Hello everyone,
Am trying to tinker with CUDA on a new MacBook Pro machine. These things ship with Intel HD graphics & nvidia GT 330M graphics.
It seems that you need to force the system to use the nvidia chip before CUDA will work. This thread helped me figure that out. You can do that in System Preferences->Energy Saver-> Uncheck “Automatic graphics switching” OR with a program like the one here: http://codykrieger.com/gfxCardStatus/
Example, with machine running on Intel graphics:
$ pwd
/Developer/GPU Computing/C/bin/darwin/release
$ $ ./deviceQuery
./deviceQuery Starting...
CUDA Device Query (Runtime API) version (CUDART static linking)
cudaGetDeviceCount FAILED CUDA Driver and Runtime version may be mismatched.
FAILED
Press <Enter> to Quit...
-----------------------------------------------------------
On GT 330M:
argon:release joel$ ./deviceQuery
./deviceQuery Starting...
CUDA Device Query (Runtime API) version (CUDART static linking)
There is 1 device supporting CUDA
Device 0: "GeForce GT 330M"
CUDA Driver Version: 3.10
CUDA Runtime Version: 3.10
...snip...
When I have some number-crunching program running full-blast on the graphics chip, system responsiveness suffers; the system is using the nvidia card to draw AND run my program. Has anyone figured out a way to make the system use the Intel chip (for graphics) while having the nvidia chip enabled (solely for CUDA use)? I’m trying to find a way.
I’ll keep playing with it and post my results, if any.