On Thu, Nov 11, 2010 at 12:36, Number Cruncher <number.cruncher@ntlworld.com> wrote:

However as commented here: https://bugzilla.redhat.com/show_bug.cgi?id=638477#c86 the valgrind memcpy implementation is overlap-safe.

Yes, of course. That's how the bug in Open MPI was originally detected. Of course you can't do production runs with valgrind.

Are you using an Intel Nehalem-class CPU?

No, Core 2 Duo and Opteron (but the Opterons still have older glibc). Reverse memcpy must only be turned on for Nehalem.

Jed