NVIDIA / NVIDIA/cutlass

[FEA] Specify L2 cache eviction in TMA copy

Open
#2,528 5 comments 3 reactions 0 assignees View on GitHub

Nobody has claimed this yet.

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

Description

Which component requires the feature?

CuTe DSL

Feature Request

I'd love to be able to control L2 cache eviction when doing TMA load and TMA store (e.g. evict_first, evict_last)

Additional context
This is important for some attention kernels, as we used it in FA3, e.g. here:
https://github.com/Dao-AILab/flash-attention/blob/413d07e9deef1e3c793c7de59d7146b43ae4d558/hopper/mainloop_fwd_sm90_tma_gmma_ws.hpp#L753

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 the CuTe DSL implementation of TMA load and store operations, using the linked FA3 hopper/mainloop_fwd_sm90_tma_gmma_ws.hpp example as context for evict_first and evict_last. Done means callers can specify L2 cache eviction behavior for TMA copy operations and the supported modes are covered by relevant tests.

Written by the indexing model from the issue text.

Assessment

Tech stack
cpp
Domain
performance
Issue type
Feature
Difficulty
4/5
Estimated time
3-5 days
Activity status
Quiet
Clarity
Mostly clear
Newbie friendliness
45/100

Get new issues in your inbox

A short digest of beginner-friendly GitHub issues.