]>;
def SVMLA_FP8_LANE_ZA16_VG2x2 : Inst<"svmla_lane_za16[_mf8]_vg2x2_fpm", "vm2di>", "m", MergeNone, "aarch64_sme_fp8_fmlal_lane_za16_vg2x2",
[IsStreaming, IsInOutZA, SetsFPMR, IsOverloadNone], [ImmCheck<3, ImmCheck0_15>]>;
def SVMLA_FP8_LANE_ZA16_VG2x4 : Inst<"svmla_lane_za16[_mf8]_vg2x4_fpm", "vm4di>", "m", MergeNone, "aarch64_sme_fp8_fmlal_lane_za16_vg2x4",
[IsStreaming, IsInOutZA, SetsFPMR, IsOverloadNone], [ImmCheck<3, ImmCheck0_15>]>;

// FMLAL (mutliple and single)
def SVMLA_FP8_SINGLE_ZA16_VG2x1 : Inst<"svmla[_single]_za16[_mf8]_vg2x1_fpm", "vmdd>", "m", MergeNone, "aarch64_sme_fp8_fmlal_single_za16_vg2x1",
[IsStreaming, IsInOutZA, SetsFPMR, IsOverloadNone], []>;
def SVMLA_FP8_SINGLE_ZA16_VG2x2 : Inst<"svmla[_single]_za16[_mf8]_vg2x2_fpm", "vm2d>", "m", MergeNone, "aarch64_sme_fp8_fmlal_single_za16_vg2x2",
[IsStreaming, IsInOutZA, SetsFPMR, IsOverloadNone], []>;
def SVMLA_FP8_SINGLE_ZA16_VG2x4 : Inst<"svmla[_single]_za16[_mf8]_vg2x4_fpm", "vm4d>", "m", MergeNone, "aarch64_sme_fp8_fmlal_single_za16_vg2x4",
[IsStreaming, IsInOutZA, SetsFPMR, IsOverloadNone], []>;
}

} // let SVETargetGuard = InvalidMode
6 changes: 6 additions & 0 deletions clang/lib/CodeGen/CGBuiltin.cpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -10183,6 +10183,8 @@ CodeGenFunction::getSVEType(const SVETypeFlags &TypeFlags) {
case SVETypeFlags::EltTyInt64:
return llvm::ScalableVectorType::get(Builder.getInt64Ty(), 2);

case SVETypeFlags::EltTyMFloat8:
return llvm::ScalableVectorType::get(Builder.getInt8Ty(), 16);
case SVETypeFlags::EltTyFloat16:
return llvm::ScalableVectorType::get(Builder.getHalfTy(), 8);
case SVETypeFlags::EltTyBFloat16:
Expand Down Expand Up @@ -11234,6 +11236,10 @@ Value *CodeGenFunction::EmitAArch64SMEBuiltinExpr(unsigned BuiltinID,
BuiltinID == SME::BI__builtin_sme_svstr_za)
return EmitSMELdrStr(TypeFlags, Ops, Builtin->LLVMIntrinsic);

// Emit set FPMR for intrinsics that require it
if (TypeFlags.setsFPMR())
Builder.CreateCall(CGM.getIntrinsic(Intrinsic::aarch64_set_fpmr),
Ops.pop_back_val());
// Handle builtins which require their multi-vector operands to be swapped
swapCommutativeSMEOperands(BuiltinID, Ops);

Expand Down
55 changes: 55 additions & 0 deletions clang/test/CodeGen/AArch64/fp8-intrinsics/acle_sme2_fp8_fmopa.c
View file
Open in desktop
Original file line number Diff line number Diff line change
@@ -0,0 +1,55 @@
// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5
// REQUIRES: aarch64-registered-target

// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sme -target-feature +sme-f8f16 -target-feature +sme-f8f32 -disable-O0-optnone -Werror -Wall -emit-llvm -o - %s | opt -S -passes=mem2reg,tailcallelim | FileCheck %s
// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sme -target-feature +sme-f8f16 -target-feature +sme-f8f32 -disable-O0-optnone -Werror -Wall -emit-llvm -o - -x c++ %s | opt -S -passes=mem2reg,tailcallelim | FileCheck %s -check-prefix=CPP-CHECK
// RUN: %clang_cc1 -DSVE_OVERLOADED_FORMS -triple aarch64-none-linux-gnu -target-feature +sme -target-feature +sme-f8f16 -target-feature +sme-f8f32 -disable-O0-optnone -Werror -Wall -emit-llvm -o - %s | opt -S -passes=mem2reg,tailcallelim | FileCheck %s
// RUN: %clang_cc1 -DSVE_OVERLOADED_FORMS -triple aarch64-none-linux-gnu -target-feature +sme -target-feature +sme-f8f16 -target-feature +sme-f8f32 -disable-O0-optnone -Werror -Wall -emit-llvm -o - -x c++ %s | opt -S -passes=mem2reg,tailcallelim | FileCheck %s -check-prefix=CPP-CHECK
// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sme -target-feature +sme-f8f16 -target-feature +sme-f8f32 -S -disable-O0-optnone -Werror -Wall -o /dev/null %s

#include <arm_sme.h>

#ifdef SVE_OVERLOADED_FORMS
#define SVE_ACLE_FUNC(A1,A2_UNUSED,A3) A1##A3
#else
#define SVE_ACLE_FUNC(A1,A2,A3) A1##A2##A3
#endif


// CHECK-LABEL: define dso_local void @test_svmopa_za16_mf8_m(
// CHECK-SAME: <vscale x 16 x i1> [[PN:%.*]], <vscale x 16 x i1> [[PM:%.*]], <vscale x 16 x i8> [[ZN:%.*]], <vscale x 16 x i8> [[ZM:%.*]], i64 noundef [[FPMR:%.*]]) #[[ATTR0:[0-9]+]] {
// CHECK-NEXT: [[ENTRY:.*:]]
// CHECK-NEXT: tail call void @llvm.aarch64.set.fpmr(i64 [[FPMR]])
// CHECK-NEXT: tail call void @llvm.aarch64.sme.fp8.fmopa.za16(i32 1, <vscale x 16 x i1> [[PN]], <vscale x 16 x i1> [[PM]], <vscale x 16 x i8> [[ZN]], <vscale x 16 x i8> [[ZM]])
// CHECK-NEXT: ret void
//
// CPP-CHECK-LABEL: define dso_local void @_Z22test_svmopa_za16_mf8_mu10__SVBool_tS_u13__SVMfloat8_tS0_m(
// CPP-CHECK-SAME: <vscale x 16 x i1> [[PN:%.*]], <vscale x 16 x i1> [[PM:%.*]], <vscale x 16 x i8> [[ZN:%.*]], <vscale x 16 x i8> [[ZM:%.*]], i64 noundef [[FPMR:%.*]]) #[[ATTR0:[0-9]+]] {
// CPP-CHECK-NEXT: [[ENTRY:.*:]]
// CPP-CHECK-NEXT: tail call void @llvm.aarch64.set.fpmr(i64 [[FPMR]])
// CPP-CHECK-NEXT: tail call void @llvm.aarch64.sme.fp8.fmopa.za16(i32 1, <vscale x 16 x i1> [[PN]], <vscale x 16 x i1> [[PM]], <vscale x 16 x i8> [[ZN]], <vscale x 16 x i8> [[ZM]])
// CPP-CHECK-NEXT: ret void
//
void test_svmopa_za16_mf8_m(svbool_t pn, svbool_t pm, svmfloat8_t zn,
svmfloat8_t zm, fpm_t fpmr) __arm_streaming __arm_inout("za") {
SVE_ACLE_FUNC(svmopa_za16,_mf8,_m_fpm)(1, pn, pm, zn, zm, fpmr);
}

// CHECK-LABEL: define dso_local void @test_svmopa_za32_mf8_m(
// CHECK-SAME: <vscale x 16 x i1> [[PN:%.*]], <vscale x 16 x i1> [[PM:%.*]], <vscale x 16 x i8> [[ZN:%.*]], <vscale x 16 x i8> [[ZM:%.*]], i64 noundef [[FPMR:%.*]]) #[[ATTR0]] {
// CHECK-NEXT: [[ENTRY:.*:]]
// CHECK-NEXT: tail call void @llvm.aarch64.set.fpmr(i64 [[FPMR]])
// CHECK-NEXT: tail call void @llvm.aarch64.sme.fp8.fmopa.za32(i32 3, <vscale x 16 x i1> [[PN]], <vscale x 16 x i1> [[PM]], <vscale x 16 x i8> [[ZN]], <vscale x 16 x i8> [[ZM]])
// CHECK-NEXT: ret void
//
// CPP-CHECK-LABEL: define dso_local void @_Z22test_svmopa_za32_mf8_mu10__SVBool_tS_u13__SVMfloat8_tS0_m(
// CPP-CHECK-SAME: <vscale x 16 x i1> [[PN:%.*]], <vscale x 16 x i1> [[PM:%.*]], <vscale x 16 x i8> [[ZN:%.*]], <vscale x 16 x i8> [[ZM:%.*]], i64 noundef [[FPMR:%.*]]) #[[ATTR0]] {
// CPP-CHECK-NEXT: [[ENTRY:.*:]]
// CPP-CHECK-NEXT: tail call void @llvm.aarch64.set.fpmr(i64 [[FPMR]])
// CPP-CHECK-NEXT: tail call void @llvm.aarch64.sme.fp8.fmopa.za32(i32 3, <vscale x 16 x i1> [[PN]], <vscale x 16 x i1> [[PM]], <vscale x 16 x i8> [[ZN]], <vscale x 16 x i8> [[ZM]])
// CPP-CHECK-NEXT: ret void
//
void test_svmopa_za32_mf8_m(svbool_t pn, svbool_t pm, svmfloat8_t zn,
svmfloat8_t zm, fpm_t fpmr) __arm_streaming __arm_inout("za") {
SVE_ACLE_FUNC(svmopa_za32,_mf8,_m_fpm)(3, pn, pm, zn, zm, fpmr);
}
Loading
Loading