POST A COMMENT

7 Comments

Back to Article

  • EnsilZah - Monday, February 25, 2013 - link

    If all the computation is done on the server and all this card does is stream pixels, what's the point of having it connected to a motherboard rather than just having standalone box with some monitor and USB ports? Reply
  • giby - Monday, February 25, 2013 - link

    It's a solution for remote workstations, typically rack-mount, and not necessarily servers. The main benefit is that it allows IT to provide a workstation without the physical requirements of a large noisy system ever having to leave the data center, also an added security benefit. Reply
  • EnsilZah - Monday, February 25, 2013 - link

    Ah, I was under the impression this card was for the client side, I guess it's bad reading comprehension on my part and expecting a server-side card to be larger. Reply
  • Paulman - Monday, February 25, 2013 - link

    It's really interesting that Teridici is providing the silicon and that AMD is using them. I've never heard of PCoIP till now, but I *have* heard of Teridici. I play hockey with some of those guys on Wednesdays! Vancouver doesn't have too many silicon houses, so it's always cool to hear news from our area :) Reply
  • ArXiv76 - Sunday, March 10, 2013 - link

    HP's got them beat. Check out the HP t410.
    All in one unit that is powered from POE.
    Reply
  • RandyDGroves - Monday, March 11, 2013 - link

    The HP t410 is one of many compatible client side devices as well as the HP t310 which would be higher performing since it is based on a Teradici client ASIC (see http://www.teradici.com/where-to-buy/search-produc... for the complete list). The R5000 plugs into the workstation that needs to be accessed remotely and is an AMD unique solution. Reply
  • AG@FirePro - Monday, March 03, 2014 - link

    right - the R5000 is a pro GPU with the Teradici PCoIP encoding chip built in. It's a single (PCI) slot solution with pretty darn fast midrange 3D performance and 2G of fast GDDR5 memory -enough to power most mainstream CAD and M&E applications/workloads. But only taking up a single slot is the big deal here.

    This is crucial for maximum user density for virtual+direct-mapped remote dekstops. You can only support as many (virtual desktop) users as there are GPUs in the chassis.

    If you want to use NV (quadro or Geforce) with PCoIP infrastructure, then you are forced to use a Terdcici host (PCI) card for each NV device. This takes up twice the number of slots which is not good for IT guys who want maximum cost efficiency (via more user density per server node).

    -Adam G., Sapphire Technologies
    Reply

Log in

Don't have an account? Sign up now