itayhubara / BinaryNet

Training Deep Neural Networks with Weights and Activations Constrained to +1 or -1
298 stars 96 forks source link

About Multi-GPU Usage #1

Closed airjimmy89 closed 8 years ago

airjimmy89 commented 8 years ago

Hi,

I am trying to run the binarized MNIST case with only one GPU, since if I use all three GPUs on my server it might cause a "GPU lost" problem after training a while...(haven't figured out why yet :(..)

However, though the default GPU setting is already for only one single GPU, I can still see from nvidia-smi that all of the three GPUs are used...

I also tried to set -nGPU to 2, but still the same result...

Could anyone help?