Is there any faster way (parfor related)?

Question

Eungyu Park 2020-5-12

0
链接

此问题的直接链接

https://ww2.mathworks.cn/matlabcentral/answers/524774-is-there-any-faster-way-parfor-related

回答： Bjarke Skogstad Larsen 2020-5-12

I want to reduce the computational time of following parfor loop. I think many MATLAB users who use techniques of MCMC, ES, or any Monte-Carlo have similar problems. Please share your thoughts. Many thanks!

A=[]; % cumulation matrix
parfor ii=1:100 % or more than 100
    b=some_function(ii); % b is very tall, say it has dimension of 1000000X5 or taller
    A=[A b]; % cumulation
end

0 个评论
显示 -2更早的评论隐藏 -2更早的评论

请先登录，再进行评论。

请先登录，再回答此问题。

Answer 1

Bjarke Skogstad Larsen 2020-5-12

1
链接

此回答的直接链接

https://ww2.mathworks.cn/matlabcentral/answers/524774-is-there-any-faster-way-parfor-related#answer_431987

You could speed this up by pre-allocating memory for A if you know the size of b:

A=nan(1000000,5,100); % cumulation matrix
parfor ii=1:100 % or more than 100
    A(:,:,ii)=some_function(ii); % b is very tall, say it has dimension of 1000000X5 or taller
end