As I get ready to head to Hamburg for the 24th running of the International Supercomputing Conference, I’m taking a last look at the conference agenda to make sure I don’t miss any interesting bits. Fortunately the conference maintains a plenary track of sessions, but there are other activities — BoF sessions, exhibitions, poster sessions, and the obligatory networking and meetings — taking place in parallel, so some planning is mandatory. With that in mind, I thought I’d point to the some of the highlights of the four-day supercomputing extravaganza.

Of course the opening session on Tuesday includes the much-anticipated announcement of the latest TOP500 list. I’ve got to admit, I love the horsepower race, but this year might not have the excitement of recent lists, when petaflop supers were breaking onto the scene. Unless IBM manages some last-minute surprise, ORNL’s Jaguar super will take over the number one spot from the Roadrunner machine at Los Alamos. There is also the expectation that the rest of the top 10 will be shuffled as the newest international supers find their place.

The most interesting aspect to watch may be the TOP500 list’s turnover rate, which has accelerated recently. Because of the crummy economy though, some are predicting a dip in TOP500 volatility. The turnover, or lack thereof, is not likely to excite the mainstream IT press very much, but this is something I’ll be watching.

At the end of the opening session is Andy Bechtolsheim’s keynote, which focuses on the evolution of HPC interconnects. Bechtolsheim, who co-founded Sun Microsystems and is now the chief development officer at Arista Networks, seems to have impeccable timing with industry trends. His original $100,000 investment in Google is now thought to be worth over $1 billion. In 2007 he co-founded 10GbE switch maker Arista Networks, just as the fortunes of his original firm, Sun Microsystems, seemed to be heading south. Bechtolsheim is always worth listening to.

Later on opening day, you have a choice of sessions: one on HPC climate modeling and the other on supercomputing challenges in aeronautics, as well as a gaggle of scientific research sessions and BoF meetings. Unless you have a particular research interest in aeronautics or one of the super-science topics, I’d go for the climate modeling session, and in particular, the panel discussion in the afternoon: “If Global Change is the Grand Challenge Application, Do We Need a World Climate Computing Center?” Great question, and in the age of supercomputing one-upmanship, it implies we’ll need some political finesse to build a global HPC capability.

Finally, don’t miss the opening day welcome party, starting at 6:00 p.m. and presumably ending when the food and beer run out. Good chance to chat with the exhibitors, future employers, close friends and despised enemies. It’s not as exhausting as it sounds. In many cases, these are all the same people.

Which brings us to Wednesday. If you only go to one presentation the whole week, I’d recommend the cloud computing session Cloud Computing in HPC — Synergy or Competition. Current deployments and use cases of cloud technology, and assessments of how the cloud model fits with high performance computing will be discussed. There’s an all-star cast of HPC movers and shakers coming from traditional vendors (IBM, Sun, HP and Microsoft) as well as the high-flying cloud companies (Google, Amazon and Yahoo). A panel discussion at the end wraps it up. Two of the panel participants are Google and Microsoft, so there is a possibility of gun-play.

Later that day is the first hot seat session, which puts some of the top brass at 18 HPC companies on the spot. The vendors are given 10 minutes to speak their piece and then field a couple of questions from “hand-picked inquisitors,” most of whom come from the HPC research community. On Wednesday, Microsoft, Intel, IBM, Bull, Spectra Logic, Fujitsu, BlueArc, Cluster Resources, and Cray get grilled. On Thursday, the second hot seat session puts Sun Microsystems, Supermicro, HP, SGI, Mellanox, Myricom, QLogic, Force10, and NEC under the spotlight. An ISC tradition.

Also on Wednesday, LSU professor and HPC maven Thomas Sterling will deliver his customary state-of-the-supercomputer address for 2009. Sterling is always informative and entertaining. Not to be missed.

On Thursday, a couple of academicians try to answer the existentialist question for the multicore obsessed: What the heck is with all these cores? The multicore session features Yale N. Patt, Professor of Electrical & Computer Engineering at the University of Texas at Austin, and Kathy Yelick, Director of NERSC and CS prof at UC Berkeley. Patt will talk about future architectural directions of multicore processors, while Yelick will give us the latest developments on the software side.

For something completely different, you’ll want to check out IBM’s Prof. Dr. Gunter Dueck’s keynote on Thursday afternoon. His presentation is about Lean Brain Management. What pray tell is that, you ask? From the session summary:

Lean Brain Management aims at economizing on intelligence by moving all the necessary brain work in to a perfect system which can be handled by completely unskilled workers, the future “morons”. Only the system must be intelligent, not the employees. We estimate that a perfect system can be designed by a few intelligent people, the so-called “moronorgs”, who dumb down all business processes to perfect foolproof Lean Brain Quality.

Groovy. Presumably mind-expanding drugs will be handed out before the talk.

On Friday, make sure you attend the morning session on what’s being done with the new petaflop (and a few near-petaflop) machines around the world. In this session, we hear from some of the masters of the petaverse from the US, China, India, Germany, and Switzerland.

That wraps up my hot picks. To see the entire conference schedule, take a look here.