does the Neural network created on GPU?

[nano1900]

what i mean is ,say you created a network and when you traning it, does the network created on GPU(during a forward pass and backward pass all just runing on the GPU , GPU device do not communicate with host)or the network created on the host
and used the GPU as just a compter device

[hughperkins]

The second. Cpu sends commands to the gpu. The data stays on the gpu in between layers, but the command to run the next layer comes from the cpu. Nuance: the program to run for each layer is called a "kernel", and is sent to the gpu driver only once.

…

On Fri, Mar 30, 2018, 05:55 nano1900 ***@***.***> wrote: what i mean is ,say you created a network and when you traning it, does the network created on GPU(during a forward pass and backward pass all just runing on the GPU , GPU device do not communicate with host)or the network created on the host and used the GPU as just a compter device — You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub <#146>, or mute the thread <https://github.com/notifications/unsubscribe-auth/AAHiqPzThZxK_k1hCqvjDFYlmO1sTawYks5tjgEFgaJpZM4TBhaU> .

[nano1900]

oh! i see.The data(between layers) stay on the GPU , Host cpu send "kernels" to the GPU and the GPU operate these data

Thank you!

[hughperkins]

:)