NVIDIA / NVIDIA/cutlass

[QST] Questions about ex77 blackwell Flash attention

Open
#2,192 9 comments 0 reactions 0 assignees View on GitHub

Nobody has claimed this yet.

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

Description

  1. Why do we need to include a first batch of padding in example 77? What should the padding size be?
  2. How does example 77 handle variable sequence length with TMA? I was skimming the code and expected to see some tensormap replace calls but don't see any.

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 with example 77 and trace its first-batch padding and variable-sequence-length handling. Determine the intended padding size and explain how TMA supports the varying lengths, including why no tensormap replace calls appear; the issue is done when both questions are answered clearly.

Written by the indexing model from the issue text.

Assessment

Tech stack
cpp
Domain
machine-learning, performance
Issue type
Documentation
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.