SANTA CLARA, CALIF. – If there was any doubt Nvidia was spoiling for a
fight with Intel, its colorful CEO dispensed with that thoroughly here at an analyst’s conference.
“The competition has been awfully preoccupied with our company lately,” Nvidia (NASDAQ: NVDA) CEO Jen-Hsun Huang told the gathering at Nvidia’s offices here just a few miles from Intel’s headquarters. “I can’t go to a conference or read an article where they don’t talk about our company, how we’re going to go out of business shortly. It seems there’s a belief that if you say something loud enough and often enough, people will believe it. I think consumers are more discriminating.”
Unlike the Intel (NASDAQ: INTC) analyst briefing a few weeks ago, where a suit and tie was standard issue, Nvidia’s was a more casual affair with Huang, Nvidia’s co-founder and CEO, doing most of the talking.
Dressed in dark jeans and a sweater, he was more Steve Jobs than (Intel CEO) Paul Otellini, although his restless pacing back and forth with the microphone was perhaps more akin to a Chris Rock standup — without the obscenities. Huang maintained he was only knocking Intel in response to comments the chip leader made about his company.
Intel apparently took more than one swipe at the GPU
Huang ate it up. “This is the most inspirational slide I can imagine
giving to our employees. Nothing fires us up more than that. I’ve been told many times Nvidia will be dead soon. It’s been 15 years and we’re still passionate about this business,” he said.
While Nvidia got its start as a gamer’s company, Huang sees the graphics processing unit (GPU) moving into the broader area of computing he called visual computing. “We believe this is one of the most important areas of advancement and excitement today,” he said. “We believe visual computing will define the market for the next ten years because this is the visual generation.”
To get there required making the GPU Nvidia developed into more than just an accelerator for videogames, which is how it started. Next was the introduction of computational graphics and the Compute Unified Device Architecture (CUDA), a C-like language that lets developers write compute-intensive applications, usually financial, life sciences or visual, that utilize both the CPU and GPU.
“Clock speeds and multi-core and symmetric multiprocessing were not
enough. It turns out that the invention of this programmable GPU and CUDA on top of it connected these two paths. Visual computing and heterogeneous
computing became tied at the hips,” said Huang.
“The CPU is still very important,” he added. “Heterogeneous computing is about the fact that both are important. You need mixtures of both to do you jobs well.”
This was where Intel came in for another whipping. Intel has made noises about getting into high performance discrete graphics, an area both Nvidia and ATI specialize in. But Huang scoffed at the notion that the integrated graphics chipsets Intel provides could ever be competitive. He displayed benchmarks that showed Nvidia’s GPUs had 35 times the performance over Intel’s 965G and G35 chipsets.
Huage argued that PC sales have moved from the generic to the specialized design. “You already have the right machine for Excel. You bought it four years ago. Why are you buying another machine for Excel,” he said.
Later he discussed Gateway’s P-6831 FX notebook. The $1,249 computer was built with a slower processor than other notebooks in its price range, but used a faster Nvidia graphics chip instead, making it ideal for games and Windows Vista Aero interface. It proved a big hit and sold out quickly. “The fight will be over the visual computing part, not general computing,” said Huang.
Page 2 of 2
Mikako Kitagawa, principal analyst in Gartner’s client computing group, said enthusiasts know the difference between integrated and discrete graphics chipsets, but the mass market as yet doesn’t.
“The challenge for Nvidia is to educate OEMs and consumers because they
are looking at CPUs and thinking they need better CPUs [for better
performance]. Some people don’t really know about the importance of the
graphics card,” she told InternetNews.com.
“It’s really clear Intel is trying to get into Nvidia’s space with
integrated graphics to try and capture that market,” she went on to say. “As Jen-Hsun was saying, the whole world is going to visual and you need to have a better experience. But I don’t think Intel can get to Nvidia’s level. Nvidia is way too evolved.”
Ignoring AMD/ATI
Interestingly, while chip giant Intel took repeated verbal floggings, Huang didn’t mention ATI, Nvidia’s other rival, once.
AMD is working on an ambitious plan called Fusion, that will integrate ATI’s graphics technology with the chip maker’s CPU line. AMD bought ATI for $5.4 billion in 2006.
