Just got a fancy new 3060 w/12GB RAM, and everything's working fine - but I can't do two things at the same time.
If I run a training process, then (in another tab) start the GUI to work on cleaning up some alignment files (Tools -> Manual), the second task crashes with tensorflow.python.framework.errors_impl.ResourceExhaustedError.
If I do the opposite - start the GUI work first, then start training job via CLI - same thing. The second task I start fails to run.
Is this normal? Am I only supposed to have one thing using my CUDA cores at a time?
I tried playing with the more obvious options in config/*.ini, like allow_growth and convert_batchsize, but it made no difference.