Stories
Slash Boxes
Comments

SoylentNews is people

posted by Fnord666 on Wednesday February 28 2018, @02:13AM   Printer-friendly
from the is-that-an-AI-in-your-pocket-or.... dept.

Arthur T Knackerbracket has found the following story:

Neural networks are powerful things, but they need a lot of juice. Engineers at MIT have now developed a new chip that cuts neural nets' power consumption by up to 95 percent, potentially allowing them to run on battery-powered mobile devices.

Smartphones these days are getting truly smart, with ever more AI-powered services like digital assistants and real-time translation. But typically the neural nets crunching the data for these services are in the cloud, with data from smartphones ferried back and forth.

That's not ideal, as it requires a lot of communication bandwidth and means potentially sensitive data is being transmitted and stored on servers outside the user's control. But the huge amounts of energy needed to power the GPUs neural networks run on make it impractical to implement them in devices that run on limited battery power.

Engineers at MIT have now designed a chip that cuts that power consumption by up to 95 percent by dramatically reducing the need to shuttle data back and forth between a chip's memory and processors.

-- submitted from IRC


Original Submission

 
This discussion has been archived. No new comments can be posted.
Display Options Threshold/Breakthrough Mark All as Read Mark All as Unread
The Fine Print: The following comments are owned by whoever posted them. We are not responsible for them in any way.
  • (Score: 3, Interesting) by Schafer2 on Wednesday February 28 2018, @08:51AM (1 child)

    by Schafer2 (348) on Wednesday February 28 2018, @08:51AM (#645043)
    95% more efficient compared to their analog predecessors. See the MIT source article here:

    http://news.mit.edu/2018/chip-neural-networks-battery-powered-devices-0214 [mit.edu].

    Not compared to digital neural networks, which are still much more efficient overall, according to a review by the Skeptics' Guide to the Universe podcast. About zero chance you are going to have an analog NN in your phone anytime soon.

    https://www.theskepticsguide.org/podcast/sgu/658 [theskepticsguide.org]

    See show notes, Science or Fiction Item # 1
    Starting Score:    1  point
    Moderation   +1  
       Interesting=1, Total=1
    Extra 'Interesting' Modifier   0  
    Karma-Bonus Modifier   +1  

    Total Score:   3  
  • (Score: 1, Interesting) by Anonymous Coward on Wednesday February 28 2018, @11:31AM

    by Anonymous Coward on Wednesday February 28 2018, @11:31AM (#645091)

    > About zero chance you are going to have an analog NN in your phone anytime soon.

    You'll see it within 5y for image preprocessing and 10y for general processing. Both are in the pipeline now. The video imaging stuff is pretty mindblowingly fluid (besides phone cameras, another segment heavily R&D'ing with this kind of processor is VR) but then again so were HDR and anti-shake and smile-capture and so on, for the first week. So not "soon" like "next week," no. But yes, fast and ultra low power analog NN chips, coming fast.