Asked by Brian
on 11 Jun 2012

I am attempting to run a ParFor statement with a step function and Matlab is giving me an error "The range of a parfor statement must be increasing consecutive integers. See Parallel Computing Toolbox, "parfor"." I'm wondering if this is possible or if I have to do some manipulation to my loop to get this to work in parallel. Would someone be able to make a suggestion as to how to modify one of these loops so that it will work in parallel?

Simplified Example

parfor (i = 1:10:75,8) test3(i)= i*3; end

Real Example

parfor (i = 1:10:75,8) tic recnum = min(i+9,length(test)); datainsert(conn,'dbo.tbl_FactorExposure_Security',{'CUSIP','MktDate','Factor_Exp','Factor_Num'},MyData(i:recnum,:)); disp(i) toc end

Thanks a lot! Brian

Answer by Edric Ellis
on 12 Jun 2012

Accepted answer

As the doc states, PARFOR needs to work with unit-step ranges. So you do need to re-work the loop slightly, something like this:

idx = 1:10:75; parfor (i = 1:numel(idx), 8) val = idx(i); test3t(i) = val * 3; end

That example leaves 'test3t' slightly different to 'test3' in your example, you can rearrange things by doing

test3(idx) = test3t;

Edric Ellis
on 12 Jun 2012

MATLAB and Simulink resources for Arduino, LEGO, and Raspberry Pi test

Learn moreOpportunities for recent engineering grads.

Apply Today
## 2 Comments

## Walter Roberson (view profile)

Direct link to this comment:http://nl.mathworks.com/matlabcentral/answers/40881#comment_83998

Sorry? What do you intend

for (i = 1:10:75,8)

to mean? Is it intended as

for i = [1:10:75, 8]

meaning

for i = [1 11 21 31 41 51 61 71 8]

?

## Brian (view profile)

Direct link to this comment:http://nl.mathworks.com/matlabcentral/answers/40881#comment_84011

Yeah, sorry for being unclear. In my real example I am breaking up the size of my datainsert command by using a step function. I'm doing this because I'm inserting about 28 million records and the java heap size in Matlab is running out of memory. SInce I'm breaking it up, I'd like to do the datainsert in parallel so that it's not so slow.