Mexcuda Error on Windows 10 (Matlab 2015 b)

I already installed MingW64 compiler based on Matlab instruction, and installed Cuda v7.5. I am able to use Matlab GPU in-built syntax like GPUarray and ..., but when it comes to compile mexcuda "mexGPUExample.cpu", I get the following error:

>> mexcuda -v mexGPUExample.cu
Warning: The selected C++ compiler is not supported for CUDA compilation. Searching for a supported compiler. 

    > In mexcuda (line 89) 
    Trying MEX options 'C:\Program Files\MATLAB\R2015b\toolbox\distcomp\gpu\extern\src\mex\win64\nvcc_msvc120.xml'...FAILED
    Trying MEX options 'C:\Program Files\MATLAB\R2015b\toolbox\distcomp\gpu\extern\src\mex\win64\nvcc_msvc110.xml'...FAILED
    Warning: No supported host compiler found, or other problem with the environment.
    Continuing with selected compiler to provide detailed diagnosis. 
    > In mexcuda (line 130) 
    mex -largeArrayDims -f C:\Program Files\MATLAB\R2015b\toolbox\distcomp\gpu\extern\src\mex\win64\nvcc_msvc120.xml NVCC_FLAGS="" -v mexGPUExample.cu 
    Verbose mode is on.
    ... Looking for compiler 'NVIDIA CUDA Compiler' ...
    ... Looking for registry setting 'HKLM\SOFTWARE\Microsoft\VisualStudio\SxS\VS7' 12.0 ...No.
    ... Looking for registry setting 'HKCU\SOFTWARE\Microsoft\VisualStudio\SxS\VS7' 12.0 ...No.
    ... Looking for registry setting 'HKLM\SOFTWARE\Wow6432Node\Microsoft\VisualStudio\SxS\VS7' 12.0 ...No.
    ... Looking for registry setting 'HKCU\SOFTWARE\Wow6432Node\Microsoft\VisualStudio\SxS\VS7' 12.0 ...No.
    Did not find installed compiler 'NVIDIA CUDA Compiler'.
    Error using mex
    No supported compiler or SDK was found. You can install the freely available MinGW-w64 C/C++ compiler; see Install MinGW-w64 Compiler. For more options, see 
    http://www.mathworks.com/support/compilers/R2015b/win64.html.

    Error in mexcuda (line 157)
        [varargout{1:nargout}] = mex(mexArguments{:});

Answer 1

[Joss Knight]

As you can see from the first line of your output, your chosen host compiler is not supported by the CUDA compiler (nvcc) and you don't have a supported compiler installed (namely, MS Visual Studio).

I'm not sure what people do nowadays when they need a free version of MSVC, is it the community edition? Perhaps that will work for you.

Also worth pointing out, you'll need the 7.0 version of the CUDA toolkit to work with R2015b.

Answer 2

[Richa Gupta]

This error might be when "mexcuda" has trouble locating the NVIDIA compiler (nvcc), it might be installed in a non-default location. You can specify the location of nvcc on your system by storing it in the environment variable "MW_NVCC_PATH". You can set this variable using the MATLAB "setenv" command. For example,

>> setenv('MW_NVCC_PATH','<installed location>\CUDA\bin')

After you execute the above command, restart MATLAB and then compile the example using "mexcuda".

Note: Please use CUDA Toolkit version 7.0 as the R2015b parallel computing products are now using CUDA Toolkit version 7.0. Please refer to the header "Upgraded CUDA Toolkit version" in the documentation link below for release notes of Parallel Computing toolbox for release R2015b. http://www.mathworks.com/help/distcomp/release-notes.html#R2015b

[Joss Knight]

I think the main point is that MingW64 is not supported by the CUDA compiler, and therefore not by mexcuda.

[Joss Knight]

Oh, and by the way, if you restart MATLAB after calling setenv, your environment variable will be reset to the state it was in when you called MATLAB, so don't do that. Just call setenv and carry on.