how to interprete CL_NONE value

Accelerated Computing CUDA CUDA Programming and Performance
1

Hi, I’m pretty new to openCL and i’d like to ask (surely) stupid question

What is the way to interprete CL_NONE value returned by clGetDeviceInfo when querying for CL_DEVICE_GLOBAL_MEM_CACHE_TYPE ??
does it mean that my GF 8600 GT does not have any cahce memory or it is not availible for opencl??

2

For that card, both. What might be confusing you & me as well, is that even on Fermi where the value is CL_READ_WRITE_CACHE & cache line size is 128, Global Mem Cache Size returns the same value as Global Mem Size. This makes little sense.

Edit: Checking what the OSX Intel CPU reports, it also reports the same #. 3072 MB. WTF??

3

My bad, in code copying for a little report I wrote, that was me. I now show 245760 for GTX 480.

4

thanks, it was really helpfull :)