Skip to content

[mlir][nvgpu] Improve verifier of ldmatrix #77807

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
merged 1 commit into from
Jan 12, 2024
Merged

[mlir][nvgpu] Improve verifier of ldmatrix #77807

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
3 changes: 3 additions & 0 deletions mlir/lib/Dialect/NVGPU/IR/NVGPUDialect.cpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -321,6 +321,9 @@ LogicalResult LdMatrixOp::verify() {
if (isTranspose && !(elementBitWidth == 16))
return emitError()
<< "nvgpu.ldmatrix transpose works only at 16b granularity";
if (resShape.size() != 2) {
return emitError() << "results must be 2 dimensional vector";
}
if (!(resShape[1] == numElementsPer32b))
return emitError() << "expected vector register shape[1] = "
<< numElementsPer32b;
Expand Down
8 changes: 8 additions & 0 deletions mlir/test/Dialect/NVGPU/invalid.mlir
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -40,6 +40,14 @@ func.func @ldmatrix_trans_f32_x4(%arg0: memref<128x128xf32, 3>) -> vector<4x1xf
}
// -----

func.func @ldmatrix_trans_f32_x4(%arg0: memref<128x128xf32, 3>) -> vector<4x1xf32> {
%c0 = arith.constant 0 : index
// expected-error @+1 {{results must be 2 dimensional vector}}
%a = nvgpu.ldmatrix %arg0[%c0, %c0] {transpose = false, numTiles = 4 : i32} : memref<128x128xf32, 3> -> vector<4xf32>
return %a : vector<4xf32>
}
// -----

func.func @ldmatrix_type_x4(%arg0: memref<128x128xf32, 3>) -> vector<4x2xf16> {
%c0 = arith.constant 0 : index
// expected-error @+1 {{'nvgpu.ldmatrix' op failed to verify that srcMemref and res have same element type}}
Expand Down