cancel
Showing results for 
Search instead for 
Did you mean: 

Archives Discussions

gbilotta
Adept III

OpenCL issues with 13.1 on Debian

Hello all, I recently tried upgrading the AMD OpenCL platform on my Debian unstable (which carries stuff from Catalyst 12.6) to the 13.1 stuff from Debian experimental. I noticed two things:

(1) the clinfo command segfaults while printing the information of the Intel platform (I have AMD, Intel, and NVIDIA platforms on this machine). This also happens if I keep the OpenCL platform from 12.6, so it seems specific to clinfo from 13.1

(2) the AMD platform from 13.1 expects to find fglrx. I have no AMD cards on this specific machine, so I only use the AMD platform for CPU. I get these error messages:

Setting of real/effective user Id to 0/0 failed
FATAL: Module fglrx not found.
Error! Fail to load fglrx kernel module! Maybe you can switch to root user to load kernel module directly

While I understand the opportunity of trying to load the fglrx module if it's not loaded, the console messages (which get printed at every run) are quite a bother. Some way to suppress the message, or even set the amd platform to not look for gpus, would be nice.

0 Likes
3 Replies
gbilotta
Adept III

I just noticed that both the issues have been reported, discussed and (partially) solved in other threads. Sorry for not noticing this before.

0 Likes

Yes, clinfo will segfault if you have OpenCL platforms with different versions (like 1.1 and 1.2 together).

You can temporarily workaround by just moving the Intel or whatever1.1 ICD file (usually in /etc/OpenCL/vendors/ directory) out.

As far as the console messages, it has been reported and will be solved.

Thanks,

This should be fixed in next SDK Release.