IBM Eye

Are there awards for least informative headlines? Just wondering.

Anyhow, supercomputer rankings came out today, and yet again IBM tops the list.

HP has passed IBM in the number of supercomputers in operation and enjoys the largest market share, according to a list of the Top 500 supercomputers that was compiled by university computer researchers in the U.S. and Germany. But IBM’s Blue Gene/L supercomputer, installed at Lawrence Livermore National Laboratory in California, ranked first on the list and IBM claims it is the market share leader when vendors are ranked by combined computing performance.

So crummy old HP has grown their install base this year (Watch your price point IBM). The big news this year is that IBM has finally broken the petaflop barrier (the ability to execute a thousand trillion mathematical operations a second) with Blue Gene/P (not on the list yet). That really is amazing. I don’t understand how all that works, but I can’t imagine that there is a whole lot of use for that kind of raw processing power, given that the RAM allocated for a given processor is so low. And it seems that a problem would need to be easily partitionable thousands of times to use the machine’s enormous power anyway. I guess for things like the SETI project where there’s so much data and it can be logically broken up into small pieces… or whatever the NSA does. Anyhow, I think these computers are as much a learning tool for researchers as they are a practical solution for people. And the buzz it generates positions IBM in people’s mind as the best of the best. When you need a computer, you go to Dell, but when you need something done fast and well, you talk to IBM. If building supercomputers solidifies that perception of the brand, I’m all for it.

(This next paragraph wanders a bit, but I think my point comes accross.)
A lot of press attention has gone of late to Sun and the efforts of Andreas Bechtolsheim to build the Sun Constellation System. Competition is fine, and if IBM loses the title for a year (or more), no big deal. IBM comes accross as the old-guard foil to the heroic newcomer Sun. I hate IBM being cast as the Microsoft vs. Sun as the Google. Everyone (not me but it seems like everyone else) loves Google. When people think about the brand “Google”, the percieve them as doing no wrong. A bright eyed scrappy underdog turned behemoth who still remembers his roots. When they launch a product, people expect it to work well and fast and for free. Very few people love Microsoft… it’s in the business of making money rather than helping people. So that’s how I think the brands are percieved. In all the press I read about it, Sun is cast in the Google light and IBM as the doomed has-been — check out the quote (the last bit) from the NYTimes:

Mr. Bechtolsheim’s newest machine will ultimately be tested against his most powerful rival, I.B.M., in the $10 billion market for high-performance computers. I.B.M., based in Armonk, N.Y., now dominates the high end of the fastest computing ranks and expects to maintain that position when the newest Top500 supercomputer rankings are announced today in Dresden.

Indeed, I.B.M. will introduce a redesigned version of its BlueGene supercomputer, to be named BlueGene/P today at the conference, saying that the new machine, scheduled to be installed next year, will finally break the petaflop computing barrier — the ability to execute a thousand trillion mathematical operations a second.

Executives at I.B.M. are skeptical about the new Sun supercomputer, noting that the system is late to be installed. “Having done six generations of machines,” said Dave Turek, the company’s vice president of Deep Computing. “I have come to realize that very little goes right the first time.”

A number of Silicon Valley technologists are, however, betting on Mr. Bechtolsheim. “He’s a perfectionist,” said Eric Schmidt, Google’s chief executive, who worked with Mr. Bechtolsheim beginning in 1983 at Sun. “He works 18 hours a day and he’s very disciplined. Every computer he has built has been the fastest of its generation.”

So IBM may get it from both sides next year - upstaged at the top next year by scrappy Sun, while HP continues to steal marketshare, leaving IBM PR to coming up with increasingly strained measurements to still claim the #1 spot. I don’t know the future though. but this year (as in the past 3) IBM is on top.