The effect of using a GPU does not seem to be good.

Kismuz / btgym

Scalable, event-driven, deep-learning-friendly backtesting library

https://kismuz.github.io/btgym/

GNU Lesser General Public License v3.0

988 stars 261 forks source link

The effect of using a GPU does not seem to be good. #48

Closed sharpwood closed 6 years ago

sharpwood commented 6 years ago

Very great project, thank you for your pay. In the example of unreal_stacked_lstm_strat_4_11, is it faster to use GPU computing than CPU? The effect of using a GPU does not seem to be good.

Kismuz commented 6 years ago

@sharpwood, see comments at #26

sharpwood commented 6 years ago

I also saw this issue, but this is talk about A3C. Isn't it better to run LSTM with a GPU? I have a NVIDIA GTX 1060 6G graphics card and an AMD AMD Ryzen 1700 8-core CPU. Is it better to use a CPU to run btgym?

Kismuz commented 6 years ago

There is a lot of traffic and other processes running so yes, It is better to use multi-core CPU to run included algorithms; one should tweak AAC framework to avoid bottlenecks and exploit GPU power.