lllyasviel / lllyasviel/ControlNet

Multi GPU slower than One GPU several time

Open
#556 0 comments 0 reactions 0 assignees View on GitHub
Dominant language
Python
Stars
34.1k
Forks
3k
PR merge metrics
No merged PRs in 30d

Description

HI

When I use multi-gpu command, I found the training time is slower several times than that of one gpu.
```
trainer = pl.Trainer(strategy="ddp", accelerator="gpu", gpus=2, precision=16, callbacks=[logger])
```

Contributor guide

No contributing guide indexed for this repository

Research direction

Start by reproducing the reported comparison using the provided Trainer configuration with one GPU and with DDP on two GPUs. Record the training times and environment details, then investigate the multi-GPU execution path to identify the source of the slowdown. Done means the cause is isolated and the two-GPU configuration no longer takes several times longer than one GPU.

Written by the indexing model from the issue text.

Assessment

Tech stack
python
Domain
machine-learning
Issue type
Bug
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.