Perfect SLI Scaling w/ rFactor 2 - Great Job ISI!

Discussion in 'General Discussion' started by Spinelli, Jul 12, 2015.

  1. Spinelli

    Spinelli Banned

    Joined:
    Jan 28, 2012
    Messages:
    5,290
    Likes Received:
    32
    I see what you're saying, Noel. What I think WhiteShadow actually meant was the original style multiview VS the new style multiview. I thereore think he meant to write "UseSubViewParams" rather than "SubViews".

    As you pointed out, I can assure everyone I am indeed using "proper" multiview rather than one large camera.

    WhiteShadow, can you confirm exactly what you meant? "UseSubViewParams" or "SubViews"?
     
    Last edited by a moderator: Jul 14, 2015
  2. Lazza

    Lazza Registered

    Joined:
    Oct 5, 2010
    Messages:
    12,345
    Likes Received:
    6,572
    SLI scaling is about non-SLI vs SLI performance (and, secondarily, GPU utilisation).

    Assuming your figures are correct, if SLI gave you 70fps in that scenario and non-SLI gave you 40fps, SLI scaling is good. If particular graphics settings gave you 5fps without SLI and 10fps with SLI, then performance sucks but SLI scaling is (remarkably) good. To use your words, the "SLI experience is broken" but only because the experience overall is broken - SLI is actually doing its job very well.

    So - again - in an SLI scaling context, the results Spinelli has talked about are good.

    If you point out that your results are 70fps (SLI) and something like 60-65fps (non-SLI), then I think we can agree the SLI scaling isn't good for you or for that scenario. But just coming in and pointing out that Spinelli is getting higher fps than he would get with different detail/graphical settings is not contributing at all to an SLI scaling discussion.

    And as others have pointed out, increasing the CPU load will only tend to mask an SLI performance increase, just like it will mask a graphics card upgrade.
     
  3. WhiteShadow

    WhiteShadow Registered

    Joined:
    Feb 16, 2015
    Messages:
    681
    Likes Received:
    3
    Sorry I made typo I meant subviews=0
     
  4. WhiteShadow

    WhiteShadow Registered

    Joined:
    Feb 16, 2015
    Messages:
    681
    Likes Received:
    3
    I meant subviews=0 but made typo sorry about that :(

    There is no way 970 SLI beats 780Ti SLI with 60 fps same settings, I get 70 fps with my system. I know that you have PCI-E 2 x16 and better CPU but in gaming you cain maybe 5 fps vs my system when game don`t have true quad core support.

    Your PC spes. 3x BenQ XL2720T>3x1920>5760. Post #1 > 6000 > Nvidia bezel correction > v1&v2 > 120 http://gaming.benq.com/gaming-monitor/xl2720t/specification/#skip
     
    Last edited by a moderator: Jul 14, 2015
  5. WhiteShadow

    WhiteShadow Registered

    Joined:
    Feb 16, 2015
    Messages:
    681
    Likes Received:
    3
    PS. Single 780Ti gives with same settings & triple screens 100fps, Titan X 140fps :)
     
  6. Lazza

    Lazza Registered

    Joined:
    Oct 5, 2010
    Messages:
    12,345
    Likes Received:
    6,572
    So identical settings gives you 70fps in SLI and 100fps non-SLI?
     
  7. WhiteShadow

    WhiteShadow Registered

    Joined:
    Feb 16, 2015
    Messages:
    681
    Likes Received:
    3
    Yes
     
  8. Noel Hibbard

    Noel Hibbard Registered

    Joined:
    Oct 5, 2010
    Messages:
    2,744
    Likes Received:
    40
    Without any AI correct? Also, do you have the new multi core threading enabled? I'm curious if that has any impact. If I remember correctly you are one of the people that had bad performance with the new threading.
     
  9. Spinelli

    Spinelli Banned

    Joined:
    Jan 28, 2012
    Messages:
    5,290
    Likes Received:
    32
    You're right :) I've owned 5 different 780 Tis and I agree the 780 Ti is definitely a bit faster than the 970, in-fact I regret selling my 780 Tis for the 970s.

    When comparing to your tests, you have to remember something specifically regarding rFactor 2; rFactor 2, unlike almost every other game, mysteriously gains massively when going to PCI-E 3.0 @ 16x. Most games may benefit a few percent - let's say around 5% - but repeated tests (from a year or so ago) show that RF2 can gain anywhere from 10-40 %. That's massive and, to be frank, not the normal but it is what it is.

    Or maybe it's something to do with the model of the cards, who knows.
     
    Last edited by a moderator: Jul 15, 2015
  10. WhiteShadow

    WhiteShadow Registered

    Joined:
    Feb 16, 2015
    Messages:
    681
    Likes Received:
    3
    Yes no AI. I do not remember to have any issue with multi core I have i7 4790k.
     
  11. WhiteShadow

    WhiteShadow Registered

    Joined:
    Feb 16, 2015
    Messages:
    681
    Likes Received:
    3
    YES I remember reading that thread some time back. Wasn't that gain (10-40%) with single card? I think SLI is different story.
    Please don`t get me wrong I am not suspecting you to lying, I just like to know why my system is performing so much worse then yours. :)
     
  12. Spinelli

    Spinelli Banned

    Joined:
    Jan 28, 2012
    Messages:
    5,290
    Likes Received:
    32
    New Test - (older content)
    Same gfx settings as Test #3 from my OP (all graphics @ max except FXAA disabled and AA @ Lvl 2)
    - Historic Eve F3
    - Historic Spa
    - In garage
    - Night time
    - Rain


    Results

    1x GPU = 60 fps
    2x GPU (SLI) w/ default RF2 comp. bits = 52 fps
    2x GPU (SLI) w/ "Lego" comp. bits = 103 fps



    Nope, same story with SLI. SLI lost a lot of frames as well. I did SLl tests (in Nvidia 3D Vision mode since that is what I normally play with and was the only time SLI would ever work/scale properly prior to b982) and had the same PCI-E results. One time I even forgot to put the PCI-E speeds back to normal and was wondering why my framerates where so low until I finally remembered.

    I wish I could tell you why your getting different results than me. I wouldn't think a 780 Ti instead of a 970 would be the difference between good SLI scaling and poor SLI scaling. Are you using the exact same driver version, graphics settings? Try everything back at default from the Nvidia control panel then in Inspector only change the SLI comp bits, nothing else. I know this sounds stupid but don't forget to click "Apply changes" in Inspector (top right) and make sure you change the "SLI compatibility bits" not the "SLI compatibility bits (DX1x)".
     
    Last edited by a moderator: Jul 17, 2015
  13. Kristoff Rand

    Kristoff Rand Registered

    Joined:
    Dec 23, 2011
    Messages:
    1,055
    Likes Received:
    10
    I'm going to give it a go and reply back with results.
     
  14. Kristoff Rand

    Kristoff Rand Registered

    Joined:
    Dec 23, 2011
    Messages:
    1,055
    Likes Received:
    10
    HOLY SHIAT... It worked.

    before this I had 50fps with sli off and like 25 with it on.

    I went into nvidia inspector and reset my rfactor2.exe profile to default.
    selected the lego sli bits and applied the changes.

    started up and now I have 106fps with sli on.

    finally...

    I'm going back to spa to play with GT3's...

    Thanks Spinelli...
     
    Last edited by a moderator: Jul 15, 2015
  15. Kristoff Rand

    Kristoff Rand Registered

    Joined:
    Dec 23, 2011
    Messages:
    1,055
    Likes Received:
    10
  16. Noel Hibbard

    Noel Hibbard Registered

    Joined:
    Oct 5, 2010
    Messages:
    2,744
    Likes Received:
    40
    So you have SubViews=0. Spinelli had SubViews=2 so it looks like it scales about the same with MultiView on or off.
     
  17. Noel Hibbard

    Noel Hibbard Registered

    Joined:
    Oct 5, 2010
    Messages:
    2,744
    Likes Received:
    40
    We really need a reporting tool for these sorts of threads. A tool that gathers up the system specs, config.ini, player.json and Nvidia settings. I took a stab at this a while back and stopped when I found out that Nvidia stores their settings in some sort of database that I can't easily read. It sucks see people report different results but never knowing for sure exactly what settings they are running. If everyone had a consistent way of reporting results and settings we may stand a better chance of figuring out why some people get completely different results.

    Edit: I guess I could just use Nvidia Inspector to export the settings. I need to look into this.
    Edit2: I don't see any command line options to automate the dumping of the rFactor2.exe profile.
     
  18. Kristoff Rand

    Kristoff Rand Registered

    Joined:
    Dec 23, 2011
    Messages:
    1,055
    Likes Received:
    10
    no, you're right... it need to be all settings. wheel, video, car... whatever. and have the ability to turn off whichever you don't want. Strange... I was just thinking of this today. We were discussing wheel settings and how it would be nice to transfer some settings without losing other settings.
     
  19. rodger

    rodger Registered

    Joined:
    Apr 9, 2013
    Messages:
    43
    Likes Received:
    0
    Makes a huge difference on my system. Thank you
     
  20. WhiteShadow

    WhiteShadow Registered

    Joined:
    Feb 16, 2015
    Messages:
    681
    Likes Received:
    3
    Well, 106 fps at garage is maybe fine if you have 60hz/60fps but it is not enough if you have 120hz or 144hz monitors. You can just imagine how it is when there is more cars at start of the race when your fps is going to drop from 120/144 fps to 55-60 fps.
    I have EVGA GeForce GTX 780Ti Classified Kingpin SLI and it is easy for me to get locked 120fps in any other game with similar settings as OP is using in this thread.

    EVGA GeForce GTX 780Ti Classified Kingpin SLI > OC, GPU 1345mhz, Ram 3700mhz
    View attachment 17354

    :(
     

Share This Page