How to make the code run default on GPU?

[aasharma90]

Hi @Skylion007

I'm observing that the default trainer function SeparateGANTrainer is running on CPU, which is quite slow. Is there a way to run it on multi-GPU?

Although, I see that you've provided GPU supported trainers such as MultiGPUGANTrainer, but I'm wondering why these functions do not support the 2:1 training strategy you are using in SeparateGANTrainer. Is there a wayout?

Also, please let me know if I'm misunderstanding anything. Thanks!

[Skylion007]

SeperateGANTrainer should only really be used on GPU. Tensorpack/Tensorflow will always default to the GPU. Did you make sure you installed the proper gpu enabled version of Tensorflow? In fact, I am surprised it doesn't error when trying to run it on CPU.

[aasharma90]

I also thought so, but running the code, I found it to be running on CPU by default. I tried it with 1.14.0 version of both tensorflow and tensorflow-gpu, and 0.8.9 version of tensorpack.

[Skylion007]

That's odd, even the normal GAN trainer shouldn't run in CPU mode. You can try manually overriding device placement, but it really does sound like a configuration issue (with Tensorflow being unable to see / load the GPUs or such).

…

On Thu, Jun 11, 2020 at 2:11 AM Aashish Sharma ***@***.***> wrote: I also thought so, but running the code, I found it to be running on CPU by default. I tried it with 1.14.0 version of both tensorflow and tensorflow-gpu, and 0.8.9 version of tensorpack. — You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub <#22 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AAPVMX6FAKNUGFJWG2GE6W3RWBYPVANCNFSM4N2VHXUQ> .