MPI Is Dead … and It’s About Time

First of all, this is from 2007, since then the number of openMPI deployments has increased steadily, the latest release is of the 4th of may of 2010, so to declare it dead is realy a bit premature, it looks like MPI is alive and kicking.

Appistry apparently makes its money by selling cloud management software. This is fine and good but does not overlap 100% with the main field of application for MPI, which is large scale number crunching in the scientific world.

This includes large scale simulations of all kinds, including fluids, particle systems, molecular and sub-molecular simulations.

While it is possible to run these on a different architecture by porting them, the amount of code that is written for and around MPI means that MPI is here to stay for a long long time. After all, rewriting all that stuff would take ages and cost a fortune.

Probably MPI will be around just a little longer than FORTRAN.

I'm confused. Maybe I'm stupid, but I didn't even understand what the supposedly better "other way" is.

The MPI codes I've seen are scientific codes where the data is a large domain that can only fit distributed across many nodes, and where there are interactions that must be passed between these. Typical examples are hydrodynamics codes, N-body codes, radiation transfer codes, or combinations thereof.

Can someone explain to me in an understandable way how they think these applications should be coded?

Summary: "MPI is useless for all those cases except those cases where MPI is useful. It's possible that our service, about whose details our website provides astoundingly little information, might be more use in some of the cases where MPI is overkill. Quick, let's stick on a trollish headline so that everybody who currently uses MPI (I'm looking at you, hugh3) will click on this link."

Thanks, appistry link-whores! Tell ya what, rewrite my code in your non-MPI environment and we'll talk. You do have a ScaLAPACK equivalent up and running, right?

The summary here seems to be that this guy is pretty familiar with embarrassingly parallel problems, correctly points out that MPI is an error-prone and labor-intensive way to solve embarrassingly parallel problems, and incorrectly assumes that there aren't really any important problems that aren't EP.