Not capping thread count when MoE inference is running on CPU #5419
Conversation
This file contains 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
slaren
approved these changes
Feb 8, 2024
|
Will see if this fixes my Mixtral prompt processing regression on my intel i5. |
|
Before: After: Thread usage seems a lot more balanced now (as it used to be), and prompt processing for pure CPU q8_0 Mixtral drops to 100ms per token instead of 125ms per token. More specifically, I get 9-10t/s prompt processing instead of 7-8t/s prompt processing on pure CPU alone for q8_0. @JohannesGaessler it seems the regression for partial offloading that I mentioned to you was in fact CPU related, and this PR should fix it :) |
ggerganov
approved these changes
Feb 9, 2024
jordankanter
pushed a commit
to jordankanter/llama.cpp
that referenced
this pull request
Mar 13, 2024
* Not capping thread count when MoE inference is running on CPU * Whitespace
hodlen
pushed a commit
to hodlen/llama.cpp
that referenced
this pull request
Apr 1, 2024
* Not capping thread count when MoE inference is running on CPU * Whitespace
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
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Outdated suggestions cannot be applied.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.
This seems to fix a performance ceiling when inferring MoE on CPU, as described here: #5417