Skip to content

Fix sparse CUTLASS-based kernels #150023

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 2 commits into from
Closed

Fix sparse CUTLASS-based kernels #150023

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
2 commits
Select commit Hold shift + click to select a range
59b496a
Update
alexsamardzic Mar 26, 2025
6e886bc
Update
alexsamardzic Mar 26, 2025
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
8 changes: 1 addition & 7 deletions aten/src/ATen/native/sparse/cuda/SparseSemiStructuredLinear.cu
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -75,12 +75,6 @@ Tensor two_four_sgemm(
using LayoutC = LayoutOutput;
constexpr int AlignmentC = 128 / cutlass::sizeof_bits<ElementC>::value;

using BiasTileThreadMap = cutlass::epilogue::threadblock::OutputTileThreadLayout<
ThreadblockShape,
WarpShape,
ElementC,
AlignmentC,
NumEVTEpilogueStages>;
using OutputTileThreadMap = cutlass::epilogue::threadblock::OutputTileThreadLayout<
ThreadblockShape,
WarpShape,
Expand All @@ -94,7 +88,7 @@ Tensor two_four_sgemm(
cutlass::epilogue::threadblock::VisitorScalarBroadcast<ElementC>;
using BiasTensor =
cutlass::epilogue::threadblock::VisitorColBroadcast<
BiasTileThreadMap,
OutputTileThreadMap,
ElementC,
cute::Stride<cute::_1, cute::_0, int64_t>>;
using Bias = std::conditional_t<use_bias, BiasTensor, BiasScalar>;
Expand Down
8 changes: 1 addition & 7 deletions aten/src/ATen/native/sparse/cuda/SparseSemiStructuredOps.cu
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -70,12 +70,6 @@ void spgemm_cutlass(
using LayoutC = LayoutOutput;
constexpr int AlignmentC = 128 / cutlass::sizeof_bits<ElementC>::value;

using TensorCTileThreadMap = cutlass::epilogue::threadblock::OutputTileThreadLayout<
ThreadblockShape,
WarpShape,
ElementC,
AlignmentC,
NumEVTEpilogueStages>;
using OutputTileThreadMap = cutlass::epilogue::threadblock::OutputTileThreadLayout<
ThreadblockShape,
WarpShape,
Expand Down Expand Up @@ -105,7 +99,7 @@ void spgemm_cutlass(
cutlass::epilogue::threadblock::VisitorScalarBroadcast<ElementC>;
using TensorCTensor =
cutlass::epilogue::threadblock::VisitorColBroadcast<
TensorCTileThreadMap,
OutputTileThreadMap,
ElementC,
cute::Stride<cute::_1, cute::_0, int64_t>>;
using TensorC = std::conditional_t<use_tensor_c, TensorCTensor, TensorCScalar>;
Expand Down
Loading