CUDA doesn't work on new Dell Precision w/ Tesla C1060

Accelerated Computing CUDA CUDA Programming and Performance
1

Hi there, first post. My company has gotten their first Tesla machine. It’s a Dell Precision with 2 XEON, 12GB RAM, Quadro card, and a Tesla C1060 card. The latest drivers have been installed. CUDA 3.1 also has been installed. I have tried to run the devicequery sample but I cannot get past the first cuda function call. cudaGetDeviceCount fails.

I believe we have installed the 257.21 display driver on Windows 7 64bit. Unfortunately I cannot run the NVidia control center since I’m remoting into the machine.

I’m running a little low on what else I could try. The Tesla card and quadro card both show up in the device manager. I tried to run devicequery in admin mode but nothing works.

Any help would be appreciated!

Thanks,
Christian

2

Small Update. We have just installed the 258.96-Tesla-win7-winvista-64bit-international-whql.exe driver. Nothing has changed. When stepping over cudaGetDeviceCount inside the debugger I get the following messages in my output window:

First-chance exception at 0x000007fefd6caa7d in deviceQuery.exe: Microsoft C++ exception: cudaError_enum at memory location 0x0012f878…
First-chance exception at 0x000007fefd6caa7d in deviceQuery.exe: Microsoft C++ exception: cudaError at memory location 0x0012f894…

Any idea?

Christian

3

Small Update. We have just installed the 258.96-Tesla-win7-winvista-64bit-international-whql.exe driver. Nothing has changed. When stepping over cudaGetDeviceCount inside the debugger I get the following messages in my output window:

First-chance exception at 0x000007fefd6caa7d in deviceQuery.exe: Microsoft C++ exception: cudaError_enum at memory location 0x0012f878…
First-chance exception at 0x000007fefd6caa7d in deviceQuery.exe: Microsoft C++ exception: cudaError at memory location 0x0012f894…

Any idea?

Christian

4

Seems like I cannot use MS Remote Desktop. When I try VNC it works. The only confusing thing is that I can run some graphic demos using Remote Desktop but I cannot start NVidia control center nor any cuda applications. Weird.

Did anyone find a way around this issue. Not sure if VNC is a viable solution. It’s very slow.

Thanks,
Christian

5

Seems like I cannot use MS Remote Desktop. When I try VNC it works. The only confusing thing is that I can run some graphic demos using Remote Desktop but I cannot start NVidia control center nor any cuda applications. Weird.

Did anyone find a way around this issue. Not sure if VNC is a viable solution. It’s very slow.

Thanks,
Christian

6

If I recall correctly, when you use MS Remote Desktop the video driver is not loaded. In this case, that means you CUDA/nVidia driver is not loaded.

7

If I recall correctly, when you use MS Remote Desktop the video driver is not loaded. In this case, that means you CUDA/nVidia driver is not loaded.

8

CUDA 3.2 has major improvements for Tesla cards and Remote Desktop.

9

CUDA 3.2 has major improvements for Tesla cards and Remote Desktop.