ggml : prevent builds with -ffinite-math-only#7726
Merged
Conversation
Contributor
Contributor
|
The issue that I reported in #7154 (comment) has not been fixed by this PR. For the minimal reproduction I was not using |
Member
Author
|
Yes, I didn't expect it to be fixed. We need a non-unified KV cache implementation to have deterministic results for |
Contributor
|
Could it be a memory barrier issue? x86 guarantees about acquire / release semantics change when you switch to xmm/ymm ops. |
Member
Author
|
The problem is not a data race or a race condition. Rather the same set of tokens can produce slightly different floating point results, depending on the position they get assigned in the unified KV cache due to the reduce operations in the attention |
|
btw, this breaks |
Seunghhon
pushed a commit
to Seunghhon/llama.cpp
that referenced
this pull request
Apr 26, 2026
This enforces a check that -fno-finite-math-only was set and that the operating compiling mode is not in finite maths mode. This is because during rewriting of silu and softmax for cpu ggml-org#7154 there emerged an issue where the result that was observed when >1 slot was nondeterministic as found by @JohannesGaessler. @LostRuins narrowed the problem down to -ffinite-math-only which was theorised to be due to SiLU, instead of flushing small values to 0, returns NaN or some other garbage. @jart proposed a fix that @ggerganov then implemented in this fix ref ggml-org#7154 (comment)
phuongncn
pushed a commit
to phuongncn/llama.cpp-gx10-dgx-sparks-deepseekv4
that referenced
this pull request
Apr 28, 2026
This enforces a check that -fno-finite-math-only was set and that the operating compiling mode is not in finite maths mode. This is because during rewriting of silu and softmax for cpu ggml-org#7154 there emerged an issue where the result that was observed when >1 slot was nondeterministic as found by @JohannesGaessler. @LostRuins narrowed the problem down to -ffinite-math-only which was theorised to be due to SiLU, instead of flushing small values to 0, returns NaN or some other garbage. @jart proposed a fix that @ggerganov then implemented in this fix ref ggml-org#7154 (comment)
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
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 enforces a check that -fno-finite-math-only was set and that the operating
compiling mode is not in finite maths mode. This is because during rewriting of
silu and softmax for cpu #7154 there emerged an issue where the result that was
observed when >1 slot was nondeterministic as found by @JohannesGaessler.
@LostRuins narrowed the problem down to -ffinite-math-only which was theorised
to be due to SiLU, instead of flushing small values to 0, returns NaN or some
other garbage. @jart proposed a fix that @ggerganov then implemented in this fix
ref #7154 (comment)