MB-71833: Add num_threads omp pragma to IVFRaBitQ#78
Merged
Conversation
Nischal1729
approved these changes
May 15, 2026
deepkaran
approved these changes
May 15, 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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
This change brings IndexIVFRaBitQ in line with the rest of the Faiss codebase by adding num_threads(num_omp_threads) to all OpenMP parallel regions in the class. Previously, the OMP pragmas in encode_vectors, decode_vectors, add_core, and sa_decode did not respect the global num_omp_threads setting (configured via set_num_omp_threads()), causing those operations to always use OpenMP's default thread count regardless of any user-configured limit.
Changes:
• Added #include <faiss/OMPConfig.h> to expose num_omp_threads
• Applied num_threads(num_omp_threads) to four #pragma omp parallel regions in IndexIVFRaBitQ.cpp