Where do I post bug with OpenCL/Tesla/Linux bug issue using OpenCL on Fedora12

Accelerated Computing CUDA CUDA Programming and Performance
#1

Hi,

I have been looking around for an official place to lodge a bug report, and this may or may not be the place for it.

I am having a problem using __constant memory on Linux Fedora12 x86_64 Tesla C1050 drivers 195.36.15, where as identical code works nicely on WinXP64 with GT9600 (driver 197.45).

It seems to be an issue with clEnqueueCopyBuffer() copying from a larger cl_mem object to another cl_mem object that is the size of the __constant memory on the card.

Should I post code here? Please point me to the the best place for this problem otherwise.

Thanks,

Josh.

#2

As much as I like to try and blame things on drivers, once again, i don’t think I can.
I cut down code to just the function of interest and it seems to work fine.
I’ll keep on bug hunting,
Josh.