Skip to content

fix(vpto): use typed vstus LLVM callees (#1124) #1136

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

Merged
mouliangyu merged 1 commit into from
Aug 6, 2026
+63 −7
Merged

fix(vpto): use typed vstus LLVM callees (#1124) #1136

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
1 commit
Select commit
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: 12 additions & 3 deletions lib/PTO/Transforms/VPTOCANN900LLVMEmitter.cpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -2410,8 +2410,16 @@ static FailureOr<StringRef> buildPstuCallee(MLIRContext *context, pto::PstuOp op
return failure();
}

static StringRef buildVstusCallee(MLIRContext *context) {
return StringAttr::get(context, "llvm.hivm.vstus").getValue();
static FailureOr<StringRef> buildVstusCallee(MLIRContext *context,
Type valueType) {
std::string vec =
getMemoryElementTypeFragment(getElementTypeFromVectorLike(valueType));
auto lanes = getElementCountFromVectorLike(valueType);
if (vec.empty() || !lanes)
return failure();
return StringAttr::get(context, "llvm.hivm.vstus.v" +
std::to_string(*lanes) + vec)
.getValue();
}

static FailureOr<StringRef> buildVstusPostCallee(MLIRContext *context,
Expand Down Expand Up @@ -7512,7 +7520,8 @@ class LowerVstusOpPattern final : public OpConversionPattern<pto::VstusOp> {
"unexpected converted vstus operand/result types");
}

FailureOr<StringRef> calleeName = buildVstusCallee(op.getContext());
FailureOr<StringRef> calleeName =
buildVstusCallee(op.getContext(), op.getValue().getType());
if (usePostIntrinsic)
calleeName =
buildVstusPostCallee(op.getContext(), op.getValue().getType());
Expand Down
15 changes: 12 additions & 3 deletions lib/PTO/Transforms/VPTOLLVMEmitter.cpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -2438,8 +2438,16 @@ static FailureOr<StringRef> buildPstuCallee(MLIRContext *context, pto::PstuOp op
return failure();
}

static StringRef buildVstusCallee(MLIRContext *context) {
return StringAttr::get(context, "llvm.hivm.vstus").getValue();
static FailureOr<StringRef> buildVstusCallee(MLIRContext *context,
Type valueType) {
std::string vec =
getMemoryElementTypeFragment(getElementTypeFromVectorLike(valueType));
auto lanes = getElementCountFromVectorLike(valueType);
if (vec.empty() || !lanes)
return failure();
return StringAttr::get(context, "llvm.hivm.vstus.v" +
std::to_string(*lanes) + vec)
.getValue();
}

static FailureOr<StringRef> buildVstusPostCallee(MLIRContext *context,
Expand Down Expand Up @@ -8111,7 +8119,8 @@ class LowerVstusOpPattern final : public OpConversionPattern<pto::VstusOp> {
"unexpected converted vstus operand/result types");
}

FailureOr<StringRef> calleeName = buildVstusCallee(op.getContext());
FailureOr<StringRef> calleeName =
buildVstusCallee(op.getContext(), op.getValue().getType());
if (usePostIntrinsic)
calleeName =
buildVstusPostCallee(op.getContext(), op.getValue().getType());
Expand Down
3 changes: 2 additions & 1 deletion test/lit/vpto/vreg_low_precision_memory_vpto_llvm.pto
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -95,7 +95,8 @@ module attributes {pto.target_arch = "a5", pto.kernel_kind = #pto.kernel_kind<ve
// CHECK: call <32 x i8> @llvm.hivm.vldas
// CANN: call { <256 x i8>, <32 x i8>, ptr addrspace(6) } @llvm.hivm.vldus.v256s8
// BETA: call { <256 x i8>, <32 x i8>, ptr addrspace(6) } @llvm.hivm.vldus.v256hif8
// CHECK: call <32 x i8> @llvm.hivm.vstus
// CANN: call <32 x i8> @llvm.hivm.vstus.v256s8
// BETA: call <32 x i8> @llvm.hivm.vstus.v256hif8
// CHECK: call <32 x i8> @llvm.hivm.vstur
// CHECK: call void @llvm.hivm.vstar

Expand Down
37 changes: 37 additions & 0 deletions test/lit/vpto/vstus_mixed_types_vpto_llvm.pto
View file
Open in desktop
Original file line number Diff line number Diff line change
@@ -0,0 +1,37 @@
// Copyright (c) 2026 Huawei Technologies Co., Ltd.
// This program is free software; you can redistribute it and/or modify it under the terms of
// the CANN Open Software License Agreement Version 2.0 (the "License").
// Please refer to the License for the specific language governing permissions and limitations under the License.

// Keep vstus callees type-specific so different element types can coexist in
// one kernel without conflicting LLVM declarations.
// RUN: ptoas --pto-arch=a5 --pto-backend=vpto --emit-vpto-llvm-ir %s -o - 2>&1 | FileCheck %s --check-prefix=DEFAULT
// RUN: ptoas --cann-output-version=9.0.0 --pto-arch=a5 --pto-backend=vpto --emit-vpto-llvm-ir %s -o - 2>&1 | FileCheck %s --check-prefix=CANN900

module attributes {pto.target_arch = "a5", pto.kernel_kind = #pto.kernel_kind<vector>} {
func.func @vstus_mixed_types(%f32_dst: !pto.ptr<f32, ub>,
%u8_dst: !pto.ptr<ui8, ub>,
%f32_value: !pto.vreg<64xf32>,
%u8_value: !pto.vreg<256xui8>) attributes {pto.kernel} {
%c0_i32 = arith.constant 0 : i32
pto.vecscope {
%f32_align = pto.init_align : !pto.align
%f32_align_out = pto.vstus %f32_align, %c0_i32, %f32_value, %f32_dst :
!pto.align, i32, !pto.vreg<64xf32>, !pto.ptr<f32, ub> -> !pto.align
pto.vstas %f32_align_out, %f32_dst, %c0_i32 : !pto.align, !pto.ptr<f32, ub>, i32
%u8_align = pto.init_align : !pto.align
%u8_align_out = pto.vstus %u8_align, %c0_i32, %u8_value, %u8_dst :
!pto.align, i32, !pto.vreg<256xui8>, !pto.ptr<ui8, ub> -> !pto.align
pto.vstas %u8_align_out, %u8_dst, %c0_i32 : !pto.align, !pto.ptr<ui8, ub>, i32
}
return
}
}

// DEFAULT-LABEL: define void @vstus_mixed_types_mix_aiv
// DEFAULT: call <32 x i8> @llvm.hivm.vstus.v64f32
// DEFAULT: call <32 x i8> @llvm.hivm.vstus.v256i8

// CANN900-LABEL: define void @vstus_mixed_types_mix_aiv
// CANN900: call <32 x i8> @llvm.hivm.vstus.v64f32
// CANN900: call <32 x i8> @llvm.hivm.vstus.v256i8
Loading