Results 1 to 3 of 3

Thread: Nvidia announces TensorRT Hyperscale Platform at GTC Japan

  1. #1
    HEXUS.admin
    Join Date
    Apr 2005
    Posts
    27,608
    Thanks
    0
    Thanked
    1,705 times in 591 posts

    Nvidia announces TensorRT Hyperscale Platform at GTC Japan

    The "industry's most advanced inference accelerator" for voice, video, imaging.
    Read more.

  2. #2
    Long member
    Join Date
    Apr 2008
    Posts
    775
    Thanks
    32
    Thanked
    149 times in 110 posts
    • philehidiot's system
      • Motherboard:
      • Father's bored
      • CPU:
      • Cockroach brain V0.1
      • Memory:
      • Innebriated, unwritten
      • Storage:
      • Big Yellow Self Storage
      • Graphics card(s):
      • Semi chewed Crayola Mega Pack
      • PSU:
      • 20KW single phase direct grid supply
      • Case:
      • Closed, Open, Cold
      • Operating System:
      • Cockroach
      • Monitor(s):
      • The mental health nurses
      • Internet:
      • Please.

    Re: Nvidia announces TensorRT Hyperscale Platform at GTC Japan

    I dread to think what they're going to do with AI in medical imaging. I have enough problems trying to figure out how they've post-processed an image and what weirdness that's creating (or masking) without it getting all neural netty. If it's going to use AI to try and interpret imaging for quicker reports.... we've tried that and it's a disaster as people who aren't qualified to interpret the imaging rely on the report and assume it's correct. This is why for a lot of the qualitative stuff we dual report. What we need are reliable methods of automated quantification of things we normally assess qualitatively which are quick and easy to employ under pressure but let you make the interpretation based on relatively raw output data. That way you can spot errors quickly and easily.
    Last edited by philehidiot; 13-09-2018 at 11:37 AM. Reason: remove swearing - I got carried away

  3. Received thanks from:

    CAT-THE-FIFTH (13-09-2018)

  4. #3
    root Member DanceswithUnix's Avatar
    Join Date
    Jan 2006
    Location
    In the middle of a core dump
    Posts
    9,257
    Thanks
    438
    Thanked
    939 times in 800 posts
    • DanceswithUnix's system
      • Motherboard:
      • M5A-97 EVO R2.0
      • CPU:
      • FX-8350
      • Memory:
      • 16GB ECC 1333
      • Storage:
      • 500GB Linux, 1TB Games (Win 10)
      • Graphics card(s):
      • Asus Strix RX Vega 56
      • PSU:
      • 650W Corsair TX
      • Case:
      • Antec 300
      • Operating System:
      • Fedora 28 + Win 10 Pro 64 (yuk)
      • Monitor(s):
      • Benq XL2730Z 1440p + Samsung 2343BW 2048x1152
      • Internet:
      • Zen 80Mb/20Mb VDSL

    Re: Nvidia announces TensorRT Hyperscale Platform at GTC Japan

    Quote Originally Posted by philehidiot View Post
    I dread to think what they're going to do with AI in medical imaging. I have enough problems trying to figure out how they've post-processed an image and what weirdness that's creating (or masking) without it getting all neural netty. If it's going to use AI to try and interpret imaging for quicker reports.... we've tried that and it's a disaster as people who aren't qualified to interpret the imaging rely on the report and assume it's correct. This is why for a lot of the qualitative stuff we dual report. What we need are reliable methods of automated quantification of things we normally assess qualitatively which are quick and easy to employ under pressure but let you make the interpretation based on relatively raw output data. That way you can spot errors quickly and easily.
    There are some things they seem good at, and that should improve over time.

    https://www.theregister.co.uk/2018/0...mind_eye_scan/

Thread Information

Users Browsing this Thread

There are currently 1 users browsing this thread. (0 members and 1 guests)

Posting Permissions

  • You may not post new threads
  • You may not post replies
  • You may not post attachments
  • You may not edit your posts
  •