I have 2 Geforce 8800M gtx in my laptop (Vista), SLI is disable but cudaGetDeviceCount() detects only one graphic card. Could someone help me?
I haven’t had experience with this, but from my understanding, there is a physical connection between 2 gpus when running in SLI. Disabling SLI via software can’t eliminate that and I’m assuming you’re just in fact running 1 GPU. I’m not 100% sure though since I’ve never used SLI.
With the latest SDK (CUDA 2.3), you can supposedly enumerate all GPUs in SLI, so how bout trying to enable SLI and see what comes up? Again, you have to be using 2.3.
Thanks, but… latest driver aren’t aviable for my graphic card (those needed by CUDA 2.3)… I will wait.
If it doesn’t work now, you likely just need a modified INF file.
Take a look at this site, it’s likely you can get 2.3 running with their INF.
So? Does it solve anything? Cause im having practically the same problem (only i have 3 geforces and 1 quadro, not a laptop). cudaGetDeviceCount returns 2 devices (1 quadro and 1 geforce ), so does the deviceQuery. SLI (according to nvidia control panel) is turned off. The thing is when i tried to install the latest driver aviailable for my devices I got “out of range” on my monitor :">