CUDA D870 setup problem on CentOS 5

Hi,

I recently got D870 but I am unable to set it up correctly to use it. I installed the system with CentOS 5.0 x86_64. I installed the latest driver from the nVidia web site. I did everything as per the install guide except the nvidia-xconfig. My primary display is on board ATI ES1000. So if I do nvidia-xconfig, it will mess up my X config.

deviceQuery gives the following error:

NVIDIA: could not open the device file /dev/nvidiactl (No such file or directory).
There is no device supporting CUDA.

Excerpt from lspci:
01:04.0 VGA compatible controller: ATI Technologies Inc ES1000 (rev 02)
02:00.0 PCI bridge: nVidia Corporation Unknown device 01b3 (rev a3)
03:00.0 PCI bridge: nVidia Corporation Unknown device 01b3 (rev a3)
03:01.0 PCI bridge: nVidia Corporation Unknown device 01b3 (rev a3)
05:00.0 PCI bridge: nVidia Corporation Unknown device 01b3 (rev a3)
06:00.0 PCI bridge: nVidia Corporation Unknown device 01b3 (rev a3)
06:01.0 PCI bridge: nVidia Corporation Unknown device 01b3 (rev a3)
07:00.0 3D controller: nVidia Corporation Unknown device 0197 (rev a2)
08:00.0 3D controller: nVidia Corporation Unknown device 0197 (rev a2)

Here is an excerpt from dmesg:
nvidia: module license ‘NVIDIA’ taints kernel.
PCI: Setting latency timer of device 0000:07:00.0 to 64
PCI: Setting latency timer of device 0000:08:00.0 to 64
NVRM: loading NVIDIA UNIX x86_64 Kernel Module 174.55 Thu Mar 27 19:12:02 PST 2008
PCI: Setting latency timer of device 0000:07:00.0 to 64
PCI: Setting latency timer of device 0000:08:00.0 to 64
NVRM: loading NVIDIA UNIX x86_64 Kernel Module 174.55 Thu Mar 27 19:12:02 PST 2008

excerpt from /var/log/messages:
May 22 17:56:14 compute-0-15 kernel: Memory for crash kernel (0x0 to 0x0) notwithin permissible range
May 22 17:56:14 compute-0-15 kernel: PCI: Failed to allocate mem resource #6:20000@b0000000 for 0000:07:00.0
May 22 17:56:14 compute-0-15 kernel: PCI: Failed to allocate mem resource #6:20000@c0000000 for 0000:08:00.0

Any help is sincerely appreciated.
My config:
D870/x16 card
Dual Opteron Quadcore
Mellanox Infiniband InfiniHost III LX HCA single port
OFED stack
CentOS 5.0 x86_64

Thanks,
Kalyan

You need to create those device nodes yourself since you aren’t loading the nvidia drivers into X. See the script in the release notes: [url=“http://developer.download.nvidia.com/compute/cuda/1_1/CUDA_Toolkit_Release_Notes.txt”]http://developer.download.nvidia.com/compu...lease_Notes.txt[/url]

You could also use the runlevel version of the script posted by jimh: [url=“The Official NVIDIA Forums | NVIDIA”]The Official NVIDIA Forums | NVIDIA

Awesome! Using the runlevel script, everything works fine. Thank you guys.