Heuristics for mmq_id -> original threshold #734
Merged
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
Nexesenex
added a commit
to Nexesenex/ik_llama.cpp.nxs
that referenced
this pull request
Nov 6, 2025
This reverts commit 966a6ce.
Nexesenex
added a commit
to Nexesenex/ik_llama.cpp.nxs
that referenced
this pull request
Nov 6, 2025
This reverts commit 966a6ce.
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Outdated suggestions cannot be applied.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 is a follow up of #728.
For large enough u-batches the original implementation of the fused
ffn_up_exps+ffn_gate_expsop becomes faster than themmq_idimplementation added in #728. In #728 a fixed threshold ofu-batch = 2048was used to transition to the original implementation. I have now investigated the speed of original vsmmq_idfor 3 models with different number of total and active experts, and it looks like the best heuristics is to usemmq_idforu-batch <= 32 * total_experts. This PR makes this simple change.