Files
nixos-config/pkgs/openwhispr
Miguel Palhas a7c74d5a9f
CI / lint (push) Successful in 26s
CI / eval (push) Successful in 1m48s
fix(openwhispr): target CUDA 12 for the GPU whisper sidecar
The previous commit wired CUDA 13 against the bundled onnxruntime provider,
which only accelerates diarization/embeddings — transcription still fell back
to CPU with "GPU could not be activated". The GPU whisper-server is fetched at
runtime and links CUDA 12 SONAMEs, so it needs cudaPackages_12 on
LD_LIBRARY_PATH (plus libstdc++/libgomp, which it also could not resolve).

Verified: whisper-server now reports "found 1 CUDA devices ... RTX 4060" and
loads the model on the CUDA0 backend. Drops the closure 4.6 -> 3.7 GiB.
2026-09-07 16:52:11 +01:00
..