I am trying to update the CUDA driver. I am a newbie to both Terminal commands and to updating CUDA drivers, so please bear with me.
I am running Mountain Lion (10.8.2) on a dual quad core Mac Pro (mid 2010 model).
I have an NVIDIA GeForce GTX 570 card for speeding up video processing (Davinci Resolve, FCPX, etc.).
(FYI, the card is installed in a Cubix Desktop 4 PCI expander).
I have been following the guide here:
http://docs.nvidia.com/cuda/cuda-getting-started-guide-for-mac-os-x/index.html
To install gcc w/o having to install Xcode (I don’t need all of Xcode and so don’t want it installed), I downloaded the “Command Line Tools (OS X Mountain Lion) for Xcode – November 2012” from the Apple Developer Downloads page:
xcode452cltools10_86938211a.dmg
I also downloaded the “Mac OS X CUDA 5 Production Release” from the NVIDIA Developers Download page:
cuda_5.0.36_macos.pkg
I installed the Command Line Tools, and then installed the CUDA 5 Release (driver, toolkit, samples). Both installed successfully.
To verify that gcc was installed, I entered the Terminal command:
/usr/bin/gcc –-help
The resulting message was:
i686-apple-darwin11-llvm-gcc-4.2: –-help: No such file or directory
i686-apple-darwin11-llvm-gcc-4.2: no input files
I take it this means I did not successfully install gcc.
Just to see what happens, I went ahead with the following verification steps:
I went to “Define the environment variables”.
I entered the command
export PATH=/Developer/NVIDIA/CUDA-5.0/bin:$PATH
I got no message
I then entered the command:
export DYLD_LIBRARY_PATH=/Developer/NVIDIA/CUDA-5.0/lib:$DYLD_LIBRARY_PATH
I got no message
To verify that the CUDA kernel extension is loaded, I then entered the command:
kextstat | grep -i cuda
I got this message:
123 0 0xffffff7f81e69000 0x2000 0x2000 com.nvidia.CUDA (1.1.0)
I then checked the CUDA Toolkit version by entering the command:
nvcc –V
I got the message:
nvcc: NVIDIA (R) Cuda compiler driver
Copyright (c) 2005-2012 NVIDIA Corporation
Built on Fri_Sep_28_16:10:16_PDT_2012
Cuda compilation tools, release 5.0, V0.2.1221
To change the directory, I then entered:
cd /Developer/NVIDIA/CUDA-5.0/samples
Then to compile the samples, I entered:
make
The resulting message was:
make[1]: Nothing to be done for all'. make[1]: Nothing to be done for all’.
make[1]: Nothing to be done for all'. make[1]: Nothing to be done for all’.
make[1]: Nothing to be done for all'. make[1]: Nothing to be done for all’.
make[1]: Nothing to be done for all'. make[1]: Nothing to be done for all’.
make[1]: Nothing to be done for all'. make[1]: Nothing to be done for all’.
make[1]: Nothing to be done for all'. make[1]: Nothing to be done for all’.
make[1]: Nothing to be done for all'. make[1]: Nothing to be done for all’.
make[1]: Nothing to be done for all'. make[1]: Nothing to be done for all’.
make[1]: Nothing to be done for `all’.
Makefile:79: *** MPI not found, not building simpleMPI… Stop.
make: *** [0_Simple/simpleMPI/Makefile.ph_build] Error 2
To run the binaries, I changed the directory by entering:
cd /Developer/NVIDIA/CUDA-5.0/samples/C/bin/darwin/release
and then entered
deviceQuery
I got the message:
-bash: cd: /Developer/NVIDIA/CUDA-5.0/samples/C/bin/darwin/release: No such file or directory
I stopped at this point.
I have gone through these steps twice.
Obviously, I am doing something wrong.
Please tell me what I need to do.
Thanks!