If you thought NVIDIA Corporation's (NASDAQ:NVDA) Tegra K1 was fast, you're gonna love this.
On Sunday -- just as expected ahead of CES 2015 -- the graphics chip specialist unveiled the Tegra X1, a 20-nanometer mobile super chip with a 256-core Maxwell GPU and 8 CPU cores, the latter of which includes 4 ARM Cortex A57s and 4 Cortex A53s. Consequently, NVIDIA says the Tegra X1 offers twice the performance of its 192-core Tegra K1, which was launched exactly one year ago on the eve of last year's Consumer Electronics Show.
Specifically, the Tegra X1 packs over one full teraflops of computing power, offers 1.3 gigapixels of camera throughput, and can deliver 60 frames per second of 4K video. For perspective, that's faster than ASCI Red, which was the world's first teraflops system and the fastest supercomputer up until the year 2000. What's more, ASCI Red required 1,600 square feet of floor space and used 500,000 watts of power. Each thumbnail-sized Tegra X1 draws under 10 watts.
So what does NVIDIA plan to do with all that efficient computational might?
It's no mystery the company would love to pack the Tegra X1 into as many high-end tablets and smartphones as possible. But while it's likely that the Tegra X1 will follow in its predecessor's footsteps by earning some solid design wins along the way, fellow Fool Ashraf Eassa has already pointed out that the market for the X1 will remain constrained, as it's effectively limited to non-Apple, non-Samsung devices.
Supercomputing in your car
That's where NVIDIA's second CES announcement comes into play: NVIDIA DRIVE automotive computers.
For now, NVIDIA DRIVE comes in two varieties: NVIDIA DRIVE CX and NVIDIA DRIVE PX.
The first, DRIVE CX, is described as a "cockpit computer" for running next-gen digital instrument clusters and infotainment systems inside the car. But that's also an area where NVIDIA already has a significant presence, with previous Tegra-series chips already being used in over 6.2 million vehicles as of only a few months ago. However, available with either the Tegra X1 or the Tegra K1, Drive CX can power up to 16.8 million pixels across multiple displays -- more than 10 times the number of pixels that current-model cars are capable of crunching.
Even more enticing is NVIDIA DRIVE PX, an auto-pilot platform powered by dual Tegra X1 processors and "a host of advanced algorithms created by NVIDIA's computer vision engineering team." DRIVE PX also supports inputs for up to 12 high-res cameras, the combined video streams from which it uses to create an incredibly detailed real-time obstacle map as the vehicle drives, while at the same time controlling the car's throttle, brake, and steering.
NVIDIA DRIVE PX not only receives and processes the images collected by the cameras, but it also interprets them to a startling degree, down to individual speeds, types, and statuses of objects. As a result, the system could be programmed, for example, to more effectively respond to much more elaborate scenarios, such as anticipating the increased likelihood that a service or delivery truck might pull into traffic from its parked position in a side lane.
That said, the automotive chip market is obviously still in its infancy, with even the most optimistic of industry players looking at a 5 to 10-year time frame before their self-driving vehicle efforts are expected to have a meaningful impact.
At the time time, assuming NVIDIA's existing auto industry partners embrace this new tech, NVIDIA DRIVE also brings investors that much closer to NVIDIA CEO Jen-Hsun Huang's prediction of having each and every vehicle on the road contain several NVIDIA processors to handle computationally intense tasks. Considering more than 80 million vehicles were sold worldwide last year, NVIDIA could easily be sitting on a multi-billion dollar market just ripe for the picking over the long term.