ggml-quants: use _mm256_testz_si256 for mask checks in AVX2 #17641
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
`_mm256_testz_si256` directly checks if all bits of a vector are zero, which is a more efficient approach for conditional branching than extracting an 8-bit mask and then checking if the mask is non-zero. This optimization leverages specific AVX2 instruction capabilities, potentially reducing instruction latency and improving overall performance by avoiding unnecessary register transfers for the mask. References: * When to use _mm256_testz_si256 vs _mm256_movemask_epi8: [https://stackoverflow.com/questions/27643534/when-to-use-mm256-testz-si256-vs-mm256-movemask-epi8](https://stackoverflow.com/questions/27643534/when-to-use-mm256-testz-si256-vs-mm256-movemask-epi8) * AVX2: _mm256_testz_si256 vs _mm256_cmpeq_epi32 and _mm256_movemask_epi8: [https://stackoverflow.com/questions/43206253/avx2-mm256-testz-si256-vs-mm256-cmpeq-epi32-and-mm256-movemask-epi8](https://stackoverflow.com/questions/43206253/avx2-mm256-testz-si256-vs-mm256-cmpeq-epi32-and-mm256-movemask-epi8) * Intel Intrinsics Guide for `_mm256_testz_si256`: [https://www.intel.com/content/www/us/en/docs/intrinsics-guide/index.html#text=_mm256_testz_si256](https://www.intel.com/content/www/us/en/docs/intrinsics-guide/index.html#text=_mm256_testz_si256) * Intel Intrinsics Guide for `_mm256_movemask_epi8`: [https://www.intel.com/content/www/us/en/docs/intrinsics-guide/index.html#text=_mm256_movemask_epi8](https://www.intel.com/content/www/us/en/docs/intrinsics-guide/index.html#text=_mm256_movemask_epi8) * Efficiently checking for zero vectors with AVX2: [https://lemire.me/blog/2018/06/18/efficiently-checking-for-zero-vectors-with-avx2/](https://lemire.me/blog/2018/06/18/efficiently-checking-for-zero-vectors-with-avx2/)
github-actions
bot
added
the
ggml
pwilkin
added
the
vibe-coded
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
2 participants
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.
_mm256_testz_si256directly checks if all bits of a vector are zero, which is a more efficient approach for conditional branching than extracting an 8-bit mask and then checking if the mask is non-zero. This optimization leverages specific AVX2 instruction capabilities, potentially reducing instruction latency and improving overall performance by avoiding unnecessary register transfers for the mask.References:
When to use _mm256_testz_si256 vs _mm256_movemask_epi8:
https://stackoverflow.com/questions/27643534/when-to-use-mm256-testz-si256-vs-mm256-movemask-epi8
AVX2: _mm256_testz_si256 vs _mm256_cmpeq_epi32 and _mm256_movemask_epi8:
https://stackoverflow.com/questions/43206253/avx2-mm256-testz-si256-vs-mm256-cmpeq-epi32-and-mm256-movemask-epi8
Intel Intrinsics Guide for
_mm256_testz_si256:https://www.intel.com/content/www/us/en/docs/intrinsics-guide/index.html#text=_mm256_testz_si256
Intel Intrinsics Guide for
_mm256_movemask_epi8:https://www.intel.com/content/www/us/en/docs/intrinsics-guide/index.html#text=_mm256_movemask_epi8
Efficiently checking for zero vectors with AVX2:
https://lemire.me/blog/2018/06/18/efficiently-checking-for-zero-vectors-with-avx2/
Co-Authored-By: Gemini 2.5 Pro (References and description commit changes)