How to use parallel computing

Question

Stéphane 2014-9-24

0
链接

此问题的直接链接

https://ww2.mathworks.cn/matlabcentral/answers/156029-how-to-use-parallel-computing

评论： Stéphane 2014-9-25

Hello,

I'm very new in parallel computing and I can't find what I want in the matlab help (even if it's probably in there...).

I have a code that computes things on two different (big) matrices and then use both of them to compute something else :

for i = 1:N

M1 = computeMatrix1(M1);
M2 = computeMatrix2(M2);
X = doSomething(M1,M2).
end

computeMatrix1() and computeMatrix2() can be processed independently.

My question is then : is there a way to ask matlab to process these two functions on different processors, and then process doSomething() when the previous functions are done ? Or is my question silly because I'm so ignorant in parallel processing...

Thanks

2 个评论
显示无隐藏无

Raymond Norris 2014-9-24

Are computeMatrix1 and computeMatrix2 the same function called with different input arguments? Or are they completely different functions?

The for loop itself is then not independent since for example, M1 is dependent on the previous iteration, correct?

Stéphane 2014-9-25

They are different functions at the moment, but I could probably find a way to merge them into one function.

And yes, the M1 is dependent on the previous iteration.

Thanks for your help

请先登录，再进行评论。

请先登录，再回答此问题。

Answer 1

José-Luis 2014-9-24

0
链接

此回答的直接链接

https://ww2.mathworks.cn/matlabcentral/answers/156029-how-to-use-parallel-computing#answer_152755

在 MATLAB Online 中打开

In order for Matlab to compute the functions on different cores (not the same as different processors), you could always:

 a = cell(1,2),
 a{1} = M1;
 a{2} = M2;
 parfor ii = 1:2
     a{ii} = computeMatrix(a{ii});
 end
 X = doSomething(a{1},a{2}); %You would need to take this out of the loop