is it possible to replace for loops to fast it up?

2 次查看（过去 30 天）

显示更早的评论

elis02 2018-9-27

0
链接

此问题的直接链接

https://ww2.mathworks.cn/matlabcentral/answers/420951-is-it-possible-to-replace-for-loops-to-fast-it-up

回答： Jan 2018-9-27

在 MATLAB Online 中打开

Hello,

Example of something i have is:

for ang=1:length(angle)
parfor dist=1:length(distance)
sol(ang,dist)=abs(pulsecompress_new_3620_check([distance(dist) angle(ang)]));
end
end

every alteration in the function takes a long time. I have long arrays for "distance" and "angle" these two "for" loops might take several days. Is it possible to use the matrix nature of matlab to make this code run faster?

Thanks,

3 个评论
显示 1更早的评论隐藏 1更早的评论

KSSV 2018-9-27

We need to check the function pulsecompress_new_3620_check to make it fast.

elis02 2018-9-27

attaching the files

请先登录，再进行评论。

请先登录，再回答此问题。

回答（1 个）

Jan 2018-9-27

0
链接

此回答的直接链接

https://ww2.mathworks.cn/matlabcentral/answers/420951-is-it-possible-to-replace-for-loops-to-fast-it-up#answer_338703

Start the optimization of the code by using the profiler to determine the bottleneck. It is a waste of time to improve a piece of code, which takes a negligible amount of the total run time only: If you accelerate a part be a factor 2, which takes 2% of the run time, the total time is reduced by 1% only.

After finding the most time consuming part, post it here together with some relevant inputs, e.g. created by rand() if possible. The typical dimensions matter.