how to rum m file with GPU

2 次查看（过去 30 天）
Shehab Tarek
Shehab Tarek 2020-7-22
评论： Walter Roberson 2020-7-22
hello guys i am asking for how to active the GPU to run the m file faster
what the i can do

请先登录，再回答此问题。

采纳的回答

Walter Roberson
Walter Roberson 2020-7-22
You cannot do that. When MATLAB talks to the GPU, it mostly does so in terms of a bunch a specialized pre-compiled code kernels (that are in the binary programming language of the GPUs.) The GPU does not run MATLAB code, only the pre-compiled snippets.
You can do some processing on GPU by using GPU Coder product; https://www.mathworks.com/products/gpu-coder.html but do not expect it to handle MATLAB code in general.
  5 个评论
显示 3更早的评论
Edric Ellis
Edric Ellis 2020-7-22
I wouldn't say that "the GPU does not run MATLAB code". Of course, MATLAB code doesn't run directly on the GPU - but if you use the GPU version of arrayfun, then the function you're running is converted into an optimised kernel and runs directly on the GPU. (That's why the GPU version of arrayfun can be blazingly fast).
Walter Roberson
Walter Roberson 2020-7-22
I wouldn't say that "the GPU does not run MATLAB code".
I would say it without hesitation, just as I would say that Intel x64 architecture chips do no run C code. The code has to be compiled to run on the target, with the instructions generally not even in the same order as implied by the original code.
gpuArray support is only available for the operations Mathworks has already pre-built GPU support for. Anything beyond that requires GPU Coder product.

请先登录，再进行评论。

更多回答（0 个）

类别

在 Help Center 和 File Exchange 中查找有关 Get Started with GPU Coder 的更多信息

标签

Community Treasure Hunt

Find the treasures in MATLAB Central and discover how the community can help you!

Start Hunting!

Translated by