See https://github.com/dstackai/gpuhunt/issues/142 Ensure to pass both `cuda_max_good` and `compute_cap`
See dstackai/gpuhunt#142
Ensure to pass both
cuda_max_goodandcompute_cap