Framework not working on 3000 series GPUs? #154
Description
We are trying to use this framework for a research project, but whenever we try to run the Windows branch*,
it takes approximately 7 hours on an RTX 3700 and 1 hour with 2x RTX 3900.
We tried the same branch on a Geforce 1070, where it took a total of approximately 5 minutes.
Has anyone else experienced something similar on Linux or Windows? Any tips on fixing this?
Could it be that CUDA might need an update or anything like that?
*#61