Hi,
with tensorflow I can set a limit to gpu usage, so that I can use 50% of gpu and my co-workers (or myself on another notebook) can use 50%
I just have to do this:
config = tf.ConfigProto(gpu_options=tf.GPUOptions(per_process_gpu_memory_fraction=0.7))
sess = tf.InteractiveSession(config=config)
Do you know how to do this with pytorch ?
Thanks
3 Likes
smth
September 11, 2017, 1:44pm
2
there is no way to do this in pytorch. However, PyTorch doesn’t pre-occupy the GPU’s entire memory, so if your computation only uses 50% of GPU, only that much is locked by PyTorch
2 Likes
simonp
(Simon P.)
April 26, 2021, 5:05pm
4
In case someone still stumbles upon this question: With PyTorch 1.8, there is set_per_process_memory_fraction
1 Like