ParFor - how to rewrite For loop

Is it possible to rewrite the following code to that the main loop (a 1:NumData) use Parfor and allows the main loop to be processed in parallel. The core prpoblem is the need to access the 'a' variable which is state dependent and therefore excludes the use of Parfor.
Suggestions would be appreciated.
row = -1; high = 1;
NumData = 100000;
% start timer
tic
% pre allocate memory
RandData= ones(1:80,1:NumData);
% make random numbers
for a=1:NumData % <-- use ParFor ??
for b=1:80
RandData(a,b) = round(low + (high-low) * rand);
end
end
% end timer
toc

回答(1 个)

Have you tried it? It works fine for me
parfor a=1:NumData
for b=1:80
RandData(a,b) = round(low + (high-low) * rand);
end
end

2 个评论

I tried this, but it did not work as I think that in parallel mode (using matlabpool) the parallel processing has no way of what the variabel 'a' is globally when numerous instances are being run, unless they all just mimick oneanother or in fact the Parfor is not being run in parallel.
How many workers were involved in the matlabpool?
I don't see the state dependency? What you have posted looks like straight-forward use of parfor to me.

请先登录,再进行评论。

类别

帮助中心File Exchange 中查找有关 Parallel for-Loops (parfor) 的更多信息

提问:

2011-10-4

Community Treasure Hunt

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

Start Hunting!

Translated by