sgl-project/sglang

[Kernel] cuDNN attention backend

Open

#2,272 opened on Nov 30, 2024

View on GitHub
 (5 comments) (1 reaction) (2 assignees)Python (6,216 forks)auto 404
enhancementgood first issuehelp wantedhigh priorityinactive

Repository metrics

Stars
 (28,442 stars)
PR merge metrics
 (Avg merge 2d 1h) (1,000 merged PRs in 30d)

Description

cuDNN provides very fast attention implementation and it is well maintained by NVIDIA. We would like to add a new attention backend based on cudnn.

Steps

  1. Learn this cudnn paged attention python api. https://github.com/NVIDIA/cudnn-frontend/blob/v1.8.0/samples/python/52_scaled_dot_product_attention_with_paged_caches.ipynb
  2. Add a new attention backend "cudnn" here https://github.com/sgl-project/sglang/tree/main/python/sglang/srt/layers/attention
  3. We should be able to use it with python3 -m sglang.launch_server --model meta-llama/Llama-3.1-8B-Instruct --attention-backend cudnn

Contributor guide

Research direction
Understand the cuDNN paged attention Python API and integrate it as a new attention backend in the codebase.
Tech stack
python
Domain
backendmachine learning
Issue type
Feature
DifficultyEstimated implementation difficulty for a new contributor, from 1 for very small changes to 5 for expert-level work.
3
Estimated timeA rough time range for an experienced contributor to investigate, implement, test, and prepare a pull request.
1-2 days
Activity statusHow available the issue appears right now: fresh, active, stale, blocked, or waiting on maintainer input.
Active
ClarityHow clearly the issue explains the expected change, acceptance criteria, and next step.
Clear
Prerequisites
PythonCUDAcuDNN
Newbie friendlinessA 1-100 score estimating how approachable this issue is for first-time contributors.
60