GTX 770 - performance lower than expected. Need help from 680/770/780/Titan owners

Discussion in 'Technical Archives' started by Marek Lesniak, Jul 6, 2013.

Page 3 of 4
  1. Marek Lesniak

    Marek Lesniak Car Team Staff Member

    Joined:
    Oct 6, 2010
    Messages:
    1,585
    Likes Received:
    101
    Thanks a lot! Looks like GTX780 gives a nice boost :)
    Test 1 is with AI so CPU could be more important here.

    EDIT to Edit:
    Looking at those scores for 780, Test 2 and 3 show almost 100% boost (compared to GTX 580) so 780 seem to work great in rF 2.
     
    Last edited by a moderator: Jul 28, 2013
    Marek Lesniak, Jul 27, 2013
    #41
  2. Natureboy

    Natureboy Registered

    Joined:
    Jan 13, 2013
    Messages:
    117
    Likes Received:
    0
    Here are some tests for you. As stated earlier, I have found the 770 to give 50-100% increase in FPS. I have performed your tests with stock CPU and OC CPU settings on 1080 and 1440 res.

    Sorry but I do not want to reinstall my GTX660, but for reference at 1440p with f3 at brianza 3xAA and everything on max I had 45-55FPS and very smoth. With same settings on 770 gives 75 to 85FPS. I did however not see this improvement right away because the 770 somehow conflicted with my g27 wheel. I had to uninstall the wheel and card drivers, then install the card then the wheel - worked brilliant after that.

    Here are the figures:

    Gigabyte Z77X-UD3H
    i5 3570k@4.4GHz
    16GB hyper-x red @ 1600MHz
    EVGA GTX770SC with ACX cooler at 1189MHz

    1920x1080 Everything on and max with 6xAA

    Test 1
    Avg: 61.266 - Min: 48 - Max: 77
    Test 2
    Avg: 92.071 - Min: 71 - Max: 116
    Test 3
    Avg: 88.775 - Min: 73 - Max: 106

    2560x1440 Everything on and max with 3xAA

    Test 1
    Avg: 54.642 - Min: 42 - Max: 67
    Test 2
    Avg: 80.468 - Min: 64 - Max: 99
    Test 3
    Avg: 78.339 - Min: 67 - Max: 91



    i5 3570k@3.4GHz, 3.8GHz turbo
    16GB hyper-x red @ 1600MHz
    EVGA GTX770SC with ACX cooler at 1189MHz

    1920x1080 Everything on and max with 6xAA

    Test 1
    Avg: 56.498 - Min: 45 - Max: 75
    Test 2
    Avg: 94.318 - Min: 73 - Max: 119
    Test 3
    Avg: 90.340 - Min: 74 - Max: 113

    2560x1440 Everything on and max with 3xAA

    Test 1
    Avg: 50.779 - Min: 40 - Max: 62
    Test 2
    Avg: 80.653 - Min: 64 - Max: 99
    Test 3
    Avg: 76.354 - Min: 62 - Max: 90
     
    Natureboy, Jul 28, 2013
    #42
  3. Marek Lesniak

    Marek Lesniak Car Team Staff Member

    Joined:
    Oct 6, 2010
    Messages:
    1,585
    Likes Received:
    101
    Thanks Natureboy a lot! I edited my previous post (thoughts abou 780 perfomance) and now as I look at them, there might be something not OK between my 770 and rF 2 (while in other games it works perfectly well!!).

    Thanks guys! Will try to investigate the problem and let you know.
     
    Marek Lesniak, Jul 28, 2013
    #43
  4. Natureboy

    Natureboy Registered

    Joined:
    Jan 13, 2013
    Messages:
    117
    Likes Received:
    0
    if you notice anything funny with FFB, it could be the same problem I had at first with the wheel. Everything appeared to work fine, then FFB would end up with lots of lag.
     
    Natureboy, Jul 28, 2013
    #44
  5. Marek Lesniak

    Marek Lesniak Car Team Staff Member

    Joined:
    Oct 6, 2010
    Messages:
    1,585
    Likes Received:
    101
    Tomorrow will have my wheel connected, so we will see (but already on a GTX 680 4GB - had a very good deal for swapping out my GTX 770 2GB for it). For the record, those scores are with Gigabyte GTX 680 4GB with factory boost at 1150MHz (Windforce 3 cooler, the card takes 3 slots of space :D ):
    Min. 52, Max. 82, Avg. 66.365

    Hopefully, Tuttle will eventually do those tests, for comparison :)

    Also, just to let you know, I managed to fill up slightly over 2,5GB of VRAM on that GTX 680. Settings as for the tests, at Silverstone, 28 cars (all GTs + Marussia and FR3.5 from ISI).
     
    Last edited by a moderator: Aug 8, 2013
    Marek Lesniak, Aug 7, 2013
    #45
  6. Tuttle

    Tuttle Technical Art Director - Env Lead

    Joined:
    Feb 14, 2012
    Messages:
    2,480
    Likes Received:
    775
    Ok here we go.

    My system status (W7 Ultimate 64Bit) - No OC, the entire system is running on stock values:

    View attachment 9206

    Fresh install of Build 240 - Video settings:

    View attachment 9207

    Ingame settings:

    View attachment 9208

    Results (car mirrors ON):

    Test 1. Megane at Mills Outer Loop C - with 24AI: Avg: 66.075 - Min: 41 - Max: 110


    Test 2. Megane at Mills Outer Loop C - no AI: Avg: 111.824 - Min: 84 - Max: 148


    Test 3. Modern F2 at Silverstone - no AI: Avg: 108.339 - Min: 89 - Max: 126



    Let me know if you need further tests. :)
     
    Tuttle, Aug 9, 2013
    #46
  7. deak1944

    deak1944 Guest

    What happens on your system when you change "Multi display performance mode" to "Single display performance mode"?
     
    deak1944, Aug 9, 2013
    #47
  8. Ingy

    Ingy Registered

    Joined:
    Oct 16, 2011
    Messages:
    13
    Likes Received:
    0
    Strange prob u have mate, just switched from Gainward 570GTX GLH II which was OC-ed above 580GTX by 10 %, bought a Gigabyte GTX 770 OC 4GB and I have almost twice the FPS(3dmark Firestrike 570 GLH score 4858, with 770 score was 7899) also noticed the lower rate of micro lag( Was using the latest Beta driver 326.41 both cases) . Last time i had the same prob was the mobo's fault. If u have the chance to check ur VGA in another board would be the best solution.
     
    Ingy, Aug 10, 2013
    #48
  9. Marek Lesniak

    Marek Lesniak Car Team Staff Member

    Joined:
    Oct 6, 2010
    Messages:
    1,585
    Likes Received:
    101
    3DMark, Unigine/Valey Benchmark, Crysis 3 etc. - scores were as expected. The problem seem to be related only to rFactor 2. Will try again on a fresh OS.
     
    Marek Lesniak, Aug 10, 2013
    #49
  10. Bart S

    Bart S Member

    Joined:
    Oct 5, 2010
    Messages:
    851
    Likes Received:
    104
    What a nice guy, must be very busy with gfx stuff, works for ISI and still got time to help you with time consuming tests.

    I will see if I can help u out too mate
     
    Bart S, Aug 12, 2013
    #50
  11. Bart S

    Bart S Member

    Joined:
    Oct 5, 2010
    Messages:
    851
    Likes Received:
    104
    Test 1. Megane at Mills Outer Loop C - with 24AI: Avg: 63.130 - Min: 46 - Max: 91

    Test 2. Megane at Mills Outer Loop C - no AI: Avg: 127.756 - Min: 103 - Max: 157

    Test 3. Modern F2 at Silverstone - no AI: Avg: 109.907 - Min: 82 - Max: 143

    Ok dude I think your test is floored my specs are similar to tuttles but somethings not right.
    Check my specs 4770k @ 4.5ghz GTX780 @ 1020mhz
    I think u need to have more accuracy in your test description.
    I basically should of smoked his stats but i only beat him just slightly.
    This show the benchmark isnt accurate enough, we need a better method for testing.

    And mate from a pol to a pol I hope these arnt the settings you use to play rfactor2 they seriously are cartoon looking, you could get much better than this using nV inspector send us a mail if u intrested in some pointers.
     
    Last edited by a moderator: Aug 12, 2013
    Bart S, Aug 12, 2013
    #51
  12. Bart S

    Bart S Member

    Joined:
    Oct 5, 2010
    Messages:
    851
    Likes Received:
    104
    just comparing a sample with much better gfx on test3

    Test 3. Modern F2 at Silverstone - no AI: Avg: 109.907 - Min: 82 - Max: 143 - your set
    AA level 6
    45fov

    vs

    nv inspector
    Test 3. Modern F2 at Silverstone - no AI: Avg: 123.017 - Min: 100 - Max: 144 - my set
    8xCSAA 4c+4cv
    2xSGSS
    26fov
     
    Bart S, Aug 12, 2013
    #52
  13. Marek Lesniak

    Marek Lesniak Car Team Staff Member

    Joined:
    Oct 6, 2010
    Messages:
    1,585
    Likes Received:
    101
    Thanks Bart. I am fully aware, my test settings are demanding... that's on purpose. Still though, I didn't wanted to use tweaks from outside of driver settings. Wanted it to be clean and straight forward and easy to replicate.

    I tried to describe my testing methods as precise as possible. I can do several passes and all give very similar results.
    I don't use nV Inspector for those tests for a reason (everything is described in my first post).

    Maybe your scores are lower because nV Inspector still fiddle with your settings (I assume you reset it to defaults and drivers are too, apart from those few options from the first post).
     
    Marek Lesniak, Aug 12, 2013
    #53
  14. Bart S

    Bart S Member

    Joined:
    Oct 5, 2010
    Messages:
    851
    Likes Received:
    104
    LesiU don't worry I fully reset nv inspector, there may of been something not set quite right tho between mine and tuttles results but I follow your first post accuratley to the description so maybe you missed something important. Perhaps Fraps wasnt started / finished at the exact moment.

    An benchmarkin app from ISI would be really useful, esspecially set to a default for everyone.


    EDIT: I just thought what the difference maybe, I have 3D drivers installed, altho not in use they still steal some fps, that may be the difference between mine an tuttles results
     
    Last edited by a moderator: Aug 12, 2013
    Bart S, Aug 12, 2013
    #54
  15. AlexJTownsend

    AlexJTownsend Registered

    Joined:
    Aug 5, 2013
    Messages:
    22
    Likes Received:
    3
    Must say that this is all very interesting for me. I'm very new to Sim racing and have only just built my PC capable of running something like this properly.

    I will carry out the tests this week, not only to supply extra data but also for my own learning and peace of mind.

    My system is similar to Tuttle and Bart, (apart from the CPU), and I have the Palit 780 GTX SuperJetStream to test so this may give different results to stock GTX780.

    I must say that so far, the game runs solid and fast with most things maxed, (if not everything), with F2 at Malaysia. Looks amazing too!
     
    AlexJTownsend, Aug 13, 2013
    #55
  16. Tuttle

    Tuttle Technical Art Director - Env Lead

    Joined:
    Feb 14, 2012
    Messages:
    2,480
    Likes Received:
    775
    Just to be 100% clear, the system where I did the test is brand new and clean like fresh snow. No antivirus running nor any other background demanding task. I did nothing in terms of system tweaks or OC, so should be very easy to get the "same" result for a similar system running a clean OS and a fresh rF2 installation.
     
    Tuttle, Aug 13, 2013
    #56
  17. AlexJTownsend

    AlexJTownsend Registered

    Joined:
    Aug 5, 2013
    Messages:
    22
    Likes Received:
    3
    Need to figure Fraps out...

    Test 1 - Don't think this is correct...

    2013-08-15 10:13:50 - rFactor2
    Frames: 117541 - Time: 1014584ms - Avg: 115.851 - Min: 0 - Max: 2631
     
    AlexJTownsend, Aug 15, 2013
    #57
  18. Marek Lesniak

    Marek Lesniak Car Team Staff Member

    Joined:
    Oct 6, 2010
    Messages:
    1,585
    Likes Received:
    101
    Interesting indeed :)
     
    Marek Lesniak, Aug 15, 2013
    #58
  19. AlexJTownsend

    AlexJTownsend Registered

    Joined:
    Aug 5, 2013
    Messages:
    22
    Likes Received:
    3
    Figured it out... (read the instructions...)

    Test 1
    2013-08-15 11:15:17 - rFactor2
    Frames: 14653 - Time: 206810ms - Avg: 70.852 - Min: 50 - Max: 110

    Test 2
    2013-08-15 11:20:49 - rFactor2
    Frames: 24474 - Time: 203597ms - Avg: 120.208 - Min: 96 - Max: 150

    Test 3

    2013-08-15 11:42:00 - rFactor2
    Frames: 28027 - Time: 275216ms - Avg: 101.836 - Min: 76 - Max: 128

    Retested... Didn't have steering wheel "off"

    Test 3
    2013-08-15 11:51:31 - rFactor2
    Frames: 29513 - Time: 275077ms - Avg: 107.290 - Min: 80 - Max: 136


    I'm not sure about test 3 as I had race settings set to the same as the previous tests so the driver didn't start in the pits but on the track to do the formation...
    Should I have changed the race settings?

    I'm also wondering if I need vsync on or not?
    I seem to be getting more tearing with it off as I'm currently using my Samsung TV through HDMI, maybe I need vsync for TV use?
     
    Last edited by a moderator: Aug 15, 2013
    AlexJTownsend, Aug 15, 2013
    #59
  20. Tuttle

    Tuttle Technical Art Director - Env Lead

    Joined:
    Feb 14, 2012
    Messages:
    2,480
    Likes Received:
    775
    I see you're getting similar values compared to my system, so this is a good confirmation. :)

    For the Test 3 you've just to choose the Practice session instead Race in the RACE SETTINGS area. This way you can start from garage and drive an entire lap grid to grid.

    Keep the VSynch off for all those tests or you're going to cap your max framerate with your monitor refresh rate.
     
    Tuttle, Aug 15, 2013
    #60
Page 3 of 4

Share This Page