Skip to main content

faiss-gpu-cu12-cuvs

Unofficial Python wheels for FAISS 1.15.0 with CUDA GPU and NVIDIA cuVS support.

This downstream package is built from the official FAISS source through the Di-Is/faiss-gpu-wheels build scaffold. It is not an official FAISS release artifact.

Project repository: https://github.com/Gabrieliam42/FAISS-GPU-CUDA-cuVS

Target

  • Python 3.12
  • Linux x86-64 / WSL2
  • CUDA 12.9 Update 1
  • NVIDIA cuVS 26.6.0
  • Native GPU code for sm_75, sm_80, sm_86, and sm_90

Install

pip install --extra-index-url https://pypi.nvidia.com faiss-gpu-cu12-cuvs==1.15.0

The host requires an NVIDIA driver compatible with CUDA 12.9. The package dynamically loads CUDA and RAPIDS libraries supplied by Python distributions. Its CUDA components are pinned to the versions used by PyTorch 2.9.1 CUDA 12.9. FAISS does not link to cuDNN, so it can coexist with the cuDNN package required by the surrounding PyTorch stack.

The exact PyTorch 2.9.1+cu129 stack requires CUDA 12.9 Update 1 components. The cuda-toolkit==12.9.0 meta-package is not compatible with that stack's exact cuBLAS 12.9.1.4 requirement.

The validation stack does not include xformers. The official PyTorch cu129 index has no xformers 0.0.33.post2 wheel, and its stable 0.0.32.post2 wheel requires Torch 2.8.0 rather than the required Torch 2.9.1.

Download files

Download the file for your platform. If you're not sure which to choose, learn more about installing packages.

Source Distributions

No source distribution files available for this release.See tutorial on generating distribution archives.

Built Distribution

If you're not sure about the file name format, learn more about wheel file names.

faiss_gpu_cu12_cuvs-1.15.0-cp312-cp312-manylinux_2_39_x86_64.whl (91.6 MB view details)

Uploaded CPython 3.12manylinux: glibc 2.39+ x86-64

File details

Details for the file faiss_gpu_cu12_cuvs-1.15.0-cp312-cp312-manylinux_2_39_x86_64.whl.

File metadata

File hashes

Hashes for faiss_gpu_cu12_cuvs-1.15.0-cp312-cp312-manylinux_2_39_x86_64.whl
Algorithm Hash digest
SHA256 20e104ce0754ce53fd5d4bee1036ab8aae2f354f8248fa055adbeeb7a38fc63f
MD5 c5f5990b50d9113580288fa8e122dd1c
BLAKE2b-256 34a94d84bf3898b36b9c0c70e48db7b985fe43c9578aad86f8fb38038f9bdd7e

See more details on using hashes here.

Supported by

AWS Cloud computing and Security Sponsor Datadog Monitoring Depot Continuous Integration Fastly CDN Google Download Analytics Sentry Error logging StatusPage Status page