GPU memory exhaustion
RobinVanSchendel opened this issue · comments
I really appreciate this software and the effort you put in maintaining it. I am now able to run localcolabfold on our GPU and it works for most combinations of proteins. For one rather large one I get this message:
Not Enough GPU memory? RESOURCE_EXHAUSTED: Failed to allocate request for 35.58GiB (38204550232B) on device ordinal 0
Is there a way to specify the maximum GPU memory a run should take?
I am trying to fold some larger proteins (>1500 aa) on my laptop with RTX 3060 GPU 6Gb. Is this just not possible, and I need to use different hardware, or is there a way?