Nvidia CEO Jen-Hsun Huang’s chip shot

August 27, 2008, 6:58 PM UTC

SAN JOSE, Calif. – Nvidia CEO Jen-Hsun Huang shows up to our chat in a black muscle shirt and jeans, and a casual attitude to match. The co-founder of the world’s largest maker of dedicated graphics chips doesn’t pull a lot of punches – but he does want to clear up something about his widely covered war of words with Intel.

In statements earlier this year, Intel , which is nearly 10 times Nvidia’s size, implied that Nvidia’s chips are becoming obsolete, which rattled investors and infuriated Huang. When an analyst raised the issue with Huang at a meeting this spring, asking if it was OK to open up that can of worms, Huang shot back, “Sure, let’s open up a can of whup-ass” – and then continued with a spirited rebuttal.


But really, Huang insists, the whole thing wasn’t a big deal. “Those kinds of comments, I thought, from a large, global company like Intel, were probably irresponsible, considering the platform that they have,” Huang says.  “They issued a public apology.”

Still, everything between the two companies isn’t champagne and roses. At Nvidia’s first annual NVISION conference this week in Silicon Valley, Huang is presenting a vision of the future that clashes with Intel’s.

Intel imagines its microprocessors dominating smartphones and business computers. In Huang’s view, Nvidia’s graphics experience will give it inroads into mobile devices with its Tegra chips, and into high-powered parallel computing with its CUDA software environment. Those products are going after the same markets that Intel has targeted with its mobile Atom processor, and its upcoming Larrabee chips.

“These companies that are working together to serve their mutual customers are also competing with each other,” says Shane Rau, analyst with IDC. “They have two very different points of view, and that leads to conflict. They both want to assert the superiority of their technology.”

That comes out as I ask Huang about the differences between Intel and Nvidia’s chips for iPhone-like devices. As Intel explained to me last week, with its Atom chips it decided to abandon the low-wattage ARM platform that’s popular in phones today, wagering instead that over time engineers could build a low-wattage version of Intel’s powerful PC chips that would work in phones. Nvidia has made the opposite bet, basing its Tegra mobile chips on ARM.

“Comparing Atom and Tegra is a little bit like comparing something with a V12, 1,000 horsepower to a hybrid.” Huang says. “What the world wants from us in the mobile device space is something like a hybrid.”

Meanwhile at the conference, a few developers also showed what they’re doing with Nvidia’s CUDA programming environment. CUDA lets software tap into the power of graphics chips for tasks that traditionally have been left to Intel’s microprocessors – at times yielding eye-popping results.

For example, OptiTex has written software code that simulates how various fabrics move in the real world, which allows computers to show virtual models twirling and bending as the virtual clothes wrinkle and stretch on their bodies.

This is more than a cute simulation – fashion makers can plug in data to mimic a specific type of silk or leather, or to change the dimensions of the model. Matt Bakhoum, who works in marketing for OptiTex, says the company is using CUDA to make the software take advantage of Nvidia’s graphics processors – but that they are also keeping an eye out for Intel’s Larrabee, which won’t be available for a couple of years.