Lucas Wilkinsonandkhluu e4c0211529 [Build] Download nvcc from NVIDIA if system CUDA < 12.9 ...

- Add nvcc toolchain override in setup.py to download nvcc 12.9.86
  from NVIDIA's redistribution server when system CUDA < 12.9
- FlashMLA requires nvcc >= 12.9 for compilation
- Use CUDA 12.8.1 base image to match PyTorch cu128 wheels
- Update pre-commit and test.txt to use cu128 torch backend
- Cache downloaded nvcc in .deps/nvcc/ for reuse
- Set VLLM_SKIP_NVCC_OVERRIDE=1 to disable

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>

Signed-off-by: Lucas Wilkinson <lwilkins@redhat.com>
(cherry picked from commit d7276b81542f4086f31ad64f6483962fcd246507)

2026-02-18 15:46:29 -08:00

..

docker-bake.hcl

Add docker buildx bake configuration (#31477)

2025-12-31 01:08:54 +00:00

Dockerfile

[Build] Download nvcc from NVIDIA if system CUDA < 12.9

2026-02-18 15:46:29 -08:00

Dockerfile.cpu

[XPU][9/N] clean up existing ipex code/doc (#34111)

2026-02-11 02:33:27 -08:00

Dockerfile.nightly_torch

Fix RoutingMethodType logic (#33919)

2026-02-06 14:03:34 -08:00

Dockerfile.ppc64le

Fix ijson build for Power. (#31702)

2026-01-08 17:12:33 +00:00

Dockerfile.rocm

[rocm][ray] Fix: Unify Ray device visibility handling across CUDA and ROCm (#33308)

2026-02-04 10:09:14 -08:00

Dockerfile.rocm_base

[ROCm] [CI] [Release] Rocm wheel pipeline with sccache (#32264)

2026-01-16 02:56:18 +08:00

Dockerfile.s390x

[CPU][IBM Z][Dockerfile] Fix IBM Z builds (#33243)

2026-02-01 23:41:29 -08:00

Dockerfile.tpu

Always use cache mounts when installing vllm to avoid populating pip cache in the image. Also remove apt cache. (#23270)

2025-08-21 18:01:03 -04:00

Dockerfile.xpu

[XPU]Replace pip in docker.xpu with uv pip (#31112)

2026-02-06 14:02:33 +08:00

versions.json

[Build] Download nvcc from NVIDIA if system CUDA < 12.9

2026-02-18 15:46:29 -08:00