search

FreshAirTonight / af2complex

Predicting direct protein-protein interactions with AlphaFold deep learning neural network models.
146 stars 19 forks source link

Specifying the GPU id on a machine with more than one gpu #1

Closed avilella closed 2 years ago

avilella commented 2 years ago

Hi,

I am attempting to run the following on a Linux machine with more than one NVIDIA GPU. Only one of the GPUs is of a large memory size, and it's not the one in GPU id=0, which is where the job below is being sent:

python3.8 -u $af_dir/run_af2c_mod.py --target_lst_path=~/af2complex/example/PET102/test.lst   --data_dir=$DATA_DIR --output_dir=$out_dir   --feature_dir=~/af2complex/example/PET102   --model_names=$model   --preset=$preset   --model_preset=$model_preset   --save_recycled=$recycling_setting

Is there a way to specify the GPU id when running af2c? Thanks in advance.

FreshAirTonight commented 2 years ago

Please try

CUDA_VISIBLE_DEVICES=<gpu_id> python -u python3.8 -u $af_dir/run_af2c_mod.py ...