[Official] DX9 vs DX11 performance difference

Discussion in 'Technical & Support' started by 4thworld, May 4, 2017.

Thread Status:
Not open for further replies.
  1. SPASKIS

    SPASKIS Registered

    Joined:
    Sep 7, 2011
    Messages:
    3,155
    Likes Received:
    1,426
    I am a mechanical engineer. I have no knowledge about CPU or GPU and I have never admitted so. My specs have been available from the very beginning. Nobody has ever mentioned my hardware specs to be the reason for fps drop.

    You say that DX11 requires extra horsepower, so according to your theory everybody should notice a performance drop.

    Please, enlighten us with your knowledge and quantify this extra horsepower. Which extra requirements does DX11 need?

    You apparently didn't read any of my reports. I am quite sure that the fps drop is related to fullscreen not properly working and not to my specs.

    Your theory does not explain that I get the same fps in all three video modes in DX11 and 30% more fps in fullscreen wrt borderless or windowed in DX9.
     
    demerzel likes this.
  2. Lazza

    Lazza Registered

    Joined:
    Oct 5, 2010
    Messages:
    12,345
    Likes Received:
    6,572
    So assuming your main issue is fullscreen not working in dx11, I don't think there's much point you posting about fps losses in dx11. You've already shown that in borderless/windowed you get better fps in dx11, which then removes any doubts about hardware etc giving you unusual results. You just don't seem to have working fullscreen.
     
    SPASKIS likes this.
  3. SPASKIS

    SPASKIS Registered

    Joined:
    Sep 7, 2011
    Messages:
    3,155
    Likes Received:
    1,426
    https://forum.studio-397.com/index....formance-difference.55385/page-11#post-897871
    In this post you can see the different behaviour of the three display modes when working on DX9 and DX11.

    As you say, in DX11 I get an improvement for borderless and windowed wrt DX9. However I don't get higher values for fullscreen as I did with DX9. In DX11 the three modes show the same fps. That's why I presume Fullscreen behaving as if it was actually Borderless in DX11.

    The fact that available resolutions and refresh frequencies are different between DX9 and DX11 also suggests that something's is not working as expected.
     
  4. Ari Antero

    Ari Antero Registered

    Joined:
    Jul 27, 2012
    Messages:
    1,882
    Likes Received:
    829
    rFactor 2 DX11 switches to borderless window mode when you alt tab. There is fullscreen mode in dx11, it just don`t work properly.
    However it is possible to play rFactor 2 DX11 in fullscreen mode if close or minimize all open apps, tick Auto Close Launcher in settings menu. ;)
     
    SPASKIS and dadaboomda like this.
  5. Stan

    Stan Registered

    Joined:
    Jan 18, 2016
    Messages:
    166
    Likes Received:
    88
    ... and we are numbers with the same fullscreen/stuttering problem... why does S397 can't simply make an official statement regarding this awfully spread problem? Nobody will moan anymore...
     
    Last edited: Oct 9, 2017
    dadaboomda and SPASKIS like this.
  6. Mulero

    Mulero Registered

    Joined:
    Jan 16, 2012
    Messages:
    351
    Likes Received:
    153
    These are my results with the last Build released at the end of July. Performance has never been so bad in DX11. 18% less than DX9.

    dx9 vs dx11.png

    Clocks are very unimportant in rF2, tested at 4.8Ghz and at 1.4Ghz and the performance is pretty much the same.

    7600k @4.8
    Frames, Time (ms), Min, Max, Avg
    17642, 270974, 44, 85, 65,106

    7600k @ 3.4
    Frames, Time (ms), Min, Max, Avg
    17570, 270865, 44, 86, 64.866

    7600k @ 2.2
    Frames, Time (ms), Min, Max, Avg
    17639, 274562, 44, 82, 64.244

    7600k @ 1.4
    Frames, Time (ms), Min, Max, Avg
    17245, 270958, 42, 82, 63.645
     
    Emery and SPASKIS like this.
  7. patchedupdemon

    patchedupdemon Registered

    Joined:
    Jan 3, 2017
    Messages:
    1,602
    Likes Received:
    1,151
    Wouldn't that suggest a bottle neck then in systems where clock rates do not effect fps,I know iracing is very independent on clock frequency of cpu,why is rf2 different,I'm not challenging your claim just wanting to know more about it.

    Edit just checked your sig,your gpu would suggest to me that it bottlenecks your cp,you have a great cpu but that gpu makes your system not balanced,it's stopping you getting the most out of your cpu,I've seen this too in iracing

    check to see the loads on both cpu and gpu in while in game,if your cpu isnt running at high loads,your gpu is holding it back,maybe it is the game,but ive seen this too many time in iracing and everytime its the gpu holding back the cpu
     
    Last edited: Oct 10, 2017
  8. Ari Antero

    Ari Antero Registered

    Joined:
    Jul 27, 2012
    Messages:
    1,882
    Likes Received:
    829
    I can see that you have AMD RX 570 4GB which should run rFactor 2 in single screen fine and I haven't seen people who has AMD cards complaining much here in forums.
    I know that you have been around a long time but I must ask if you are aware about fullscreen issue (My post #344) and how to run rFactor 2 in Borderless or fullscreen mode?
    Yes I know that I am like a parrot but sometimes gaming is like sex and it is better to be safe then sorry :)
     
    Mulero likes this.
  9. SPASKIS

    SPASKIS Registered

    Joined:
    Sep 7, 2011
    Messages:
    3,155
    Likes Received:
    1,426
    I have Launcher Auto Close Selected and I don't have any open apps in the windows bar apart from rF2 itself. However I seem to be in permanent borderless mode.

    Are you sure your method applies to every computer?
    How can I effectively know whether I am in FullScreen or Borderless?
     
  10. Ari Antero

    Ari Antero Registered

    Joined:
    Jul 27, 2012
    Messages:
    1,882
    Likes Received:
    829
    Well, I can not know how and which apps people are using but 3 PC`s I have to test with, Launcher Auto Close or alt + tab is working fine, how you can test if you are in Borderless or Fullscreen I don`t know.
    With 1080Ti Borderless vs fullscreen DX11 I am not able to see any performance difference, fps is equal. 780Ti and 680 has about 15% lower performance in borderless mode. @stonec mentioned that Kepler GPU he tested had 15% FPS lost vs Pascal GPU. With my Kepler GPU`s performance is 10-15% when game is in Borderless DX11. Fullscreen mode Pascal vs Kepler has no performance lost DX9 vs DX11 with hardware I own.
    However I am not happy when game does not work proper in fullscreen mode simply becourse Windowed/Borderless is running trough OS Syncbuffer causing input lag,stutters etc. and it is logical that I7-6950X , 10 core, 20 threads at 5,4 ghz, DDR4@3000mhz with 1080Ti has equal performance with Borderless and fullscreen DX11 but 680/I7-3770k and 780Ti/I7-4770k are struggling when game is in borderless DX11 mode.

    ;)
     
    Last edited: Oct 10, 2017
    SPASKIS likes this.
  11. Mulero

    Mulero Registered

    Joined:
    Jan 16, 2012
    Messages:
    351
    Likes Received:
    153
    I do not know if it's good or bad but I'm not having any changes in any way (fullscreen, borderless, windowed) by pressing alt + tab. FPS are always the same, 91 in pits, except when you see the image of rF2 but still on the desktop that goes down to 30FPS.

    Performance is not bad in general to be rF2, which has always been a great devourer of resources ... Lowering something is kept at all times above 60 FPS and once advanced the race does not usually fall below 90 FPS. With a 144hz monitor and freesync the fluidity is excellent. What bothers me is that in DX11 the yield is 20% lower with the same configuration when at least it should be the same.

    I hope all these months without news bring some improvements in this section with the new build. The current requirements for playing DX11 are tremendously disproportionate.
     
    Rui Santos likes this.
  12. stonec

    stonec Registered

    Joined:
    Jun 19, 2012
    Messages:
    3,399
    Likes Received:
    1,488
    Yeah, fullscreen vs windowed mode shouldn't cause any FPS drops, I think the -30% that @SPASKIS gets in DX9 is an anomaly with DX9. The FPS should be the same because the desktop isn't actively causing a big load for modern GPUs, otherwise nobody would use windowed/borderless. The only difference with borderless and fullscreen for me is that FPS is a bit more smooth and less input lag due to game not running through Windows desktop buffer. But FPS stays the same.

    I think DX11 just fundamentally is 20% or so slower at the moment with AMD and older Nvidia generations. So if you want similarish performance to DX9, it seems only way to achieve it at the moment is to buy a GTX 900 or 1000 series GPU. Partly lack of performance is explained by Ambient Occlusion and slightly better shadows in DX11, which are new features that you cannot turn off.
     
  13. Ari Antero

    Ari Antero Registered

    Joined:
    Jul 27, 2012
    Messages:
    1,882
    Likes Received:
    829
    @stonec have you edited Windows visual effects? I haven't done that with any of my PC`S which could be explanation why I have FPS drop when borderless mode with my 680/780Ti?
     
  14. Lazza

    Lazza Registered

    Joined:
    Oct 5, 2010
    Messages:
    12,345
    Likes Received:
    6,572
    @stonec I haven't compared recently, but a year or two ago I found around a 10% difference in FPS when changing from fullscreen to windowed (borderless). It's always been a battle between being able to alt-tab freely (for motec and admin stuff, especially during/after qual) and the slightly better FPS (and general smoothness, yes, but I was hovering around 55-60fps so that extra 10% could make a big difference too, as it can then reach the screen refresh rate).

    Besides, @SPASKIS has shown an FPS increase from Dx9 -> Dx11 in windowed mode. So either the drop in dx9 is unexpected (not my personal experience, but I'll yield to the majority if that's the case), or his dx11 is underperforming in fullscreen. One thing's for sure - he can't both be getting a global drop in performance in dx11 due to his hardware, and have better fps in windowed mode with it.
     
    SPASKIS likes this.
  15. SPASKIS

    SPASKIS Registered

    Joined:
    Sep 7, 2011
    Messages:
    3,155
    Likes Received:
    1,426
    @Ari Antero (or someone else).
    Could you benchmark fullscreen vs borderless both in dx9 and dx11 to see if the results are different from mine?

    @Lazza reports 10% drop in dx9. DX11?
    I report 30% drop in dx9. 0% drop in DX11.
     
    Last edited: Oct 12, 2017
  16. Ari Antero

    Ari Antero Registered

    Joined:
    Jul 27, 2012
    Messages:
    1,882
    Likes Received:
    829
    @SPASKIS Test is made with I7-4770k, 780Ti two-way SLI. It is huge task to me remove second card becourse all my old PC`s are water cooled. I make new test with this PC and single card when new build is released. I don`t see any point to test with I7-6950X, 1080Ti three-way SLI, perhaps I make test later with single 1080Ti.

    All in-game settings are max, no AA, single screen res. 2560x1080

    DX9 Borderless:
    2017-10-12 07:41:23 - rFactor2
    Frames: 29843 - Time: 126531ms - Avg: 235.855 - Min: 178 - Max: 297

    DX9 Fullscreen:
    2017-10-12 07:46:40 - rFactor2
    Frames: 32261 - Time: 127078ms - Avg: 253.868 - Min: 174 - Max: 332

    DX11 Borderless:
    2017-10-12 08:00:25 - rFactor2
    Frames: 30295 - Time: 120703ms - Avg: 250.988 - Min: 226 - Max: 293

    DX11 Fullscreen:
    2017-10-12 08:05:15 - rFactor2
    Frames: 36507 - Time: 126438ms - Avg: 288.734 - Min: 257 - Max: 349

    :)
     
    SPASKIS likes this.
  17. SPASKIS

    SPASKIS Registered

    Joined:
    Sep 7, 2011
    Messages:
    3,155
    Likes Received:
    1,426
    Thanks for these results Ari.
    As I expected, there is difference in favour of fullscreen in both DX9 and DX11.

    This would "confirm" that my system is running in Borderless when set to fullscreen in DX11 considering I get exactly the same fps in both modes. I measure a huge 30% difference between them in DX9.
     
  18. Mulero

    Mulero Registered

    Joined:
    Jan 16, 2012
    Messages:
    351
    Likes Received:
    153
    I do not understand anything. I returned to DX9 to do the test with the Borderless mode and the first surprise has been that the FPS were a little lower than I remembered. I have changed the settings I use to perform the benckmark (max) and woooowwww ..... The performance drop since the last time I tested in DX9 is 30% less.

    Early thoughts were that it was a problem in my end, but after several tests results have always been the same. The performance drop has only been in the DX9 version of rF2, in the rest of games and the DX11 version of rF2 the performance is the same as always.

    Second surprise came when trying in Borderless mode, FPS were higher than in Fullscreen mode. And when I returned to DX11 to perform the test, the result was the same, more FPS in Borderless mode than in Fullscreen, when yesterday they were exactly the same.

    As I said .... I do not understand anything. :(

    chart.png
     
  19. SPASKIS

    SPASKIS Registered

    Joined:
    Sep 7, 2011
    Messages:
    3,155
    Likes Received:
    1,426
    When things appear illogical the best is to repeat the tests. Beware of Windows 10!!!. Check all affecting settings. Sometimes Murphy changes them!
    Try to do things exactly the same. Use preferably cockpit view and avoid TV views. I have found inconsistent fps with it.
     
  20. patchedupdemon

    patchedupdemon Registered

    Joined:
    Jan 3, 2017
    Messages:
    1,602
    Likes Received:
    1,151
    Windows ten had an update the other day for me,did yours update today
     
Thread Status:
Not open for further replies.

Share This Page