Memory usage in codistrubuted array

I am wondering if in the following code, the total memory usage is twice that needed for the matrix A, since the local part of A be being copied into lA in each worker.

spmd
    A = codistributed.rand(10000);
    lA = getLocalPart(A);
    lA = lA + labindex;
    A = codistributed.build(lA, getCodistributor(A));
end

I think it is because when I run the lines separately, and look at the memory usage in Windows Task Manger, I see the memory doubling after the lA=lA+labindex; line.

Is there a way to modify local parts of a codistributed array without making copies?