I’m having trouble using OpenCL with my Nvidia-enabled system. First, here are the particulars for my setup:
-Hardware: Dell Precision M6500 with an Nvidia Quadro FX 3800M video card.
-OS: Centos 6.4 running the 2.6.32-358.14.1.el6.x86_64 kernel
-Video Driver: Uninstalled and blacklisted Nouveau driver and installed the Nvidia version 304.54 driver. Working and stable.
-SDKs: Installed the CUDA 5.0 SDK. Code samples built and running. Intel OpenCL SDK installed and running.
-CUDA OpenCL Samples: I found an archive of the 4.2 SDK version that had OpenCL samples that I’ve successfully built. I’m also using the samples from “OpenCL Programming Guide” by Munshi, et. al.
The problem I’m having is that I can’t seem to use the GPU from any of my code samples. e.g. The cuda sdk samples all terminate at run time with the following message:
WARNING: NVIDIA OpenCL platform not found - defaulting to first platform!
Running the other samples (from the programming guide) also fails to find the nvidia platform.
As a further check, the contents of the /etc/OpenCL/vendors/ looks sane:
/etc/OpenCL/vendors/nvidia.icd
/etc/OpenCL/vendors/intel64.icd
And nvidia.icd points to the correct (and installed) shared library: libnvidia-opencl.so.1
So, I’m stumped. Any idea what’s going on? My ultimate objective is really just to get a simple hello world type OpenCl application working with the Nvidia GPU.
Thanks in advance for any help.