[Native] Use vdotq_s32 for int7u/int8 distances on ARM#144505
Merged
ldematte merged 5 commits intoelastic:mainfrom Mar 19, 2026
Merged
[Native] Use vdotq_s32 for int7u/int8 distances on ARM#144505ldematte merged 5 commits intoelastic:mainfrom
ldematte merged 5 commits intoelastic:mainfrom
Conversation
Replace vmull_s8 + vpadalq_s16 with vdotq_s32 in doti8_inner, cosi8_inner, cosi8_inner_bulk, and call_i8_bulk (dot path). Make call_i8_bulk accumulator type parametric via acc_ops traits. Bump -march to armv8.2-a+dotprod and add HWCAP_ASIMDDP runtime check in caps.cpp. Made-with: Cursor
Collaborator
|
Pinging @elastic/es-search-relevance (Team:Search Relevance) |
Collaborator
|
Hi @ldematte, I've created a changelog YAML for you. |
michalborek
pushed a commit
to michalborek/elasticsearch
that referenced
this pull request
Mar 23, 2026
Replace vmull_s8 + vpadalq_s16 with vdotq_s32 in int7/int8 native implementation on ARM. Benchmarks show a 40 to 60% speedup in scoring.
This was referenced Apr 1, 2026
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
3 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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Replace
vmull_s8+vpadalq_s16withvdotq_s32in int7/int8 native implementation on ARM.Benchmarks:
Single-vector (ops/us, dims=1024)
Bulk (ops/s, dims=1024, numVectors=1500, bulkSize=32)