Skip to content

Update cagra config yaml for cuvs_bench #1822

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.

Sign up for GitHub

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

Open
tfeher wants to merge 3 commits into
base: main
Choose a base branch
from
Open

Update cagra config yaml for cuvs_bench #1822

Show file tree
Hide file tree
Changes from 2 commits
Commits
Show all changes
3 commits
Select commit Hold shift + click to select a range
7052f7f
Update cagra config yaml for cuvs_bench
tfeher Feb 19, 2026
473ed74
Update max iterations
tfeher Feb 19, 2026
e06e1ee
Merge branch 'main' into cuvs_bench_cagra_config
cjnolet May 13, 2026
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
11 changes: 6 additions & 5 deletions python/cuvs_bench/cuvs_bench/config/algos/cuvs_cagra.yaml
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -5,17 +5,18 @@ constraints:
groups:
base:
build:
graph_degree: [32, 64, 96, 128]
intermediate_graph_degree: [32, 64, 96, 128]
graph_build_algo: ["NN_DESCENT"]
graph_degree: [32, 64, 96]
intermediate_graph_degree: [128]

@tarang-jain tarang-jain Feb 19, 2026

Copy link
Copy Markdown
Contributor

Choose a reason for hiding this comment

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

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.

@tfeher tfeher May 15, 2026

Copy link
Copy Markdown
Contributor Author

Choose a reason for hiding this comment

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

Ideally yes, intermediate_graph_degree = 2 * graph_degree would be sufficient. Our benchmark framework does not adjust this automatically. I would not expect recall drop by using larger intermediate degree. Do we have any detail on that that issue?

graph_build_algo: ["IVF_PQ"]
search:
itopk: [32, 64, 128, 256, 512]
search_width: [1, 2, 4, 8, 16, 32, 64]
max_iterations: [0, 16]
search_width: [1, 2, 4, 8]
test:
build:
graph_degree: [32]
intermediate_graph_degree: [32]
graph_build_algo: ["NN_DESCENT"]
graph_build_algo: ["IVF_PQ"]
search:
itopk: [32]
search_width: [1, 2]
11 changes: 6 additions & 5 deletions python/cuvs_bench/cuvs_bench/config/algos/cuvs_mg_cagra.yaml
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -5,17 +5,18 @@ constraints:
groups:
base:
build:
graph_degree: [32, 64, 96, 128]
intermediate_graph_degree: [32, 64, 96, 128]
graph_build_algo: ["NN_DESCENT"]
graph_degree: [32, 64, 96]
intermediate_graph_degree: [128]
graph_build_algo: ["IVF_PQ"]
search:
itopk: [32, 64, 128, 256, 512]
search_width: [1, 2, 4, 8, 16, 32, 64]
max_iterations: [0, 16]
search_width: [1, 2, 4, 8]
test:
build:
graph_degree: [32]
intermediate_graph_degree: [32]
graph_build_algo: ["NN_DESCENT"]
graph_build_algo: ["IVF_PQ"]
search:
itopk: [32]
search_width: [1, 2]