parfor memory usage inefficiency

I'm using a parfor loop in my code. One of the things I pass to the workers is a very large data cell array that matlab copies in entirety to each worker. Each worker works on a subset (determined on certain groupings I calculate beforehand) of the data cell array. This is obviously very inefficient and I sometimes run out of memory. I do not change anything in the original data cell array. Is there any way I can tell matlab not to create a copy of the original data cell array and instead reference the original to get around this issue?

[Matt J]

We'd have to see your code. If the data is a sliced variable, only the slice used by each worker should be copied.

Answer 1

[Jason Ross]

You might find Edric's FEX submission useful:

http://www.mathworks.com/matlabcentral/fileexchange/31972-worker-object-wrapper

Answer 2

[Shishir Moudgal]

Perfect ... I'll give it a shot. Thank you. I was working on an alternative using matfile.