MATLAB Answers

Which Visual Studio 2019 package should I install to work with CUDA?

57 views (last 30 days)
CaG
CaG on 3 Dec 2019
Commented: Joss Knight on 4 Dec 2019
I would like to use my nVidia GPU to speed up my computations. For this reason, I'd like to install CUDA with Microsoft Visual Studio 2019 Enterprise.
Unfortunately, Visual Studio 2019 has a lot of workloads and packages and I do not know which one I should install in order to use the GPU coder.
Is the "Desktop Development with C++" workload enough for this purpose or additional packages are required?

  0 Comments

Sign in to comment.

Answers (1)

Joss Knight
Joss Knight on 4 Dec 2019
To accelerate your MATLAB code with an NVIDIA GPU, you do not need to install a C++ Compiler.

  4 Comments

Show 1 older comment
Joss Knight
Joss Knight on 4 Dec 2019
But you do need to use code generation? Or is PCT and gpuArray sufficient? I ask because sometimes we find people are under the impression that GPU Coder is a requirement for GPU capabilities in MATLAB.
CaG
CaG on 4 Dec 2019
I'd like to have a MEX file to be executed on the GPU. I do not think, for the kind of code I have, that PCT and gpuArray are enough.
Joss Knight
Joss Knight on 4 Dec 2019
But you are writing the CUDA code entirely yourself? In which case you should be using mexcuda, not GPU Coder.

Sign in to comment.

Sign in to answer this question.

Products


Release

R2019b

Translated by