#Using gpu causes freeze

1 messages · Page 1 of 1 (latest)

stark totem
#

I cannot use my rtx 3060 when loading a model, completely hangs when doing so. Happens right after typedstorage is deprecated warning pops up.

fiery fog
stark totem
#

No error at all, just goes on about how to use the typedstorage crap.

#
2025-04-22 01:18:47,204 INFO     [EmbedderManager] Loading embedder hubert_base
2025-04-22 01:18:48,001 INFO     [PitchExtractorManager] Loading pitch extractor crepe_full_onnx
2025-04-22 01:18:48,062 INFO     [PipelineGenerator] Loading index...
2025-04-22 01:18:48,062 INFO     [PipelineGenerator] Try loading "model_dir/0/added_IVF1732_Flat_nprobe_1_KORONE_v2.index"...
/home/unknownu/voice-changer/server/venv/lib/python3.10/site-packages/faiss/contrib/torch_utils.py:51: UserWarning: TypedStorage is deprecated. It will be removed in the future and UntypedStorage will be the only storage class. This should only matter to you if you are using storages directly.  To access UntypedStorage directly, use tensor.untyped_storage() instead of tensor.storage()
  x.storage().data_ptr() + x.storage_offset() * 4)```
fiery fog
stark totem
#

Hangs on gpu and doesn't use it.

fiery fog
#

then restart it GatorHUHH

stark totem
#

Rehangs till I delete the config.

fiery fog
#

delete stored_setting.json then try again

#

also try loading another model then

stark totem
#

That I tried too, same issue with the other model.

fiery fog
#

try (re)installing CUDA toolkit or the GPU driver (either the latest or Dec 2024 version)

#

the last resort is to fresh install windows and everything if ur cooked

sour sierra
#

If running with CPU is fine, but the issue happens with GPU, it could indicate that either your GPU is failing. Or you're running the older version of fork W-Okada.

stark totem
#

Forgot to mention I'm using arch linux. I wonder if my drivers are the problem?

sour sierra
#

That could be the reason why.

fiery fog