Results 1 to 8 of 8

Thread: [News] Google benchmarks its Tensor Processing Unit (TPU) chips

  1. #1
    Join XS BOINC Team StyM's Avatar
    Join Date
    Mar 2006
    Location
    Tropics
    Posts
    16,492

    [News] Google benchmarks its Tensor Processing Unit (TPU) chips

    http://hexus.net/tech/news/industry/...nit-tpu-chips/

    As Google relies heavily on compute-intensive machine learning for its core activities it has designed and rolled out its own Tensor Processing Unit (TPU) accelerator chips in recent years. The need for such a chip became evident about six years ago with the growth in computational expense due to technologies like Google voice search taxing the deep neural net speech recognition systems.
    • On our production AI workloads that utilize neural network inference, the TPU is 15x to 30x faster than contemporary GPUs and CPUs.
    • The TPU also achieves much better energy efficiency than conventional chips, achieving 30x to 80x improvement in TOPS/Watt measure (tera-operations [trillion or 1012 operations] of computation per Watt of energy consumed).
    • The neural networks powering these applications require a surprisingly small amount of code: just 100 to 1500 lines. The code is based on TensorFlow, our popular open-source machine learning framework.





  2. #2
    Xtreme Enthusiast
    Join Date
    Aug 2008
    Posts
    889
    Comparing against a K80? Tumb on the scale or elbow?
    Intel 8700k
    16GB
    Asus z370 Prime
    1080 Ti
    x2 Samsung 850Evo 500GB
    x 1 500 Samsung 860Evo NVME


    Swiftech Apogee XL2
    Swiftech MCP35X x2
    Full Cover GPU blocks
    360 x1, 280 x1, 240 x1, 120 x1 Radiators

  3. #3
    Xtreme Enthusiast
    Join Date
    Dec 2005
    Posts
    766
    Given the point in which these devices were determined to be made, K80 and Haswell v3 were top tier. The whitepaper is quite interesting in its approach to WHY, within a given thermal envelope, a TPU makes more sense. the ancillary discussion of FPGA vs. ASIC is interesting as well.
    Heat: 50 - 0 - 0 under "Argus333"

  4. #4
    Xtreme Guru
    Join Date
    May 2007
    Location
    Ace Deuce, Michigan
    Posts
    3,955
    It also helps that the logic instructions were probably highly optimized to the type of work Google does. Nothing wrong with that (in fact it's really smart), but it doesn't necessarily mean a TPU is all that great for other applications.
    Quote Originally Posted by Hans de Vries View Post

    JF-AMD posting: IPC increases!!!!!!! How many times did I tell you!!!

    terrace215 post: IPC decreases, The more I post the more it decreases.
    terrace215 post: IPC decreases, The more I post the more it decreases.
    terrace215 post: IPC decreases, The more I post the more it decreases.
    .....}
    until (interrupt by Movieman)


    Regards, Hans

  5. #5
    Xtreme Mentor
    Join Date
    Aug 2006
    Location
    HD0
    Posts
    2,646
    So nVidia's growth is mainly primed on doing AI/ML work...
    What's to stop something like this from impacting nVidia's server segment? I can see something like this dominating when it comes to doing DNNs.

  6. #6
    Xtreme Guru
    Join Date
    May 2007
    Location
    Ace Deuce, Michigan
    Posts
    3,955
    Quote Originally Posted by xlink View Post
    So nVidia's growth is mainly primed on doing AI/ML work...
    What's to stop something like this from impacting nVidia's server segment? I can see something like this dominating when it comes to doing DNNs.
    My best guess is lack of driver support and dev environment/toolboxes. CUDA is finally well established; switching to a TPU without a quality dev environment/toolbox would probably not be worth the headache right now.
    Quote Originally Posted by Hans de Vries View Post

    JF-AMD posting: IPC increases!!!!!!! How many times did I tell you!!!

    terrace215 post: IPC decreases, The more I post the more it decreases.
    terrace215 post: IPC decreases, The more I post the more it decreases.
    terrace215 post: IPC decreases, The more I post the more it decreases.
    .....}
    until (interrupt by Movieman)


    Regards, Hans

  7. #7
    Xtreme Enthusiast
    Join Date
    Aug 2008
    Posts
    889
    Intel 8700k
    16GB
    Asus z370 Prime
    1080 Ti
    x2 Samsung 850Evo 500GB
    x 1 500 Samsung 860Evo NVME


    Swiftech Apogee XL2
    Swiftech MCP35X x2
    Full Cover GPU blocks
    360 x1, 280 x1, 240 x1, 120 x1 Radiators

  8. #8
    One-Eyed Killing Machine
    Join Date
    Sep 2006
    Location
    Inside a pot
    Posts
    6,334
    Pwned
    Coding 24/7... Limited forums/PMs time.

    -Justice isn't blind, Justice is ashamed.

    Many thanks to: Sue Wu, Yiwen Lin, Steven Kuo, Crystal Chen, Vivian Lien, Joe Chan, Sascha Krohn, Joe James, Dan Snyder, Amy Deng, Jack Peterson, Hank Peng, Mafalda Cogliani, Olivia Lee, Marta Piccoli, Mike Clements, Alex Ruedinger, Oliver Baltuch, Korinna Dieck, Steffen Eisentein, Francois Piednoel, Tanja Markovic, Cyril Pelupessy (R.I.P. ), Juan J. Guerrero

Bookmarks

Bookmarks

Posting Permissions

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