-
Notifications
You must be signed in to change notification settings - Fork 199
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
OpenCL fails with 2 fully connected layers #105
Comments
When calling:
Without neural.cl in same directory as the executable: With neural.cl in same directory as the executable: I think that perhaps one of the problem is that my directory 'neural' is not in '../../../neural' but in ../neural |
I just try same program on a Linux Mint 20.2 I add "-dUseCThreads" in Custom Options {$IFDEF UseCThreads} to {$IFDEF UseCThreads} It works, but I don't see any acceleration. |
OpenCL is actually slower in this experiment. I'm wondering if the number of weights/neurons is so small in this experiment that OpenCL has no advantage. |
I don't know why it crashes on Windows after clCreateContext OK! |
I'm about to start working on this. |
On dense (fully connected layers), OpenCL is called only when there is enough neurons/weights to compensate the overhead that it adds:
Depending on how many neurons you have on each layer, maybe its not even in use. |
I've just tested the following and it works for me:
|
I've just tested the following and it also works for me:
and
Given that I can't reproduce, you'll need to share a full Lazarus project source code that provokes the error. |
In the case that it helps, this is how neural.cl is loaded:
|
Hi, Thank you very much for your tests :) You have to change the path of /neural and add neural.cl in the same directory as the executable. Just train (500-100 epoch) and test B-> |
I'm having problem to adapt one of my program using CAI with OpenCL
I've tested "SimpleImageClassifierGPU" and it's working on my computer (removing the option -dAVX, because my CPU is old)
When I try to add OpenCL in my test program with 2 fully connected layers (no convolution, it fails).
The text was updated successfully, but these errors were encountered: