Vectorize search_n
for CPUs with SSE4.2 but not AVX2 support; handle AVX2 tail
#5544
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Outdated suggestions cannot be applied.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.
Copy AVX2 path to SSE4.2, adapt it as follows:
movemask
output fits 16 bit (despite using 32-bit register), the mask with carry fits 32 bitCopy AVX2 path to AVX2 tail path, adapt it as follows:
_Carry
with shifted_Msk_with_carry
to take into account previous_Carry
for tails smaller thann
.SSE4.2 benchmark results
These results are actually fake. AVX2 path is artificially disabled on an AVX2-capable CPU.
Also they suffer much from random variation.
Still they are useful enough to see that SSE4.2 path is indeed helpful.
I've highlighted the relevant results.