Intel has had some rocky times bringing Itanium to market, but Pat Gelsinger believes the company is over the hump.
Gelsinger, general manager of Intel's Digital Enterprise Group, has no shortage of bullish predictions for Itanium. "We will look back at this as the inflection point of the platform," he said in an interview at Intel's launch event for Itanium 2, code-named Montecito, on Tuesday.
Montecito arrived well after the 2005 launch date Intel had envisioned, but it is unquestionably an improvement over its predecessors. It has double the performance and yet consumes 20 percent less electrical power. Much of the performance boost is from the use of dual-processing engines, a feature already present when IBM introduced Power4 in 2001 and Sun introduced UltraSparc IV in 2004.
Gelsinger is responsible for improving not just Itanium technology, but also its image. The chip family's debut five years ago was beset by performance problems, delays and software incompatibility with mainstream chips. More recent ignominy included decisions by Dell and IBM to drop Itanium servers.
Gelsinger, a veteran Intel employee, is undaunted. He's risen through the ranks since joining Intel in 1979, and has held titles such as 80486 processor chief architect and, more recently, chief technology officer. He talked to ZDNet UK’s sister site, CNET News.com, at the launch event.
Q: Intel today launched the Montecito processor version of its Itanium 2 processor family. What's important about the Montecito launch?
A: This is the exciting announcement of the dual-core, 1.7 billion-transistor, multithreaded, virtualised, secure, reliable processor. What we're trying to do with the Itanium family is position it as the unquestioned solution for mission-critical computing. We are joined by eight OEMs (original equipment manufacturers, a term used to describe computer makers) announcing systems to go with it. We continue to build our ISV (independent software vendor) momentum. There now are over 8,000 applications on Itanium.
You described Montecito as an "inflection point". Why is that?
There are a couple of aspects to today's announcement. We've had to labour to build a new architecture. Today, it's unquestionably at critical mass, with eight OEMs shipping it. We have the operating systems available for it, the applications momentum around it, the Itanium Solutions Alliance. It's credible. We posted incredibly good benchmarks.
And we're seeing it's up and to the right: Systems revenue is growing very consistently, over the last couple of years. We're now half the size of the rivals we're aiming at, Power and Sparc. Clearly, we've arrived at the point where this thing is gaining momentum. There was pent-up demand for Montecito. It's a socket drop-in: It drops right into the Madison (Itanium) systems today, so customers can immediately start to take advantage of this great new technology.
It wasn't all smooth sailing, though. In October last year, you announced a delay. You initially wanted to launch Montecito in 2005, but you started shipping it in June 2006, and servers with it won't generally be available until September. What happened in that delay, and what was its effect?
We were embarrassed by the delay, of course. We're a microprocessor company. We should be able to deliver these things when we say we are. Unfortunately, we hit some difficult spots, so we announced the delay last year. Since then, we've executed exactly against what we said we would. I'm pretty proud of that. The team has worked hard to get to this point.
Secondly, there are customers. They were anxious. They were ready to go. They had the sockets, the systems ready to go. There really is some of that pent-up enthusiasm for it.
Thirdly, we had to convince the industry, "Yes, wait, this thing really is great". Now that they've begun seeing it, and (Intel has begun) seeding it with customers and the ISVs, they're saying, "Wow, this is really a breakthrough platform". They are really embracing it in quite an aggressive way. As we see the system announcements from the OEMs over the next 30 to 90 days, I think you're going to see some really exciting stuff.
A few weeks ago, you launched a new Xeon processor, the Woodcrest model, which is a pretty big redesign. In the scheme of things, which is more important for you, Woodcrest or Montecito?
Yes.
No, come on! One or the other.
It's really two different markets. In terms of the revenue impact, the Xeon family is a much bigger near-term revenue impact for Intel, so that's really big and important for us, and it's why we put a lot of emphasis on that announcement.
But the market is really almost cut in half, as you think about the market for servers. About half of it is volume servers where Xeon is king. Then you have the back-end data centre, mission-critical machines which are about $25bn as well. That's the space of the big mainframes, the big iron. The volumes are much lower, but the system prices are much higher. That's where Itanium is the target.
Is the Itanium business meeting your financial goals?
Our goal is to grow it. We've laboured to get the platform to critical mass, as we believe it is now, and we've seen the systems revenue really start to take off. It's not yet up to the level of profitability that we hope to see it, but we are very encouraged by the strong demand, and we are ahead of our forecast for the product line this year.
