llama : switch KQ multiplication to use F32 precision by default #10015
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
Nexesenex
added a commit
to Nexesenex/croco.cpp
that referenced
this pull request
Oct 24, 2024
…rganov#10015 Co-Authored-By: Georgi Gerganov <ggerganov@gmail.com>
This was referenced Oct 24, 2024
|
Hi, just forwarding some findings from community members - apparently this change to using F32 precision causes a significant speed regression (-30% T/s) on HIPBLAS. I personally use a Nvidia GPU and there's been no issues there for me. They are using ROCM rx 6800xt and 5800x3d, 7900 xtx (driver 24.9.1) Regression was found by bisecting commits until arriving at this one. Not sure if anything can be done, but just thought I'd highlight it to see if anyone else using AMD GPUs has observed similar issues. Also tagging @YellowRoseCx as they use AMD and can assist in testing/verifying if needed. |
LostRuins
added a commit
to LostRuins/koboldcpp
that referenced
this pull request
Nov 4, 2024
… only for now pending ROCM investigation re ggerganov#10015
arthw
pushed a commit
to arthw/llama.cpp
that referenced
this pull request
Nov 15, 2024
arthw
pushed a commit
to arthw/llama.cpp
that referenced
this pull request
Nov 18, 2024
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Outdated suggestions cannot be applied.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.
ref #10005, #9991 (comment)
The list of models that require higher floating point range in the attention keeps growing, so to be on the safe side, default to F32 for the KQ multiplication.