FluxML/Flux.jl

Benchmark optimizer latency on GPU

Open

#1,699 opened on Aug 24, 2021

View on GitHub
 (0 comments) (0 reactions) (0 assignees)Julia (619 forks)batch import
cudahelp wantedperformance

Repository metrics

Stars
 (4,725 stars)
PR merge metrics
 (Avg merge 3d 6h) (6 merged PRs in 30d)

Description

Ref. https://github.com/JuliaGPU/GPUCompiler.jl/issues/65#issuecomment-903155469. We should do this for both the current mutating optimizers and for Optimisers.jl. It may be that TTFG is only part of the problem.

Contributor guide

Research direction
Set up benchmarking scripts for optimizer latency on GPU, focusing on both current mutating optimizers and Optimisers.jl. Use CUDA.jl for GPU timing and compare time to first gradient (TTFG) and iteration latency.
Tech stack
None
Domain
machine learningbackend
Issue type
Performance
DifficultyEstimated implementation difficulty for a new contributor, from 1 for very small changes to 5 for expert-level work.
2
Estimated timeA rough time range for an experienced contributor to investigate, implement, test, and prepare a pull request.
1-3 hours
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.
Mostly clear
Prerequisites
JuliaGPU programming
Newbie friendlinessA 1-100 score estimating how approachable this issue is for first-time contributors.
50