Nvidia 1080ti

Discussion in 'Distributed Computing' started by LANMAN, Mar 13, 2017.

  1. LANMAN

    LANMAN Platinum Member

    Joined:
    Oct 10, 1999
    Messages:
    2,800
    Likes Received:
    19
    Well, guess I should have followed my instincts when Nvidia had the 1080ti's for sale. :( Now everyone is out of stock. Dilemma.. do I get the 1080 since the prices have gone down or wait until the stock pile is replenished? :rolleyes:
     
  2. Loading...

    Similar Threads - Nvidia 1080ti Forum Date
    NVIDIA 1080 & 980TI in same system Distributed Computing Dec 6, 2016
    Does NVIDIA 376.19 Release work with Folding@home? Distributed Computing Dec 5, 2016
    Nvidia GPU Drivers and F@H Distributed Computing Nov 27, 2016
    PSA: Beware Nvidia's 364 drivers Distributed Computing Apr 23, 2016

  3. TennesseeTony

    TennesseeTony Elite Member

    Joined:
    Aug 2, 2003
    Messages:
    2,008
    Likes Received:
    399
    I don't know man, 35% increase in performance for 39% increase in power consumption...Hard sell for 24/7 operation.......

    I ebay'ed a 1080 for $437 just hours ago....
     
  4. Markfw

    Markfw CPU Moderator VC&G Moderator Elite Member
    Super Moderator

    Joined:
    May 16, 2002
    Messages:
    14,475
    Likes Received:
    709
    Well, I think the performance may be greater than 35% due to bonuses. When I get the one I want, I will post the results.
     
  5. StefanR5R

    StefanR5R Senior member

    Joined:
    Dec 10, 2016
    Messages:
    588
    Likes Received:
    229
    Re PPD increase vs. power increase: It very much depends on the point in the frequency/power curve at which you drive the card. I guess it should be possible to get more PPD/W out of 1080Ti at optimum operating point (compared to 1080 at its own optimum operating point) in F@H with its bonuses. But maybe not in BOINC projects with linear points/work.

    Finding the optimum op point is difficult though: AFAICS it involves playing with power target, clock offset, and fan speed. (OK, the last one is easy: Most is best for GPU perf/Watt.) And then measuring PPD takes a lot of time in a project like F@H where you get wildly differing work units.