Avatar

Today’s the day.

Today marks 10 years since the first commit in the original Open MPI CVS source code repository (which was later converted to Subversion):

$ svn log -r 1 http://svn.open-mpi.org/svn/ompi
------------------------------------------------------------
r1 | jsquyres | 2003-11-22 11:36:58 -0500 (Sat, 22 Nov 2003)
First commit
------------------------------------------------------------

As I look right now, I see that we’re up to r29729.

Wow!

The Ohloh.net site gives detailed statistics from the Open MPI Subversion history.

Before I made that first commit ten years ago, we had had many, many hours of technical and logistical discussions between the LAM/MPI, LA-MPI, and FT-MPI developer teams.  Working together seemed like a good idea, but none of us really had any idea at all whether the whole thing would work as a combined project (PAC-X MPI joined us shortly after we started).

The idea was fairly radical: merge four existing MPI implementations; take the best ideas from each of them to start both a whole new code base, and a new community of MPI implementors.  This involved taking software developers with different backgrounds, different organizational biases, and different goals and making them all work towards a common goal.

As history has shown, not only did this idea work, it worked really, amazingly, astoundingly well.

Open MPI both as a software project and as a community is alive, kicking, and more awesome than ever.  It’s been a great ride; I’ve been privileged to work with oodles of waaaay-smarter-than-me MPI implementors, MPI application developers, and system administrators over the past 10 years.

I look forward to the next 10.

Long live Open MPI!