codistributed array, parallel computing

Hello!

I have to compute a large matrix of data. For example: for i=1:3E5 A(:,i) = ... end Do to this I use the parallel computing toolbox. The amount of memory is in my opinion very large, because the array A is copied to every worker. If I define A as a codistributed array, the for loop becames very slow (I use for drange). Are there any suggestions to speed up this computation? Thx!

[Walter Roberson]

Just to check: you _are_ preallocating A, right?

Answer 1

[Jill Reese]

Could you post the code that you are currently using? It would be easier to make suggestions based on what you have already tried.

Thanks,

Jill