Skip to content

[WebAssembly] Support BUILD_VECTOR with F16x8. #108117

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.

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

Merged
merged 1 commit into from
Sep 11, 2024
Merged
Show file tree
Hide file tree
Changes from all commits
Commits
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
15 changes: 15 additions & 0 deletions llvm/lib/Target/WebAssembly/WebAssemblyISelLowering.cpp
Original file line number Diff line number Diff line change
Expand Up @@ -208,6 +208,9 @@ WebAssemblyTargetLowering::WebAssemblyTargetLowering(
MVT::v2f64})
setOperationAction(ISD::BUILD_VECTOR, T, Custom);

if (Subtarget->hasFP16())
setOperationAction(ISD::BUILD_VECTOR, MVT::f16, Custom);

// We have custom shuffle lowering to expose the shuffle mask
for (auto T : {MVT::v16i8, MVT::v8i16, MVT::v4i32, MVT::v4f32, MVT::v2i64,
MVT::v2f64})
Expand Down Expand Up @@ -2055,6 +2058,18 @@ static SDValue LowerConvertLow(SDValue Op, SelectionDAG &DAG) {

SDValue WebAssemblyTargetLowering::LowerBUILD_VECTOR(SDValue Op,
SelectionDAG &DAG) const {
MVT VT = Op.getSimpleValueType();
if (VT == MVT::v8f16) {
// BUILD_VECTOR can't handle FP16 operands since Wasm doesn't have a scaler
// FP16 type, so cast them to I16s.
MVT IVT = VT.changeVectorElementType(MVT::i16);
SmallVector<SDValue, 8> NewOps;
for (unsigned I = 0, E = Op.getNumOperands(); I < E; ++I)
NewOps.push_back(DAG.getBitcast(MVT::i16, Op.getOperand(I)));
SDValue Res = DAG.getNode(ISD::BUILD_VECTOR, SDLoc(), IVT, NewOps);
return DAG.getBitcast(VT, Res);
}

if (auto ConvertLow = LowerConvertLow(Op, DAG))
return ConvertLow;

Expand Down
7 changes: 7 additions & 0 deletions llvm/test/CodeGen/WebAssembly/half-precision.ll
Original file line number Diff line number Diff line change
Expand Up @@ -28,6 +28,13 @@ define <8 x half> @splat_v8f16(float %x) {
ret <8 x half> %v
}

; CHECK-LABEL: const_splat_v8f16:
; CHECK: v128.const $push0=, 20800, 0, 0, 0, 0, 0, 0, 20800
; CHECK-NEXT: return $pop0
define <8 x half> @const_splat_v8f16() {
ret <8 x half> <half 42., half 0., half 0., half 0., half 0., half 0., half 0., half 42.>
}

; CHECK-LABEL: extract_lane_v8f16:
; CHECK: f16x8.extract_lane $push0=, $0, 1
; CHECK-NEXT: return $pop0
Expand Down
Loading