Skip to content

Conversation

@aartbik
Copy link
Collaborator

@aartbik aartbik commented Jun 27, 2025

No description provided.

@aartbik aartbik requested review from cliffburdick and Copilot June 27, 2025 20:04
@copy-pr-bot
Copy link

copy-pr-bot bot commented Jun 27, 2025

This pull request requires additional validation before any workflows can run on NVIDIA's runners.

Pull request vetters can view their responsibilities here.

Contributors can view more details about this message here.

Copy link
Contributor

Copilot AI left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Pull Request Overview

This PR adds stream parameters to the allocation and free functions in various CUDA sparse matrix operations to support asynchronous execution.

  • Updated allocation calls in MatVecCUSPARSEHandle_t, MatMulCUSPARSEHandle_t, Sparse2DenseHandle_t, and Dense2SparseHandle_t to pass a stream.
  • Updated free calls in the same classes to use the stream from params_.stream, aligning resource management with the new stream support.

Reviewed Changes

Copilot reviewed 4 out of 4 changed files in this pull request and generated no comments.

File Description
include/matx/transforms/matmul/matvec_cusparse.h Updated alloc/free calls to support streams.
include/matx/transforms/matmul/matmul_cusparse.h Updated alloc/free calls to support streams.
include/matx/transforms/convert/sparse2dense_cusparse.h Updated alloc/free calls to support streams.
include/matx/transforms/convert/dense2sparse_cusparse.h Updated alloc/free calls to support streams.
Comments suppressed due to low confidence (5)

include/matx/transforms/matmul/matvec_cusparse.h:153

  • Ensure that the 'stream' passed to matxAlloc is consistent with the stream used in matxFree (params_.stream). Consider using a unified stream reference for clarity.
                stream);

include/matx/transforms/matmul/matmul_cusparse.h:159

  • Verify that the stream used in allocation ('stream') matches the stream retrieved via params_.stream in the free call to maintain consistency.
                stream);

include/matx/transforms/convert/sparse2dense_cusparse.h:127

  • Ensure that the 'stream' provided to matxAlloc is the same underlying stream used later in matxFree to avoid potential mismatches.
                stream);

include/matx/transforms/convert/dense2sparse_cusparse.h:136

  • Confirm that the 'stream' passed during allocation is correctly aligned with the stream used in the free call (params_.stream), ensuring consistent resource management.
                stream);

include/matx/transforms/convert/dense2sparse_cusparse.h:182

  • Verify that params_.stream is intended to be used here and is consistent with the stream passed in the allocation call to avoid mismatches.
      matxFree(workspace_, params_.stream);

@cliffburdick
Copy link
Collaborator

/build

2 similar comments
@cliffburdick
Copy link
Collaborator

/build

@cliffburdick
Copy link
Collaborator

/build

@cliffburdick cliffburdick merged commit 90470c6 into NVIDIA:main Jun 30, 2025
1 check passed
@cliffburdick cliffburdick deleted the buk branch June 30, 2025 14:09
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

2 participants