Geek asks and answers

Cannot start CUDA or OpenCL on GeForce 540M

I have problems with running samples provided by Nvidia in their Computing SDK SDK (there is a library of compiled code samples there).

For cuda, I get the message “CUDA-enabled device was not found”, for OpenCL there is an error from a function that should find modules that support OpenCL.

I installed all three parts from Nvidia for development with OpenCL - devdriver for win7 64bit v.301.27, cuda toolkit 4.2.9 and gpu computing sdk 4.2.9.

I think this may be related to Optimus technology, which redirects the output from the Nvidia GPU to Intel for visualization (this laptop also has an Intel 3000HD accelerator), but in the Nvidia control panel I decided to use the high-performance Nvidia GPU, set the power so maximize performance, and PhysX I changed again from automatic selection to an Nvidia processor. However, nothing has changed; these samples will not work (even those designed for GF8000 cards).

I would like to work a little with OpenCL and see what it is capable of, but without the ability to verify that it is useless. I found some information about this in the forums, but mostly for Linux users, where you need a Bumblebee to access the Nvidia GPU. However, there are no such problems on Windows, drivers are better, so you can access it without dark magic (or I thought so until I found this problem).

+3
source share
3 answers

, . , - , , , - AV. Comodo +, . , + .

, .

+2

GeForce 540M, Optimus, Sandy Bridge Intel. CUDA :

  • NVIDIA
  • NVIDIA
  • " 3D" → " 3D" → " "
  • Preferred Graphics " NVIDIA"

, , CUDA. .

-. !

+2

This is most likely due to the Optimus architecture. Therefore, I suggest you read the NVIDIA CUDA Developer's Guide for NVIDIA Optimus platforms , especially the "Request for CUDA device" section, which, in my opinion, solves this problem.

+1
source

More articles:


All Articles