I'm using Cpdgemr2d in a loop like the following:
for (ii=1; ii<=K; ii+=1)
Cpdgemr2d (1, N, A, ii, 1, descA, PL, ii, 1,
K can be as large as a few thousand. Somehow Cpdgemr2d stalls at a
certain step when K is large (1000+ for example). Any clue why?
Lapack mailing list