Bug causing big performance issue for a part of RF2 users + old stuttering issue still there ;)

Discussion in 'Technical & Support' started by dadaboomda, Dec 31, 2017.

  1. MarcG

    MarcG Registered

    Joined:
    Oct 4, 2010
    Messages:
    5,660
    Likes Received:
    1,073
    With every evolving games comes the need for players to be of an equal mind when it comes to upgrading, no longer are games released at v1.0 with exact System Spec Requirements then development stopped so everyone is happy. These days games evolve to a point that when they become "old" like in RF2s case then those Specs from 5 years ago need to be upped, yes optimisation is needed across the board but also an open mind from the user about the software they are using.
     
  2. Mark Fuller

    Mark Fuller Registered

    Joined:
    Dec 10, 2016
    Messages:
    62
    Likes Received:
    14
    I run an i5 4690K, with 16 GB RAM, STRIX GTX 980Ti and 250 GB SSD on W10. I run DX11 with the Oculus Rift and run at medium settings. My in-game experience is fantastic and since the update to the 'full' DX11 in December, have found my system runs very smoothly around the 90 FPS area.

    I don't think it is about spending loads of money because from what I see on the interweb, that is no guarantee. Many people on here forget that they are running rF2 on 'their' system and usually have many things running. I decided to go with a PC that only runs rF2 and very little else, and do my day to day computing on an old PC. Even then, I close down any unnecessary programs before starting rF2. I have also tested my PC non-VR and was getting 200+ FPS with the same settings.

    S397 cannot test on every known computer configuration and I would anticipate that they test on dedicated machines. We/they want rF2 to be the best and not one of the run-of-the-mill sims, so the coding is at a high level/risk(?). I have used DX11 since it was first released and have seen tremendous improvement.

    Just my thoughts and I hope that everyone perseveres with rF2 and S397.
     
    krusti, David Kolody and MarcG like this.
  3. ceecee

    ceecee Registered

    Joined:
    Jan 10, 2012
    Messages:
    302
    Likes Received:
    86
    I agree totally.
    If you are running a computer only for rF2 then you might want to look at this link.
    http://www.blackviper.com/service-configurations/black-vipers-windows-10-service-configurations/
    We have been using his advice now for longer then I can remember (think Windows 98!) but I suggest you run this only on a dedicated gaming computer as a lot of services needed in everyday use will be turned off.
     
  4. Mark Fuller

    Mark Fuller Registered

    Joined:
    Dec 10, 2016
    Messages:
    62
    Likes Received:
    14
    Thanks - will certainly take a look at that ;-)
     
  5. Lazza

    Lazza Registered

    Joined:
    Oct 5, 2010
    Messages:
    4,275
    Likes Received:
    679
    The computations you're referring to are overhead, and will impact overall performance on a weaker CPU. But once the CPU is able to do the physics and calculating where everything is (for the GPU to then draw) it will have no impact on FPS if it's faster. There is widespread evidence in rF2 and games in general that throwing more CPU at it doesn't improve FPS. If dx11 in rF2 seems more reliant on CPU it's because there is some issue that artificially raises the level of that overhead.

    rF2 is graphically below par, and the dx11 implementation has extra problems too. The below par graphics will put some people off, but those extra problems are even more serious. We can do workarounds and make the best of the situation, but it's pointless to deny the core issues.
     
  6. The Iron Wolf

    The Iron Wolf Registered

    Joined:
    Feb 20, 2016
    Messages:
    534
    Likes Received:
    364
    I measured FPS in rF2 by clocking my CPU to 3.2/3.8/4.6Ghz. I used settings that do not make it bound to GPU (no AA, for example). I saw linear growth of FPS. That's in single display 2D. Note, my CPU is very old. Maybe newer ones reach GPU ceiling.
     
  7. juanchioooo

    juanchioooo Registered

    Joined:
    May 16, 2016
    Messages:
    368
    Likes Received:
    196
    @Lazza obviously spoke of medium equipment, high range supports much more, rfactor is poorly optimized, but little by little they will improve, as they improved the dx11 that at the beginning when it was very bad, it improved, I hope it continues to improve
     
    Clemenca and The Iron Wolf like this.
  8. ECAR_Tracks

    ECAR_Tracks Registered

    Joined:
    May 1, 2016
    Messages:
    340
    Likes Received:
    256
    If it's due rFactor 2, they shouldn't bother. They won't have a smooth play anyway, People in this forum like to say "I have 362763 FPS" but they play at infamous low settings, clear sky and with 2 visible AIs, so don't trust.
     
  9. Mulero

    Mulero Registered

    Joined:
    Jan 16, 2012
    Messages:
    328
    Likes Received:
    122
    I did this test in summer, on DX11 and at least with my CPU, clock speed made almost no difference.

    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
     
  10. The Iron Wolf

    The Iron Wolf Registered

    Joined:
    Feb 20, 2016
    Messages:
    534
    Likes Received:
    364
    What GPU do you have? If GPU holds rendering back, it will look like CPU does not matter. My min FPS was 65/78/90 with frequences I mentioned. What track/series did you use? How many visible vehicles? I tested Silverstone, maxed out grid of Endruacers, all vehicles visible, end of grid measurement at start. I did that because my impression that it is number of vehicles that need drawing is where CPU is swamped (mine's 3930k).
     
    Last edited: Jan 17, 2018 at 8:02 PM
  11. Amarildo Junior

    Amarildo Junior Registered

    Joined:
    Aug 7, 2017
    Messages:
    188
    Likes Received:
    126
    It's not the physics nor the graphics, it's something else. Probably related to how the graphics engine handles mirrors since the last updated.

    Mirrors enabled on Thunderstorm, 30 cars: 20-30 FPS (on pit)
    Mirrors disabled, 30 cars: 60+ FPS (on pit)

    Since the last few days I've noticed performance is a little better, so it's seems S397 are doing some optimization.
     
  12. juanchioooo

    juanchioooo Registered

    Joined:
    May 16, 2016
    Messages:
    368
    Likes Received:
    196
    I saw a review of the i5 7600k and when the OC rose just barely earned fps in several games almost no difference, 4 fps maximum difference, I was interested in acquiring it since I was playing is almost as fast as the 7700 k i7, which showed that did not take advantage of the full potential of the i7, said by the one who did the review
     
  13. Mulero

    Mulero Registered

    Joined:
    Jan 16, 2012
    Messages:
    328
    Likes Received:
    122
    The GPU is an RX570, the CPU was previously a Phenom II 965 and I changed it to the 7600K thinking that it was the cause of the bad performance although the change did not bring practically any improvement in rF2.

    The track and the cars I do not remember, probably Nola and the Indycar. I'm going to do the tests in Silverstone with the Enduracers mod and see the results.

    Most games are usually limited by GPU rather than CPU. In games limited by CPU if you can see a great improvement when performing OC or in programs that need a good CPU.

    When you have a very powerful GPU and a somewhat old CPU already, you can also notice a great improvement when performing OC since it is the CPU that is limiting the performance, although it is not yet the case the 7600K. (Still)
     
    juanchioooo likes this.
  14. The Iron Wolf

    The Iron Wolf Registered

    Joined:
    Feb 20, 2016
    Messages:
    534
    Likes Received:
    364
    If you want to eliminate GPU out of equation, you could try setting low settings, especially no AA. That should show CPU impacting FPS. But thruth is, what we're doing here is quite futile, only S397 can fix/improve this. I tested to verify if overclocking and forcing PCIE3.0 is worth it, and in my case both are.
     
    Mulero and SPASKIS like this.
  15. Mulero

    Mulero Registered

    Joined:
    Jan 16, 2012
    Messages:
    328
    Likes Received:
    122
    Tested with the usual configuration. High / Ultra. Silverstone GT, ER Mod all cars, 103 AI (41 cars) all visible.

    FPS game paused before the start of the race in the last position. Luckily Alt + Tab worked perfectly so it was pretty quick to perform the tests.

    @ 4.5Ghz - 92Fps
    @ 3.8Ghz - 92Fps
    @ 3.6Ghz - 92Fps (Stock)
    @ 3.4Ghz - 89Fps
    @ 3.0Ghz - 86Fps
    @ 2.6Ghz - 75Fps
    @ 2.2Ghz - 66Fps
    @ 1.8Ghz - 55Fps
    @ 1.4Ghz - 44Fps

    Maybe previous tests were with a single car on the track because it is how I usually do the test to avoid variables. OC is not bringing any improvement in my system. The drop of FPS with UC is quite small until 3Ghz, from there it is more significant.

    @The Iron Wolf was watching the specifications of your CPU and despite having a few years it is still a great CPU. I've seen that it only supports Pcie 2.0 so it may be that's why when performing OC you get an improvement although I do not know if it's related. What do you mean by forcing PCIE3.0?
     
    The Iron Wolf likes this.
  16. The Iron Wolf

    The Iron Wolf Registered

    Joined:
    Feb 20, 2016
    Messages:
    534
    Likes Received:
    364
    It is hard to tell how "great" 3930k is. On a single thread, Coffee Lake is not that far ahead from my processor. However, nvidia driver is mulithreaded. So I suspect if each of the cores is 20-30% faster, who knows how much better performance will be, in particular, in VR, which is what hurts me the most. I'll see as soon as next intel 8 core comes out, hopefully with Meltdown/Spectre improved in hardware. The fact that guys with 7700k struggle in rF2 VR, however, worries me.

    The PCIE 2-3 thing with 3930k is a mess. It has to do with x79 chipset. Early revision of 3930k did not support PCIE 3.0. nvidia chose to disable PCIE 3.0 on x79 chipset claiming stability problems. Which shouldn't matter in most cases, unless SLI is involved. But there's nvidia's tool to force that on x79. To my huge surprise in rF2, forcing PCIE 3.0 gives 5%-7% percent FPS boost on single 1080ti at 800x600, no AA. Your system is likely running in 3.0, it is just the x79/393(6)0k combo that's a mess. You could verify by using tool GPU-z.
     
    Last edited: Jan 17, 2018 at 10:38 PM
    Mulero likes this.
  17. Ari Antero

    Ari Antero Registered

    Joined:
    Jul 27, 2012
    Messages:
    1,437
    Likes Received:
    500
    i5 7600k vs i7 7700k Did that one who did the review mention something about minimum FPS ;)
     
  18. juanchioooo

    juanchioooo Registered

    Joined:
    May 16, 2016
    Messages:
    368
    Likes Received:
    196
    He does not, I do not either, it must be that you did not understand what I meant, once again;):):)


    pd.- I actually confirm what he said
     
  19. Ari Antero

    Ari Antero Registered

    Joined:
    Jul 27, 2012
    Messages:
    1,437
    Likes Received:
    500
    I understand what you meant and since you wrote :"I was interested in acquiring " I just wanted to point out that minimum FPS and smoothness is as important as maximum when you consider buying new CPU. If you look this video from 20:00 you may get the clue what I wanted to point out.
     
    juanchioooo and SimonJ like this.

Share This Page