PDA

View Full Version : Experience with NVidia Optimus and OpenGL ?



Nicolas Lelong
04-23-2011, 01:46 AM
Hi everyone,

I happened to have my hands for the first time on an NVidia Optimus 'enabled' laptop yesterday.

It happened that our OpenGL applications did not automatically trigger the switch to NVidia 540M GPU, and tried to run (without success, as expected) on the Intel integrated GPU. We had to force the graphics device to the NVidia board to have things running propely.

I had a look at 'CUDA Developer Guide for Optimus Platforms' (http://developer.download.nvidia.com/compute/cuda/3_2/toolkit/docs/CUDA_Developer_Guide_for_Optimus_Platforms.pdf) that mentions this problem. I thought about having a look at NVAPI but the download links are broken ATM.

AFAIK, the 'official way' would be to create an application profile in the NVidia driver, but this is cumbersome.

Is there some way using NVAPI to force the graphics device selection at application startup ? I thought about creating a CUDA context before the OpenGL context to 'tag' the process as being 'GPU hungry', but I could not test it.

Does any of you have experience or references about controlling Optimus behaviour programmatically ?

Cheers,

overlay
04-23-2011, 09:37 AM
Optimus is such a failure. Shame on nVidia to not even providing a driver for Linux.