Open
Conversation
cjnolet
added
improvement
tarang-jain
reviewed
Feb 19, 2026
Contributor
tarang-jain
left a comment
tarang-jain
left a comment
There was a problem hiding this comment.
LGTM except one comment.
| intermediate_graph_degree: [32, 64, 96, 128] | ||
| graph_build_algo: ["NN_DESCENT"] | ||
| graph_degree: [32, 64, 96] | ||
| intermediate_graph_degree: [128] |
Contributor
There was a problem hiding this comment.
Dont we want intermediate_graph_degree 64 when the graph degree is 32? @bkarsin had also raised an issue in the past wherein CAGRA can give poor recall when the intermediate graph degree is quite large in comparison to the graph degree.
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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 PR updates the CAGRA config options for the parameter sweep used by cuvs-bench.
intermediate_graph_degree: in practice we often useintermediate_graph_degree = graph_degree * 1.5. We pick a fixed value that works works with all of thegraph_degreeoptions we want to test.max_iterations: the default 0 value auto selects the iterations based on itopk. I have added value16additionally, which is a low number of iterations, and it can be useful to reach large throughput in recall < 0.9 region.search_width: we often reach around 99% recall with width 8, therefore I propose to remove higher values from the base scangraph_build_algo: ideally we wound want the algorithm automatically selected. Currently the benchmark framework would select the iterative method by default, which does not work with memory mapped data. The NN descent method can run out of memory for 10M scale dataset. Therefore I propose to fixIVF_PQ, until we enable a reliable default algorithm selection.