A tipping point was January’s Consumer Electronic Show in Las Vegas and the just-finished Mobile World Congress in Barcelona. At both shows virtually every major vendor presented gadgets with Tegra 2 chips inside.
On the heels of those mobile strides, Huang is suggesting notebooks could be up next for the disruption. Echoing a similar sentiment by Steve Jobs, Huang told CNET future laptops will eventually replicate Apple’s design with the MacBook Air:
You’ll have trouble finding one that doesn’t look like the MacBook Air. I think the Macbook Air is a good mental image of what a clamshell laptop will look like. They’ll be thin because you won’t need any heat pipes, the fan, and extra batteries to lug around.
Reading between the lines, Huang suggested his company’s silicon would drive such ultra-thin notebook designs. You could be forgiven for rejecting his words as wishful thinking. After all, silicon giant Intel with its x86 processor architecture has been the dominate force in the personal computer market for a long time.
Then again, in the mobile space processor designs from UK-based ARM Holdings plc are a norm. A fabless silicon maker, ARM licenses its blueprints to semiconductor makers providing chips that power the vast majority of today’s smartphones and tablets, like chips from Qualcomm, Texas Instruments, Samsung and Nvidia.
Those chips typically integrate graphics, memory, processing and logic capabilities on a single die to keep the costs and, more importantly, power consumption at a minimum. Nearly all of them have one thing in common, an ARM-licensed processing core, which puts Intel at a big disadvantage. Worse, deciding it doesn’t want to be held hostage of Intel and be left out of the mobile race any longer, software giant Microsoft said the upcoming Windows 8 would support both x86 and ARM architecture, a historic decision enabling Windows to run on a non-Intel consumer processor for the first time ever.
Based on a monolithic kernel, Windows 8 will run on server, desktop and mobile gear, simplifying development and maintenance costs. The announcement plays into Nvidia’s favor because the Tegra chips too are engineered around ARM’s processor designs, augmented by Nvidia’s custom graphics architecture that leapfrogs competitors in terms of efficiency, performance and power consumption. The company’s aggressive road map includes the recently unveiled quad-core Tegra 3 code-named Kal El as in the birth name for Superman.
Promising a fivefold performance increase over Tegra 2, the new Tegra 3 with its twelve-core graphics unit is capable of stereoscopic 3D with resolutions of up to 2560 by 1600, enough to drive external 30-inch displays. The first consumer tablets powered by Tegra 3 are expected in August and smartphones by the end of this year.
A pair of clips included below highlight the capabilities of the new Tegra 3 platform. The first video shows off real-time decoding of a 2560 by 1440 video stream and the other demo sees the chip rendering a whooping 650 characters in Great Battles Medieval game. Coremark performance benchmark suggests that Tegra 3 is more powerful than the half a decade old desktop Core 2 Duo processors from Intel. The extra performance does not tax battery life, according to Nvidia.
A 2014 Tegra chip codenamed the Stark promises a 75x performance improvement over Tegra 2. It should coincide with ARM-enabled Windows 8 and a next-generation of mobile gadgets. The most important consequence to note here is that notable performance increases will have soon put Tegra on par with entry-level notebook processors from Intel, enabling Tegra chips to run ultra-thin Windows 8-powered notebooks akin to the MacBook Air.
This will give Nvidia a much-needed footing in the personal computing space which has been traditionally occupied by Intel. So far, Intel hasn’t produced a comparable power-savvy chip for smartphones or tablets and it seems incapable of matching Nvidia’s graphics expertise anytime soon. Nvidia is expected to capitalize on the added amount of attention it has garnered in the past few months and focus its attention on the Tegra architecture at the upcoming 2011 GPU Technology Conference scheduled to run at the McEnery Convention Center in San Jose, California from October 11-14.