Skip to content

Add panama implementations of byte-bit and float-bit script operations #124722

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
merged 14 commits into from
Mar 25, 2025

Conversation

thecoop
Copy link
Member

@thecoop thecoop commented Mar 13, 2025

Resolves #117096

This gets about a 6-10x speedup:

Benchmark                                  (dims)  (docType)  (function)  (queryType)  (type)  Mode  Cnt    Score   Error  Units
DistanceFunctionBenchmark.benchmark          1024        BIT         DOT        FLOAT  BINARY  avgt    2  441.688          ns/op
DistanceFunctionBenchmark.benchmark          1024        BIT         DOT         BYTE  BINARY  avgt    2  357.740          ns/op
DistanceFunctionBenchmark.vectorBenchmark    1024        BIT         DOT        FLOAT  BINARY  avgt    2   48.748          ns/op
DistanceFunctionBenchmark.vectorBenchmark    1024        BIT         DOT         BYTE  BINARY  avgt    2   55.824          ns/op

Still need to determine if 128-bit versions are helpful

@thecoop thecoop added the WIP label Mar 14, 2025
@thecoop thecoop marked this pull request as ready for review March 14, 2025 11:20
@thecoop thecoop force-pushed the vector-bit-dot-product branch from e7c3f9a to d66ee15 Compare March 24, 2025 11:25
@thecoop thecoop removed the WIP label Mar 24, 2025
@elasticsearchmachine elasticsearchmachine added the Team:Search Relevance Meta label for the Search Relevance team in Elasticsearch label Mar 24, 2025
@elasticsearchmachine
Copy link
Collaborator

Pinging @elastic/es-search-relevance (Team:Search Relevance)

@benwtrent
Copy link
Member

Could you add tests to ESVectorUtilTests ? I don't think the ones I have written there actually cover indices that are big enough to be vectorized.

@thecoop
Copy link
Member Author

thecoop commented Mar 24, 2025

The PR includes changes to ESVectorUtilTests that creates random vectors up to 1024 dims

@elasticsearchmachine
Copy link
Collaborator

Hi @thecoop, I've created a changelog YAML for you.

Copy link
Member

@benwtrent benwtrent left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

marked it as enhancement as its a performance improvement :)

@thecoop thecoop merged commit 7f1203e into elastic:main Mar 25, 2025
17 checks passed
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
>enhancement :Search Relevance/Vectors Vector search Team:Search Relevance Meta label for the Search Relevance team in Elasticsearch v9.1.0
Projects
None yet
Development

Successfully merging this pull request may close these issues.

SIMD optimized asymmetric bit dot-product
3 participants