slower on GPU using tensorflow backend?

keon / deep-q-learning

Minimal Deep Q Learning (DQN & DDQN) implementations in Keras

https://keon.io/deep-q-learning

MIT License

1.29k stars 455 forks source link

Open rayset opened 5 years ago

rayset commented 5 years ago

I tried several machines, but with CUDA 9 and 10 it seems to be a lot slower on GPU. Why is this the case? Am I the only one getting this behavior?

nickion commented 4 years ago

Given a fast CPU with all cores used, simple models tend to be faster on the CPU. I suspect the overhead of setting up the GPU becomes significant.