I came back from the Open MPI collective meeting in Albuquerque. It was amazingly fruitful and we had many new ideas for non-blocking collectives. Many of the are actually new problems but it’s good to discuss them before I am challenged with those questions at conferences without preparation. The meeting was hosted by the university of New Mexico
To be precise, the Center for High Performance Computing at the UNM:
We ran through the discussions pretty quickly (actually too quickly for my taste, because we “dropped” (i.e., decided not to implement) most of the nice collv2 things we designed last year) so that we finished early. We had the last late afternoon (5pm) off and I decided to go with a guy from Sun Microsystems to do some sightseeing. We went to the Petroglyph National Monument to see some drawings of native americans.
We had to wak through the hot desert to see them
and they were really disappointing!!!! .
Anyhow, we went on and wanted to see the “Vulcanoes”.
The first one was pretty lame …
but we had a nice view over Albuquerque and the Sandia Mountains
And of course the *real* volcano!
That we finally climbed! *yeah*
So that’s pretty much it! The afternoon was over and we flew back the next day. A very productive meeting!