Skip to content

Pull requests: NVIDIA/cuvs

New pull request New
130 Open 1,534 Closed
130 Open 1,534 Closed
Author
Filter by author
Loading
Label
Filter by label
Loading
Use alt + click/return to exclude labels
or ⇧ + click/return for logical OR
Projects
Filter by project
Loading
Milestones
Filter by milestone
Loading
Reviews
Filter by reviews
No reviews Review required Approved review Changes requested
Assignee
Filter by who’s assigned
Assigned to nobody Loading
Sort
Sort by
Newest Oldest Most commented Least commented Recently updated Least recently updated Best match
Most reactions
👍 👎 😄 🎉 😕 ❤️ 🚀 👀

Pull requests list

Keep the polynomial and RBF Gram epilogues in the input precision
#2536 opened Aug 31, 2026 by maxwbuckley Contributor Loading…
2
Keep the sparse Jensen-Shannon post-processing in the input precision
#2535 opened Aug 31, 2026 by maxwbuckley Contributor Loading…
2
Remove trailing new lines from GTest names bug Something isn't working non-breaking Introduces a non-breaking change
#2534 opened Aug 31, 2026 by robertmaynard Contributor Loading…
Bound CAGRA hash table sizing loops so they cannot spin forever
#2533 opened Aug 31, 2026 by shaunakkapur Contributor Loading…
1
[Experimental] Implement dataset as a single type+specs breaking Introduces a breaking change feature request New feature or request
#2532 opened Aug 31, 2026 by achirkin Contributor • Draft
1
Compute the expanded L2 distance in fp32 instead of fp64 C++ improvement Improves an existing functionality non-breaking Introduces a non-breaking change
#2531 opened Aug 31, 2026 by maxwbuckley Contributor Loading…
3
Reuse raft::mdarray/mdspan directly in dense dataset storage
#2530 opened Aug 30, 2026 by HowardHuang1 Contributor Loading…
1
Support pre-Volta (Maxwell/Pascal) GPU architectures
#2529 opened Aug 29, 2026 by drzraf Loading…
3
Fall back to cudaMalloc on devices without stream-ordered allocation
#2528 opened Aug 29, 2026 by drzraf Loading…
1
Use JIT-LTO kernel handles through an API the toolkit actually provides
#2527 opened Aug 29, 2026 by drzraf Loading…
1
[ANN_BENCH] Use Async MR and Non Blocking Streams improvement Improves an existing functionality non-breaking Introduces a non-breaking change
#2526 opened Aug 29, 2026 by tarang-jain Contributor Loading…
4
Adapt pooled stream use to cuda::stream_ref
#2521 opened Aug 28, 2026 by bdice Contributor • Draft
1
Use raft::launch_kernel in cagra search JIT kernels Enhancement improvement Improves an existing functionality non-breaking Introduces a non-breaking change
#2520 opened Aug 28, 2026 by achirkin Contributor • Draft
3
Dry run: preprocessing::quantize::scalar Enhancement improvement Improves an existing functionality non-breaking Introduces a non-breaking change
#2519 opened Aug 28, 2026 by achirkin Contributor Loading…
@achirkin
2
Lucene: define and validate CAGRA parameter boundaries improvement Improves an existing functionality non-breaking Introduces a non-breaking change
#2516 opened Aug 27, 2026 by shaunakkapur Contributor Loading…
1
13
Fix: reduce recall calculation time at small batch
#2513 opened Aug 27, 2026 by sherylll Loading…
1
1
Fix stream ordering for brute-force L2 sqrt post-processing bug Something isn't working non-breaking Introduces a non-breaking change
#2509 opened Aug 26, 2026 by dantegd Contributor Loading…
2
Migrate CAGRA merge() to caller-concatenated dataset + offsets contract breaking Introduces a breaking change Enhancement improvement Improves an existing functionality
#2507 opened Aug 26, 2026 by HowardHuang1 Contributor Loading…
@HowardHuang1
4
Support of BBQ Dataset View C++ feature request New feature or request non-breaking Introduces a non-breaking change
#2506 opened Aug 25, 2026 by lowener Contributor Loading…
1
perf(brute_force): drop the duplicate bitset/bitmap→CSR conversion on the SDDMM path
#2505 opened Aug 25, 2026 by maxwbuckley Contributor Loading…
2
Optimize recall calculation and enhance CAGRA search features
#2502 opened Aug 25, 2026 by sherylll Loading…
6
[REVIEW] Fix padded dataset creation for aligned device tensors
#2501 opened Aug 25, 2026 by Sharon05564 Loading…
1
2
Lucene: Merge CAGRA indexes on the GPU instead of rebuilding from vectors improvement Improves an existing functionality Java Lucene non-breaking Introduces a non-breaking change
#2497 opened Aug 21, 2026 by imotov Contributor Loading…
1
@imotov
5
Rename vpq to pq and use_vq to train_coarse for datasets, factories, and params
#2494 opened Aug 20, 2026 by HowardHuang1 Contributor Loading…
3
Use rtcx ELF embedding for JIT LTO fragments improvement Improves an existing functionality non-breaking Introduces a non-breaking change
#2491 opened Aug 19, 2026 by divyegala Contributor Loading…
ProTip! Exclude everything labeled bug with -label:bug.