I have been trying to get this to work to no avail.
The last thing was to downgrade to 3.2 version with a litle bit more success but the CUDA calculations are failing. If I use the NVIDIA board as graphic output, the screen comes up all mangle. If I use the main board (ATI) as graphic output, the screen is ok with both version of the ATI driver ( nominal and FireGL). I got the same bad calculations in both cases. Even the cppIntegration program is indicating FAILED.
Is there a conflict possible if the main board has an ATI chipset? I am beginning to think so.
My setup is :
MotherBoard: ASUS M4A78LT-M
Graphic Board : Geforce GTX 260 or Geforce 9600 GSO
PSU : ULTRA LS 600
Memory : 8 GB
OS : Ubuntu 10.04 LTS
Anyone with the same configuration?
Edit: I did a test with my old config with Slackware which used to work with the MSI board and does not work with the ASUS board : conclusion = bad board for this application.
ATI chipset == NorthBridge AMD790
This is an ouput from lspci:
01:05.0 VGA compatible controller: ATI Technologies Inc 760G [Radeon 3000] Subsystem: ASUSTeK Computer Inc. Device 8388 Flags: bus master, fast devsel, latency 0, IRQ 27 Memory at c0000000 (32-bit, prefetchable) I/O ports at c000 Memory at f9ff0000 (32-bit, non-prefetchable) Memory at f9e00000 (32-bit, non-prefetchable) Capabilities:  Power Management version 3 Capabilities: [a0] Message Signalled Interrupts: Mask- 64bit+ Queue=0/0 Enable+ Kernel driver in use: fglrx_pci Kernel modules: fglrx, radeon 02:00.0 VGA compatible controller: nVidia Corporation GT200 [GeForce GTX 260] (rev a1) Subsystem: XFX Pine Group Inc. Device 2393 Flags: bus master, fast devsel, latency 0, IRQ 18 Memory at fd000000 (32-bit, non-prefetchable) Memory at d0000000 (64-bit, prefetchable) Memory at fa000000 (64-bit, non-prefetchable) I/O ports at dc00 [virtual] Expansion ROM at fea80000 [disabled] Capabilities:  Power Management version 3 Capabilities:  Message Signalled Interrupts: Mask- 64bit+ Queue=0/0 Enable- Capabilities:  Express Endpoint, MSI 00 Capabilities:  Virtual Channel <?> Capabilities:  Power Budgeting <?> Capabilities:  Vendor Specific Information <?> Kernel driver in use: nvidia Kernel modules: nvidia, nvidia-current, nvidiafb, nouveau
Any chance of getting the standalone CUDA card to operate correctly in this environnment?
Would a newer CUDA card work? GTX 5xx ?
I just gave up and order a replacement board with and nvidia chipset.
Finally got a new board M4N68T-M V2 with nvidia chipset nForce 630a.
Got it to work as a display + CUDA.
Still no luck as a standalone CUDA using the internal graphic for display,which is the configuration I had before.