Nvidia Boss Opens First NVISION With Poke At Intel3:30 PM EST Tue. Aug. 26, 2008
Advances made in graphics processing represent "the most radically innovative technology in computing over the past 10 years" and the rise of smart phones is "the second personal computing revolution," according to Nvidia CEO Jen-Hsun Huang.
Jen-Hsun, kicking off the inaugural NVISION visual computing conference Monday with a lengthy keynote that incorporated demos of cutting edge graphics technology and a chat with Battlestar Galactica star Tricia Helfer, added that "in Moore's Law terms, what we've done with GPUs is extraordinary."
It would not be the last poke at Intel during the course of Jen-Hsun's two-and-a-half hour presentation and an additional 45 minutes spent with reporters following the keynote.
NVISION is being held in San Jose, Calif. at a number of venues through Wednesday. Jen-Hsun said in a post-keynote interview session with media that the Santa Clara, Calif.-based graphics chip maker would not be releasing product news at the event, which it intended to be a celebration of visual computing writ large rather than a showcase for Nvidia alone.
"Few technologies have made the leaps that the GPU has over the past 10 years. Years ago, the GPU was really just an accelerator, an application-specific integrated circuit. Now it's a general purpose parallel computing processor," said Jen-Hsun, who is also president and co-founder of Nvidia.
"When I got started in my career in 1984, a Cray X-MP cost $1 million," he said, noting that today's GPUs cost just a few hundred dollars and "their computational capability has reached into the Teraflops, or 1,000 Cray X-MPs."
Meanwhile, smart phones represent the single most important category in computing today, according to the Nvidia boss. Asked about Nvidia's "mobile strategy," Jen-Hsun said Nvidia was "completely focused on Windows Mobile 7."
"Focusing on smart phones. That's our strategy," he said.
Nvidia's specific goals vis-'-vis smart phones include building GeForce platforms for the devices, reversing the current smart phone paradigm of "phone first, computer second," and working with low-power CPU maker VIA to optimize Nvidia products for the Taiwanese chip maker's hardware.
"We're so excited about VIA, we're optimizing our entire software stack for [new VIA CPU] Nano," Jen-Hsun said, noting that Nvidia has only done that for two other chip companies, presumably Intel and AMD.
Next: Cylons Do It Better
The statuesque Helfer, who plays the Cylon character Number Six on the Sci-Fi Channel's Battlestar Galactica television series, helped Jen-Hsun close his keynote with a discussion of the demands on actors in CGI-heavy shows like BSG. Clad in a clingy red dress that drew whistles from the crowd, Helfer joked amicably with Jen-Hsun, who himself appeared in the jeans and t-shirt the famously non-buttoned down CEO favors.
"Was she tall or what?" he joked following the keynote at San Jose's Center for Performing Arts.
Earlier in the two-and-a-half hour presentation, the Nvidia boss chatted with a succession of guest technologists specializing in visual computing fields such as digital prototyping and image processing, as well as technologies like parallel computing that have emerged from GPU advances.
Judging from audience reaction, the most popular keynote moments aside from Helfer's turn were demos of a free South Korean social networking game that generates income via "micro-transactions" for small user upgrades, and Jeff Han's presentation of a 100-inch multi-touch display featuring the hands-on multi-touch user interface technology developed by Han's company, Perceptive Pixel.
Meanwhile, Jen-Hsun's keynote reference to rival Intel's famous Moore's Law-based product strategy was clearly more than just an off-hand remark. The Nvidia boss compared the NVISION conference to the recently concluded Intel Developer Forum in San Francisco several times during a post-keynote interview session with media and seemed eager to play up Nvidia's competition with its larger rival.
"We didn't call this 'the Nvidia Developer Forum,'" Jen-Hsun joked. He also claimed that Santa Clara, Calif.-based Intel and Advanced Micro Devices of Sunnyvale, Calif. had been invited to the first NVISION event but declined.
Jen-Hsun also addressed Intel's latest discrete graphics initiative, code named Larrabee. The chip giant recently presented a paper on new graphics technology it plans to release in the next year or two at the SIGGRAPH industry conference in Los Angeles. Intel's main argument for Larrabee revolves around the programming benefits it claims will be gained from building GPUs on the x86 architecture, particularly in Web-centric applications.
Intel currently produces its own on-board graphics engines -- Intel executives are quick to point out the chip giant "ships the most graphics in the world" -- but has long left the discrete graphics market to be battled over by Nvidia and AMD's ATI division.
"Larrabee hasn't shipped so you don't know what it is and I don't know what it is," Jen-Hsun said. "That said, by the time it does ship, Nvidia's technology will be so far advanced it won't matter."
He shrugged off Intel's claims that the bulk of Internet applications were built to run on Intel architecture (IA) microprocessors, making IA, or x86-based GPUs a natural fit for visual computing programmers transferring traditionally client-based applications to hosted or "cloud computing" formats.
"The Internet doesn't run on x86. It just doesn't, no matter how often Intel says it does," Jen-Hsun said.