pirosb3 / cudaneuralnetworks Goto Github PK

Hi, thanks for a great code! I want to ask what is the speedup using cuda for training? And how to define the batch size? is there any difference between using device and global? I am a new beginner of cuda, can you help answer my questions?

Hi,
I have tried to increase the numbers of input node (i.e. 10,000 nodes) and numbers of iteration for training (i.e. 2,000). The CUDA freezes every time at the same point of iteration around 1,800. Then I have to restart the computer to reset the CUDA. Have you seen something like this before? Currently I'm using GTX590 on Windows 7.

Thanks,
Mono