Users can now execute fitting on cuda-enabled devices by settings this in the config:
from pyhdx.config import cfg
cfg.set('fitting', 'device', 'cude')
And dtypes:
from pyhdx.config import cfg
cfg.set('fitting', 'dtype', 'float32')
Supported dtypes are float32 and float64
Switching to cuda is only economical for (very) large datasets (Total ~5000 residues, exact tradeoff point not tested and depends on specific hardware use)
Users can now execute fitting on cuda-enabled devices by settings this in the config:
And dtypes:
Supported dtypes are
float32
andfloat64
Switching to cuda is only economical for (very) large datasets (Total ~5000 residues, exact tradeoff point not tested and depends on specific hardware use)