Skip to content

[SM6.10][Bugfix] Make sure DstN::Value calculation is rounded up #8359

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
hekota merged 1 commit into from
Apr 10, 2026
+9 −1
Merged

[SM6.10][Bugfix] Make sure DstN::Value calculation is rounded up #8359

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
5 changes: 4 additions & 1 deletion tools/clang/lib/Headers/hlsl/dx/linalg.h
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -186,8 +186,11 @@ __MATRIX_SCALAR_COMPONENT_MAPPING(ComponentType::U64, uint64_t)
__MATRIX_SCALAR_COMPONENT_MAPPING(ComponentType::F64, double)

template <ComponentEnum DstTy, ComponentEnum SrcTy, int SrcN> struct DstN {
// Make sure to round up in case SrcN isn't an even multiple of the number of
// elements per scalar
static const int Value =
(SrcN * ComponentTypeTraits<SrcTy>::ElementsPerScalar) /
(SrcN * ComponentTypeTraits<SrcTy>::ElementsPerScalar +
ComponentTypeTraits<DstTy>::ElementsPerScalar - 1) /
ComponentTypeTraits<DstTy>::ElementsPerScalar;
};

Expand Down
5 changes: 5 additions & 0 deletions tools/clang/test/CodeGenDXIL/hlsl/linalg/api/vectors.hlsl
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -82,4 +82,9 @@ void main(uint ID : SV_GroupID) {
half16 srcF16 = BAB.Load<half16>(128);
InterpretedVector<uint, 4, ComponentEnum::F8_E4M3FN> convertedPacked = Convert<ComponentEnum::F8_E4M3FN, ComponentEnum::F16>(srcF16);

// CHECK: call <1 x i32> @dx.op.linAlgConvert.v1i32.v3f16(i32 -2147483618, <3 x half> %25, i32 8, i32 21)
// CHECK-SAME: ; LinAlgConvert(inputVector,inputInterpretation,outputInterpretation)
half3 ThreeF16 = BAB.Load<half3>(256);
InterpretedVector<uint, 1, ComponentEnum::F8_E4M3FN> convertedPacked2 =
Convert<ComponentEnum::F8_E4M3FN, ComponentEnum::F16>(ThreeF16);
}
Loading