UPSTREAM PR #17397: vulkan: Revive MUL_MAT_ID to perf testing #275
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
loci-dev
temporarily deployed
to
PROD__AL_DEMO
GitHub Actions
Inactive
loci-dev
force-pushed
the
main
branch
5 times, most recently
from
75ebdd4 to
ec397c5
Compare
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Outdated suggestions cannot be applied.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.
Mirrored from ggml-org/llama.cpp#17397
Not sure if this was intended but the original MUL_MAT_ID perf tests were replaced by MUL_MAT_ID_FUSION tests at #17095, and MUL_MAT_ID cannot be measured by
test-backend-ops perf -o MUL_MAT_IDat the moment. We'd like to revive this since it is actively used in models like gpt-oss (until it is fully replaced by MUL_MAT_ID_FUSION ops?)