Professional Projects

I have had a pretty good career so far but like most techies I generally toil in obscurity. Here are some of the projects that I worked on that did get some external visibility.

The Blue Gene/L Supercomputer

Blue Gene/L was an IBM designed supercomputer that took first place on the TOP500 list in November 2004 and held it until June 2008 when it was displaced by the IBM Roadrunner system. BlueGene first broke into the TOP 500 list at #4 with a four rack system that we used for our early development. The first place system was a 16 rack system which eventually grew to a 64 rack system. Our customer, Lawrence Livermore National Laboratory ( eventually expanded it even further.

During my time on the project (2002 to 2006) I worked on everything from low-level debugging of the compute node kernel to the mpirun program, which was what end users used to submit jobs to the machine. When I started on the project hardware was a twinkle in somebody's eye ... by the time I left it was living in a confidential building at a national laboratory and simulating the effects of aging on the nuclear stockpile. (It is better to simulate bombs blowing up than to actually try it.)

Blue Gene/L group photo Blue Gene/L 16 rack in Rochester
Some of the BlueGene/L team (2004?) The 16 rack machine in Rochester

Here are some Blue Gene/L publications that I contributed to:

Designing a Highly-Scalable Operating System: The BlueGene/L Story Paper, Super Computing '06 Co-author
Blue Gene/L programming and operating environment Paper Co-author
Blue Gene/L: Application Development IBM "redbook" Technical content contributor
The Blue Gene/L Supercomputer: A Hardware and Software Story Paper Co-author
Unfolding the IBM Eserver Blue GeneSolution IBM "redbook" Technical content contributor

IBM Cell Broadband Engine

The Cell BE was basically a supercomputer on a chip designed for use in the Sony PlayStation 3 but also used in IBM Blade servers. I came to the project after hardware was becoming available but before IBM had decided to try to build a full scale supercomputer from individual blades. My first major project was to lead a small team that designed and implemented an FFT library that became part of the SDK. After that work I hit the road for a while doing support for our business partners, teaching classes on how to program and optimize code for the chip, and working directly with end customers on their projects.

IBM Cell BE CPU picture
Showing off the QS22 at a trade show in NYC The Cell BE chip with all 8 co-processors clearly visible

Here are some Cell related publications that I contributed to:

Roadrunner: Hardware and Software Overview IBM "redpaper" Technical content contributor
Programming the Cell Broadband Engine™ Architecture: Examples and Best Practices IBM Technical content contributor
Programming the Cell Processor: For Games, Graphics, and Computation Textbook Technical content reviewer