It’s that time of the year for Intel developers and customers alike. Several thousand will make a pilgrimage to San Francisco this week, where weather is reportedly going to be in the 90 degree range, for the Intel Developer Forum, to find out what the chip giant has in store.
This will be the first IDF without Pat Gelsinger, the former CTO who created the Intel (NASDAQ: INTC) event in 1997 and was a prominent and popular speaker at the event. He jumped ship to become President and Chief Operating Officer of storage giant EMC last week, and will be replaced on stage by Sean Maloney, executive vice president of the Internet Architecture Group. Intel CEO Paul Otellini will give the opening keynote Tuesday.
Larrabee, Intel’s GPU project, may well make an appearance at IDF to quiet fears and gossip that it’s behind schedule. Intel has consistently said Larrabee would be available in the first half of 2010 and has not deviated from that date, noted Jon Peddie, president of Jon Peddie Research.
“I expect to see something about Larrabee. I’m not sure what but it should be proof of concepts of Larrabee,” Peddie told InternetNews.com. He also had one pretty wild speculation: that Intel would show off new optical communications techniques that use fiber optics to replace the traditional cabling in computers.
“If [the big surprise of the show] is not Larrabee I don’t know what it’s going to be,” said Nathan Brookwood, research fellow with Insight 64. If Larrabee is delayed in any way, then he doubts Intel will talk much about it.
This past May during Intel’s annual meeting with financial analysts, Larrabee only came up once, and in passing. Not exactly a ringing endorsement. A spokesman for Intel said there would be an update on Larrabee at IDF, but declined to give details.
Jim McGregor, chief technology analyst for In-Stat, said he believes Larrabee is indeed delayed. “Their whole idea was they were going to change the GPU compute model. It took them two years longer to make a flash part than expected, and that was just a memory component. Any time you try to do something new in silicon, it’s just not easy,” he told InternetNews.com.
“The fun part of IDF is not what they talk about but what they don’t talk about, like Ultra Wideband. That’s when you know they dropped it. I don’t think they’ve dropped Larrabee yet but they are backing off on it,” he added.
There better be a strong showing, Brookwood said. “This IDF will be very crucial in terms of how people perceive Larrabee because the expectation earlier this year was it would be something pretty ripe for getting launched late in the year or early 2010. If Larrabee does not have a big presence at IDF, those of us who want to read tea leaves will see a lot of negative leaves,” he said.
Page 2: Core updates, Pine View and Sandy Bridge
Page 2 of 2
Core updates, Pine View and Sandy Bridge
It’s pretty much up to Larrabee to provide some excitement because, barring some unexpected announcements, the show is not shaping up to be a hot news event.
“This is going to be more of proof points. It’s not an earth shattering year,” said McGregor. “We know everything that’s going on. They are moving to 32 nanometer base, the Nehalem architecture will move throughout their product line.”
Brookwood agreed. “Intel obviously has a great story in terms of the 32nm Westmere products and Core i3/i5/i7, so they will trot that out as best they can. Some of that is not fully understood by this community and they will want to know about the broader deployment of technology to the masses. For those of us who follow this on a daily basis, a lot of this will be old hat,” he said.
Atom will likely be the topic of conversation, in particular, Pine View, a new system-on-a-chip. “That’s going to be pretty cool and Pine View is a low-end unit with graphics. Given all the market pressures Intel has, they are probably going to tweak it up a bit,” said Peddie.
He also hopes to see more discussion from Intel on lowering the price of its high-performance 250GB NAND Flash drives as well as advances in the cycle lifespan of flash drives. One of the issues facing flash drives is that the cells where the bits are stored decay with repeated overwrites. This is especially true with multi-level cell (MLC) designs. Increasing the life span of MLC drives has been a challenge before Intel and other flash vendors for some time.
Brookwood also said he hoped to hear more about Sandy Bridge, which would be the next new architecture to follow Nehalem. That won’t ship until late 2010.
“They said they taped that out a while ago so there should be silicon in house. There should be demos of Sandy Bridge to demonstrate tick/tock is alive and well, at least in the x86 computing side of the world,” said Brookwood. (Intel follows a two-part cycle — what it calls a “tick-tock” model. The “tock” represents the release of a new architecture, while the next update — the “tick” — is a shrink of that architecture).
Why did Gelsinger leave?
As for Gelsinger, his departure was a shocker for many who knew the 30-year Intel veteran. Despite conspiracy theories floated on blogs that Gelsinger was pushed out due to problems with Larrabee, Peddie, said that’s all rubbish.
“He had four tons of money thrown at him and an opportunity to be CEO in three years. Basically they made him an offer he couldn’t refuse,” he told InternetNews.com.
EMC, he argued, needed a chip wizard like Gelsinger. “Servers are the foundation, the backbone of the Web, and we’re going to need more and more of them and they need to go faster and faster. If you are in the server business and don’t have someone who knows the ins and outs of how to get the most out of a CPU, you’re not going to be in the server business very long,” said Peddie.
