NVIDIA / NVIDIA/cutlass

[QST] a confuse acout struct TensorOpMultiplicand

Open
#2,338 4 comments 0 reactions 0 assignees View on GitHub

Nobody has claimed this yet.

inactive-30d inactive-90d
Dominant language
C++
Stars
10.5k
Forks
2.1k
Avg merge
3d 11h
Merged PRs (30d)
7

Description

https://github.com/NVIDIA/cutlass/blob/6316b6f8672071ea69a1161a0d73722372e1eb57/include/cutlass/layout/tensor_op_multiplicand_sm75.h#L151

coord.contiguous() is always in [0,Crosswise)?
if so, tile_contiguous_idx seems being always 0?

    // Compute the fundamental tile being accessed
    int tile_contiguous_idx =
        vec_contiguous_idx / (TileShape::kContiguous / kFactor);

Contributor guide

No contributing guide indexed for this repository

First steps

  1. Read the whole issue, then the project's contributing guide.
  2. Comment on the issue to say you are picking it up — it saves two people doing the same work.
  3. Fork the repository and make your change on a branch.
  4. Open a pull request that references the issue number.

Research direction

Start at include/cutlass/layout/tensor_op_multiplicand_sm75.h around line 151 and trace how coord.contiguous(), vec_contiguous_idx, and TileShape::kContiguous determine tile_contiguous_idx. Confirm whether the index is always zero for the relevant layout, then document the result or identify the required correction. No test or validation command is named in the issue.

Written by the indexing model from the issue text.

Assessment

Tech stack
cpp
Domain
performance
Issue type
Bug
Difficulty
4/5
Estimated time
3-5 days
Activity status
Stale
Clarity
Needs clarification
Newbie friendliness
25/100

Get new issues in your inbox

A short digest of beginner-friendly GitHub issues.