Nvidia's 2015 CES press conference | CNET

Nvidia's 2015 CES press conference

The graphics chipmaker will talk up its latest efforts in automotive and mobile.

  • Shara Tibken 1/5/2015 4:34:26 AM
    Lots of in-depth car talk. I haven't had a car for about seven years (since I left the Midwest after college), so a lot of this stuff, I've never used.
  • Ben Fox Rubin 1/5/2015 4:34:42 AM
    As more sensors are replaced by cameras, it reduce costs and energy consumed from radar technology.
  • James Martin 1/5/2015 4:34:43 AM
  • Ben Fox Rubin 1/5/2015 4:34:54 AM
    This technology now only available in high-end cars.
  • Ben Fox Rubin 1/5/2015 4:35:19 AM
    Might be able to do "something really magical," Huang said -- such as self-driving cars.
  • Ben Fox Rubin 1/5/2015 4:35:28 AM
    But still much needs to be solved, he said.
  • Ben Fox Rubin 1/5/2015 4:35:46 AM
    Need all cameras connected to a central processor. Car needs to become self-aware.
  • Ben Fox Rubin 1/5/2015 4:36:01 AM
    "It needs to see what is happening around it," Huang said.
  • Ben Fox Rubin 1/5/2015 4:36:18 AM
    Car also has to plan its path based in what situation it's in.
  • James Martin 1/5/2015 4:36:39 AM
  • Ben Fox Rubin 1/5/2015 4:36:45 AM
    Car also must become software updatable.
  • James Martin 1/5/2015 4:36:54 AM
  • Ben Fox Rubin 1/5/2015 4:37:03 AM
    Would need a super-computing processor.
  • Ben Fox Rubin 1/5/2015 4:37:09 AM
    Introduced the Drive PX
  • James Martin 1/5/2015 4:37:14 AM
  • Shara Tibken 1/5/2015 4:37:19 AM
    We're more than half an hour into the presentation and still a ton on auto.
  • Ben Fox Rubin 1/5/2015 4:37:32 AM
    An auto-pilot car computer
  • Ben Fox Rubin 1/5/2015 4:37:41 AM
    Uses dual Tegra X1 chips
  • James Martin 1/5/2015 4:37:48 AM
  • Ben Fox Rubin 1/5/2015 4:37:58 AM
    2.3 teraflops of mobile supercomputing
  • Ben Fox Rubin 1/5/2015 4:38:23 AM
    Said will use a computer vision technology.
  • James Martin 1/5/2015 4:38:54 AM
  • Ben Fox Rubin 1/5/2015 4:39:03 AM
    Could be many cameras all around the car, Huang said, with several in the front, several around the sides and back, and on the mirrors.
  • Ben Fox Rubin 1/5/2015 4:39:15 AM
    All those cameras could then feed data into the Drive PX.
  • Ben Fox Rubin 1/5/2015 4:39:40 AM
    Could record all info into a data recorder in car, Huang said.
  • James Martin 1/5/2015 4:40:25 AM
  • Ben Fox Rubin 1/5/2015 4:40:26 AM
    Now have to figure out the car's environment and keep driver out of harm's way, Huang said, using this technology.
  • Shara Tibken 1/5/2015 4:40:56 AM
    Nvidia has really struggled to get Tegra into mobile devices, but it wants to own auto. What Huang has been outlining for the past 40 minutes is what's going to help it do that.
  • James Martin 1/5/2015 4:41:12 AM
  • Ben Fox Rubin 1/5/2015 4:41:15 AM
    One of the most important things announcing is a computer vision system, Huang said.
  • Shara Tibken 1/5/2015 4:41:49 AM
    Unlike smartphones, which are typically used for just a couple years, cars have much longer life cycles. Once you're designed in and go through certification, etc, it's more stable. It's the same with home appliances. Once you're in, you're often in for awhile.
  • James Martin 1/5/2015 4:42:02 AM
  • Ben Fox Rubin 1/5/2015 4:42:04 AM
    Need to detect a lot more things for a car to become more aware of its surroundings, Huang said.
  • Shara Tibken 1/5/2015 4:42:23 AM
    It will be interesting to hear at CES what automakers have planned. The car is really being seen as one of the next big "computers." Lots of innovation taking place there.
  • Ben Fox Rubin 1/5/2015 4:42:32 AM
    He discussed "deep learning."
  • James Martin 1/5/2015 4:42:38 AM
  • James Martin 1/5/2015 4:43:01 AM
  • Shara Tibken 1/5/2015 4:43:09 AM
    Samsung, Apple and other big electronics giants also recognize how important auto is becoming. They're also targeting that market, though in a slightly different way than a chip provider like Nvidia.
  • Ben Fox Rubin 1/5/2015 4:43:26 AM
    Huang said GPUs were found in a study to be one of the most important technologies needed for "deep learning."
  • Ben Fox Rubin 1/5/2015 4:44:39 AM
    Huang said GPUs help machines more effectively detect more images.
  • Ben Fox Rubin 1/5/2015 4:45:21 AM
    He's describing how a "deep neural network sees" -- can train it to see around it.
  • James Martin 1/5/2015 4:45:23 AM
  • Ben Fox Rubin 1/5/2015 4:45:52 AM
    Extremely complicated to train these machines, but once it works it works, Huang said.
  • Ben Fox Rubin 1/5/2015 4:47:01 AM
    So, this system is based on biology, with machines' "neurons" learning the images of specific objects.
  • Ben Fox Rubin 1/5/2015 4:47:48 AM
    These systems can build on themselves so machines can self-learn.
Powered by ScribbleLive Content Marketing Software Platform

This week on CNET News

See all