Updates to this story
Don't write off Intel's Larrabee completely, because the project will come back with a vengeance when you least expect it, according to an industry insider close to the project.
Our source, who had been a key member of the Larrabee engineering team said the announcement of Larrabee's apparent demise a few months ago had been seriously mishandled by Intel, leading people to believe the project had been completely shelved and would never see the light of day. "That is just not the case," our source told us adding "Intel is not a one-shot company."
Admitting that information about Larrabee had leaked out "like information leaking out of a leaky sieve," our source also said he'd never bet against Intel and that had Larrabee emerged on time, "it would have been industry changing.
"But there is a whole product plan and roadmap still in place," he said adding "Intel has the luxury of being able to release stuff in its own time, there aren't too many other companies that can do that."
Our source said the delay had certainly given Nvidia and AMD time to catch up and push their graphics capabilities forward, but noted that Intel had really hired the cream of the crop when it came to graphics talent for Larrabee and that many of those people were still waiting in the wings, biding their time before the project resurfaced.
"We were literally hundreds of people, Intel picked some really big hitters and a lot of those people are still hovering around waiting for Larrabee to come online again," he said.
And when it does resurface, don't expect it to be poorly bunged together, either. According to our source, some disgruntled ex-Larrabee project employees had been actively spreading fud and lies, leaking bad information to press about Larrabee's technical capabilities.
"The funniest one I heard was that Intel could use Imagination Technologies' tech to fill in the gaps, which of course is absolute nonsense. I had a good laugh over that one, and so did the folks at
Imagination Technologies."
Why would Intel stop developing Larabee? Only reason that fits with good business practice is that it wasn't working out. It was beyond Intel's capabilities to design and/or produce at a competitive price.
Right now, it's a dead-end. Maybe in the future, Intel will have a reason to restart the effort. Maybe. Until then, the people they hired are being reassigned to other efforts of more immediate profitability.
Second, as GTX480, larabee lost it's thunder by coming out too late. There is no point in releasing a top GPU which can't overtake the competition. I'd be an embarassment to intel.
But as pointed out, intel already spent a lot of money in this project and it has a chance to grab a bigger slice of GPU market (it already own it, with it's integrated (crappy) graphics).
So I'll be less than surprised if we get to see larabee as a GPU in 2011.
Hardly "industry changing".
Wargo is right, they couldn't get the cost down far enough to make it competitive. It does give the old pentium fabs something to do though.
The other piece that called my attention from that article was this piece: "...Our source said the delay had certainly given Nvidia and AMD time to catch up and push their graphics capabilities forward..."
My question is: time to catch up in what?
The only one who needs to "catch up" here is Intel. AMD and Nvidia are way ahead in their GPU efforts than intel will ever be.
You guys can do better journalism than this, so please stop spreading FUD.
From the first moment i heared intel was working on a gpu i knew they were secretly planning on merging the cpu and gpu.
Trust me, nvidia is going down. Why? Because they got a lot of power waaaay to quick and think they're invincible now... but any smart man knows the cpu makers will catch up eventually... it only takes time.