[GH-ISSUE #2017] torch.cuda.OutOfMemoryError: CUDA out of memory. #1358

New issue

Closed

opened 2026-10-20 10:08:25 +00:00 by bsakaguchi · 1 comment

bsakaguchi commented

2026-10-20 10:08:25 +00:00

Owner

Originally created by @amink69 on GitHub (Apr 24, 2024).
Original GitHub issue: https://github.com/RVC-Project/Retrieval-based-Voice-Conversion-WebUI/issues/2017

hi
i have nvidia 3050ti GPU wth amd ryzen 7 5800H
but every time i want to train a model i got an error :
torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 20.00 MiB (GPU 0; 4.00 GiB total capacity; 3.36 GiB already allocated; 0 bytes free; 3.44 GiB reserved in total by PyTorch) If reserved memory is >> allocated memory try setting max_split_size_mb to avoid fragmentation. See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF

i use the latest python
my windows is 11
and there is no problem
but i can't train

i use lowest settings but i saw this error every single time
i google for this and search in github but i didn't find any solution
i set batch size 1 and highest cpu and bla bla bla
but every time i got error like this
please help me
thanks

Originally created by @amink69 on GitHub (Apr 24, 2024). Original GitHub issue: https://github.com/RVC-Project/Retrieval-based-Voice-Conversion-WebUI/issues/2017 hi i have nvidia 3050ti GPU wth amd ryzen 7 5800H but every time i want to train a model i got an error : torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 20.00 MiB (GPU 0; 4.00 GiB total capacity; 3.36 GiB already allocated; 0 bytes free; 3.44 GiB reserved in total by PyTorch) If reserved memory is >> allocated memory try setting max_split_size_mb to avoid fragmentation. See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF i use the latest python my windows is 11 and there is no problem but i can't train i use lowest settings but i saw this error every single time i google for this and search in github but i didn't find any solution i set batch size 1 and highest cpu and bla bla bla but every time i got error like this please help me thanks

bsakaguchi

2026-10-20 10:08:25 +00:00

closed this issue
added the
help wanted
label

bsakaguchi commented

2026-10-20 10:08:26 +00:00

Author

Owner

@mathewpan2 commented on GitHub (Apr 29, 2024):

If you're getting Cuda out of memory it's because you just don't have enough vram to train. If even setting batch size to 1 still causes the error, there's not much you can do unless you upgrade. 4gb seems to be the shaky point where getting this error is very common.

@mathewpan2 commented on GitHub (Apr 29, 2024): If you're getting ```Cuda out of memory``` it's because you just don't have enough vram to train. If even setting batch size to 1 still causes the error, there's not much you can do unless you upgrade. 4gb seems to be the shaky point where getting this error is very common.