The share prices for AMD and NVIDIA rose yesterday after news broke that Intel was dropped initial efforts on a graphics processor codenamed ‘Larrabee.’ Analysts and investors saw this move as a setback for the silicon behemoth in reducing the reliance on their core business of core silicon.
What it did do was say that, basically, the big bad wolf wouldn’t come around to rain on the parade for another four-to-eight quarters,” said Wedbush Morgan Securities analyst Patrick Wang.
When Larrabee was first announced over a year ago, there was a lot of concern and speculation,” Wang said. But he added: “Clearly it’s not as easy as Intel initially thought it would be.”
So what’s next for the Intel Larrabee division? Intel spokespeople indicated that a future Larrabee product would be used for a software development platform for graphics and high performance computing. What that really means, no one outside Intel really knows. Either way, NVIDIA’s shares closed up narly 13% on Monday evening at $16.09 and AMD closed up 8.4% at $8.52.
This begs an interesting question, does Intel really control that much of the eventual fate of HPC and the silicon technology market at large? If you’re a commodity cluster vendor, one can argue that a large portion of your roadmap directly mirrors that of Intel [the other large percentages fall on AMD and the various interconnects]. HPC heavyweights such as Cray and SGI base much of their technology around a singular core silicon platform. Their development cycles are simply too long to do otherwise. Those that produce design core silicon specifically for HPC are the minority. NEC, IBM, Fujitsu and [small portions of] Cray still do so. With the exception of Cray’s vector business, all of the latter organizations have large investments outside of HPC. Does this situation bode well for the future of HPC?
I’m in no place to answer this question, but I believe its prudent at this industry juncture to provoke the discussion. For more info, read the financial article here.