local NVMe caching to speed up server start time / 本地 NVMe 缓存加速服务器启动时间

Open
#629 0 comments 0 reactions 0 assignees View on GitHub

Nobody has claimed this yet.

Assessment

Difficulty
5/5
Estimated time
Over a week
Newbie friendliness
25/100
Issue type
Feature
Clarity
Needs clarification
Activity status
Stale
Tech stack
python
Domain
infrastructure, performance

Research direction

No files, tests, or entry points are identified. Start by locating the server's model-weight loading path and local storage configuration; done would require the proposed local NVMe cache and LRU behavior to be implemented and its startup-time impact validated.

Written by the indexing model from the issue text.

Description

dev velocity p1

currently a lot of GPU time is wasted by loading weights espically at low concurrency due to parallel filesystem speeds. as an percentage of time, weight loading is an signficiant amount espcially for low concucrrency

in addition, this hurts developer velocity

peusodcode

def function(model_id):
   if model_id on local NVMe:
       return path
   else 
       if localNVMe is full:
          delete LRU
       copy from remote to local

中文说明

通过本地 NVMe 缓存加速服务器启动时间。

Dominant language
Python
Stars
1.7k
Forks
303
Avg merge
1d 13h
Merged PRs (30d)
284

Contributor guide

Open the contributing guide

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.

More from SemiAnalysisAI/InferenceX

All issues in SemiAnalysisAI/InferenceX

Similar issues

More Python issues

Get new issues in your inbox

A short digest of beginner-friendly GitHub issues.