On Feb 14, 2011, at 2:14 PM, Jed Brown wrote:

On Mon, Feb 14, 2011 at 21:12, Jed Brown <[email protected]> wrote:

...

somehow otherwise communicating through the file system

Maybe running PETSc on a subcomm with some proprietary or high-inertia code on a different communicator? IIRC, there are some climate models that "couple" using the file system.

Insane :-(

On Mon, Feb 14, 2011 at 21:12, Jed Brown <[email protected]> wrote: somehow otherwise communicating through the file system

Maybe running PETSc on a subcomm with some proprietary or high-inertia code on a different communicator? IIRC, there are some climate models that "couple" using the file system.