NVIDIA’s car supercomputers are now fully autonomous.
Graphics chip company NVIDIA (nasdaq stock code: NVDA) launched the DRIVE PX platform in early 2015, a machine learning automotive supercomputers that provide advanced driving support. More than 200 companies, including some major carmakers, work with NVIDIA to DRIVE cars around the DRIVE PX system.
NVIDIA on Tuesday announced its next move to drive future self-driving cars. GPU technology conference held in Munich, NVIDIA launched DRIVE PX Pegasus, this is a pressure to DRIVE the PX system upgrade solution, NVIDIA claims it can provide complete level 5 unmanned vehicle dynamics. As processing power has soared, NVIDIA thinks Pegasus is the brain behind the self-driving car revolution.
NVIDIA drives PX flying horses.
Photo source: NVIDIA.
Pegasus can perform 320 trillion times per second deep learning operation, than DRIVE announced in early 2016 PX 2 increased tenfold. DRIVE PX 2 itself is 10 times faster than the original DRIVE PX, this means that the pegasus 100 times faster than the NVIDIA processing speed for the first time.
For self-driving cars to be safe and effective, a large number of video and sensor data needs to be processed in real time, with built-in redundancy to reduce error opportunities. NVIDIA claims that the demand for driverless cars is “50 to 100 times lighter than today’s most advanced cars”.
There are currently 225 partners working with DRIVE PX, and 25 of them are developing fully self-driving taxis. According to NVIDIA, Pegasus is a license plate size that can replace the currently required server and enterprise GPU racks to provide enough processing power for autonomous driving.
The Pegasus contains four processors, including two Xavier SoC (with eight custom CPU cores) and two next-generation independent gpus, which achieve this level of performance. The company will offer NVIDIA’s auto partners in the second half of 2018, including dozens of companies trying to bring self-driving taxis to market as soon as possible.
Some much-needed good news.
Pegasus’s announcement came after some not-so-good news from NVIDIA’s self-driving cars. Last September, Alphabet said that Waymo, its car company, had been working with Intel since 2009 to use the Xeon processor and chip giant’s field programmable gate array.
Intel is poised to become a major competitor in self-driving cars, paying more than $15 billion for MobileEye, a computer-vision company, earlier this year. The price Intel is willing to pay shows how serious the company is, and don’t miss out on a potentially huge, lucrative market.
Also in September, there was talk that Tesla, the electric car company, had partnered with NVIDIA’s rival, AMD, to develop self-developed self-drive chips. Nothing has been confirmed, but it will not be the first time a company has turned to a custom machine learning chip. Since last year, Alphabet’s GuGe has used thousands of custom tender-processing units in its data centers, claiming a significant increase in its GPU performance.
NVIDIA’s car business is still small, with revenues of less than 7% in the latest quarter. Pegasus won’t be able to get a partner’s support until the end of 2018, and the technology to produce vehicles could take years to roll out, but the news will have little impact on NVIDIA’s performance. But graphics chip companies continue to build the foundation for a booming car business.
10 stocks we’re better than Nvidia.
When investing genius David and Tom Gardner have a stock tip, it pays to listen. After all, they have been running communications for more than a decade, and motley, a stock consultant, has tripled the market. *
David and Tom have just revealed what they think are the top 10 stocks that investors are buying right now… Nvidia is not one of them! Yes, they think the 10 stocks are a better buy.