Always enable P2P access for GPU copies #21872
Closed
Conversation
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
PR pytorch#20685 incorrectly only enabled P2P access for non-contiguous copies. This can make cudaMemcpy slow for inter-gpu copies, especially on ROCm devices.
pytorchbot
added
module: cuda
Contributor
facebook-github-bot
left a comment
facebook-github-bot
left a comment
There was a problem hiding this comment.
@colesbury has imported this pull request. If you are a Facebook employee, you can view this diff on Phabricator.
soumith
approved these changes
Jun 17, 2019
Member
Author
|
I've run the imagenet ResNet-18 example (without data loading) on four Vega 20 cards. Perf is ~0.113 ms/batch of 256 vs. ~0.175 ms/batch before this PR. (Before 320c385 perf was ~0.118 ms/batch). |
Contributor
|
@colesbury merged this pull request in cc4498a. |
zdevito
pushed a commit
to zdevito/ATen
that referenced
this pull request
Jun 18, 2019
Summary: PR pytorch/pytorch#20685 incorrectly only enabled P2P access for non-contiguous copies. This can make cudaMemcpy slow for inter-gpu copies, especially on ROCm devices. I didn't notice a difference on CUDA 10, but ngimel says it's important for CUDA too. Pull Request resolved: pytorch/pytorch#21872 Differential Revision: D15863965 Pulled By: colesbury fbshipit-source-id: 0a858f3c338fa2a5d05949d7f65fc05a70a9dfe1
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
5 participants
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
PR #20685 incorrectly only enabled P2P access for non-contiguous copies.
This can make cudaMemcpy slow for inter-gpu copies, especially on ROCm
devices. I didn't notice a difference on CUDA 10, but @ngimel says it's
important for CUDA too.