Re: [OMPI users] Send problems on AMD64

2006-04-05 Thread Eric Brunner-Williams
Charles, I was debugging something that had the same properties. I was following the comm struct for corruption-on-free in the read phase of a write-then-read test, in mpich-1.2.7p1 (and mpich-1.2.6). It only manifested on 64bit nodes, and only when compiled on 64bit nodes. The binary when compil

Re: [OMPI users] Best MPI implementation

2006-03-27 Thread Eric Brunner-Williams
Howdy all, This is kind of an oddball think to ask, but does anyone have a rhel_4_ia64 box they'd be will to let me finnish debugging mpich-1.2.7p1? There's a memory corruption bug that shows up for mpich-1.2.6 and -1.2.7p1, but only for the rhel_4_ia64, and I'd decided it was probably the comm p