I need to disable the multi-gpu mode (to run CUDA application), Is there a way to do so from the API?
The NVAPI function NvAPI_SetGpuTopologies is used to enable and disable multi GPU mode (a.k.a. SLI)
This function is available in the NDA version of the NVAPI SDK - please contact your account representative for more information.
I’m running 8 videocards and have probably SLI enabled. I get out-of-VRAM messages.
So, the best bet is to disable the multi-gpu mode (to run CUDA application), Is there a way to do so remotely?
When Accessing the NVIDIA Control Panel, I get the error message, that the NVIDIA Display settings are not available, because there is currently no display attached to an NVIDIA GPU - that is the case, it is a GPU-Rack with no monitor attached.
But, I need to deactivate the multi-gpu mode/SLI. How can this be done? It would be great to access the current status and get this displayed upfront. To see if it enabled, or disabled. I think this information could be seen, if the display settings could be accessed.