search

3dem / model-angelo

Automatic atomic model building program for cryo-EM maps
MIT License
116 stars 18 forks source link

RuntimeError: CUDA out of memory. Can it be avoided, at least for tests #15

Closed pconesa closed 1 year ago

pconesa commented 1 year ago

I'm developing on a laptop...so my GPU is under specs.

I wonder if there is a way to prevent out of memory errors, event if that sacrifices results. I'm now interested it running it and registering the output and so on and less in getting the best results possible.

Error I'm getting is:

RuntimeError: CUDA out of memory. Tried to allocate 1024.00 MiB (GPU 0; 5.94 GiB total capacity; 3.90 GiB already allocated; 457.88 MiB free; 4.86 GiB reserved in total by PyTorch) If reserved memory is >> allocated memory try

Volume is a 200x200x200. Runinng without sequence.

jamaliki commented 1 year ago

Would restricting it to the CPU work? You can do that by adding the flag "--device cpu" to your command.

pconesa commented 1 year ago

Yes, I tried and it works! thanks.