Skip to content

Vulkan-run-test: fix mmq_wg_denoms #11343

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
merged 1 commit into from
Jan 23, 2025
Merged

Conversation

AMD-dwang
Copy link
Contributor

There should be a copy-and-paste error here.
*mmq_wg_denoms should be used together with *warptile_mmq, instead of wg_denoms.

There should be a copy-and-paste error here.

*mmq_wg_denoms should be used together with *warptile_mmq, instead of
wg_denoms.
@github-actions github-actions bot added Vulkan Issues specific to the Vulkan backend ggml changes relating to the ggml tensor library for machine learning labels Jan 22, 2025
@0cc4m
Copy link
Collaborator

0cc4m commented Jan 22, 2025

Yes, I copied this wrong. It didn't make a difference because wg_denoms == mmq_wg_denoms for non-coopmat2, but it should be fixed. I'll run a test later, but it should be fine.

@0cc4m 0cc4m self-requested a review January 22, 2025 08:54
@AMD-dwang
Copy link
Contributor Author

Yes, I copied this wrong. It didn't make a difference because wg_denoms == mmq_wg_denoms for non-coopmat2, but it should be fixed. I'll run a test later, but it should be fine.

l_wg_denoms or l_mmq_wg_denoms may be modified. Please see https://github.com/AMD-dwang/llama.cpp/blob/vulkan-compMat/ggml/src/ggml-vulkan/ggml-vulkan.cpp#L1476-L1499

l_wg_denoms and l_warptile_mmq will mismatch once it is modified. I just hit this error, then the number of work group is insufficient.

@0cc4m
Copy link
Collaborator

0cc4m commented Jan 22, 2025

You're right, I forgot about that logic. Not sure it's needed anymore, but the fix is definitely necessary.

Copy link
Collaborator

@0cc4m 0cc4m left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Thank you for the fix, looks good.

@0cc4m 0cc4m merged commit 955a6c2 into ggml-org:master Jan 23, 2025
45 checks passed
anagri pushed a commit to BodhiSearch/llama.cpp that referenced this pull request Jan 26, 2025
There should be a copy-and-paste error here.

*mmq_wg_denoms should be used together with *warptile_mmq, instead of
wg_denoms.
tinglou pushed a commit to tinglou/llama.cpp that referenced this pull request Feb 13, 2025
There should be a copy-and-paste error here.

*mmq_wg_denoms should be used together with *warptile_mmq, instead of
wg_denoms.
arthw pushed a commit to arthw/llama.cpp that referenced this pull request Feb 26, 2025
There should be a copy-and-paste error here.

*mmq_wg_denoms should be used together with *warptile_mmq, instead of
wg_denoms.
mglambda pushed a commit to mglambda/llama.cpp that referenced this pull request Mar 8, 2025
There should be a copy-and-paste error here.

*mmq_wg_denoms should be used together with *warptile_mmq, instead of
wg_denoms.
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
ggml changes relating to the ggml tensor library for machine learning Vulkan Issues specific to the Vulkan backend
Projects
None yet
Development

Successfully merging this pull request may close these issues.

3 participants