search

carpedm20 / BEGAN-tensorflow

Tensorflow implementation of "BEGAN: Boundary Equilibrium Generative Adversarial Networks"
Apache License 2.0
922 stars 241 forks source link

Training on Multi-GPUs #9

Open surya501 opened 7 years ago

surya501 commented 7 years ago

I am pretty new to tensorflow although I have some experience with Keras. If I have two GPUs in my system, would it be possible to accelerate the training? If so, how? Apologies if this is a obvious question.

carpedm20 commented 7 years ago

Yes I think so. But it's quite dirty to support multiple GPUs and I don't have a server with such setting so it's hard to implement it :( But there are some good references for it like:

surya501 commented 7 years ago

Not, a exact solution, but a temp workaround -- which might be obvious to a lot of folks, but still jotting it down here in case it helps anyone. Instead of using multiple GPUs for one training, I went another route... I invoked one session per GPU like this. CUDA_VISIBLE_DEVICES=0 python main.py --dataset=Shoes --use_gpu=True CUDA_VISIBLE_DEVICES=1 python main.py --dataset=CelebA --use_gpu=True