1. Do you need support for Assetto Corsa Competizione? Please use the proper forum below and ALWAYS zip and attach the WHOLE "Logs" folder in your c:\users\*youruser*\AppData\Local\AC2\Saved. The "AppData" folder is hidden by default, check "Hidden items" in your Windows view properties. If you report a crash, ALWAYS zip and attach the WHOLE "Crashes" folder in the same directory.
  2. FOR ASSETTO CORSA COMPETIZIONE: If you report issues with saved games, please always zip and attach your entire User/Documents/Assetto Corsa Competizione/Savegame folder!
  3. If your game executable is missing, please add your entire Steam directory to the exceptions in your antivirus software, then run a Steam integrity check or reinstall the game altogether.

RTX 3080 GPU + Ryzen 5000 CPU = Minimum 80-90FPS on HP Reverb G2

Discussion in 'ACC Hardware Discussions' started by sairuscz, Oct 1, 2020.

  1. sairuscz

    sairuscz Gamer

    I spent a few hours benchmarking the RTX 3080 and here are some 1440p results to better demonstrate how badly is ACC optimised to utilize all CPU cores. The main takeaway is that when I pair Ryzen 2600 and RTX 3080, both the GPU and CPU utilization remains between 40-50% both in online and offline races (with 29 opponents). Based on my bellow calculations/estimations the upcoming Ryzen 5600 (6-core) will have better single-core performance than any current Intel CPU (WOW!) and in theory should not drop bellow 80fps in ACC during race starts with 30 opponents . Can't wait for Ryzen 5000 to become available!


    Sources:
    -Ryzen 3600 is 15% faster than Ryzen 2600 in single threaded benchmarks (CPU-Z https://valid.x86.fr/bench/1 ).
    -Based on online benchmarks (youtube L2gyA7W72hM) Ryzen 3600 achieves about 30% more FPS then Ryzen 2600, but framerate can still drop to 60fps with 30 AI opponents during race start.
    -Based on Ryzen 5900x single core benchmark leaks (https://videocardz.com/newz/alleged-cpu-z-score-of-amd-ryzen-9-5900x-leaks-out), it is about 35% faster than Ryzen 2600 in single core performance.
    -Ryzen 3900x is about 7% faster in single thread than Ryzen 3600.
    -Therefore the estimation is that the upcoming Ryzen 5600 should be about 30% faster in single core performance then my Ryzen 2600.
    -We could technically see up to 60% more FPS on future Ryzen 5600 compare to Ryzen 2600.
    -Therefore where my Ryzen 2600 gets about 50fps during race start with 30 opponents, the upcoming Ryzen 5600 should do about will manage 80fps. Still bellow 90Hz of HP Reverb G2.




    BENCHMARKS:
    Zolder, 2560*1440p, 15:00, Clear Sky

    1) Here are readings taken exactly during 'Green, Green, Green' call to compare online and offline race. The CPU limitations are essentially the same regardless of the opponent's cars being driven by humans or AI.
    Note: Only bellow two results are directly comparable.

    RACE START/Green light announcement

    ONLINE 30 Players (17th position) - EPIC - 54FPS , GPU 40%, CPU 37%
    OFFLINE 29 AI (15th position) - EPIC - 58FPS, GPU 40%, CPU 45%



    2) Here are readings taken at the very same spot on the track in various situations and graphics presents:
    Note: Only bellow nine results are directly comparable as they are taken on the very same spot between turns 7 and 8 while driving and recording in nVidia ShadowPlay. There were some deviations between opponent positioning of course :).

    ONLINE RACE - 30 players
    EPIC - 78FPS GPU 56%, CPU 38%
    HIGH - 88FPS GPU 46%, CPU 36%
    MED - 94FPS GPU 40%, CPU 42%
    LOW - 100FPS GPU 35%, CPU 43%

    ONLINE QUALI - 30 players
    EPIC - 84FPS GPU 53%, CPU 42%

    OFFLINE RACE 29 AI
    EPIC - 66 FPS, GPU 47%, CPU 46%
    LOW - 103FPS , GPU 37%, CPU 44%

    OFFLINE HOTLAP
    EPIC - 162 FPS GPU 96%, CPU 36%
    LOW - 170 FPS GPU 43% , CPU 35%
     
    FX2K, swigaffu and magictap like this.

  2. Similar Threads
    Forum Title Date
    Troubleshooting - Workarounds RTX 3080ti Stutters Nov 4, 2021
    Chit Chat Room Quest 2 VR - RTX 3080 Settings Mar 9, 2021
    ACC Hardware Discussions AMD 5900X + RTX 3080, limiting factor? Monitoring data included. Jan 15, 2021
    ACC Hardware Discussions RTX 3080 or RX 6900 XT for triple 1080p Dec 23, 2020
    ACC Virtual Reality Smooth gameplay on Reverb G2 at 8K (7760x3796) with RTX 3080 Dec 22, 2020
    ACC Hardware Discussions Ryzen 5 5600X + RTX 3080 Benchmark Videos Nov 5, 2020
    ACC Hardware Discussions ACC and RTX3080 issue? Oct 12, 2020
    ACC Hardware Discussions RTX 3080 - 4K performance - CPU/engine bottlenecking Sep 27, 2020
    ACC Virtual Reality RTX 3080 Performance with Rift S (no ASW needed) Sep 22, 2020
    ACC Virtual Reality 80fps Settings Guide RTX3090/5800x Jan 2, 2021
    ACC Virtual Reality RTX 3090 SLI for VR ? Anyone tried ? Dec 12, 2020
    ACC Hardware Discussions Rtx 3070 Nov 20, 2020
    ACC Virtual Reality Oculus Quest + RTX 2070 - VR unplayable Aug 24, 2020
    ACC Virtual Reality New RTX 3000 include dlss 3.0 with support for Temporary Antialiasing May 7, 2020
    ACC Virtual Reality VR Upgrade GTX 1080 to RTX 2080 Super Apr 13, 2020

  3. rikirk

    rikirk Racer

    Reverb G2?
     
  4. DaReaLDeviL

    DaReaLDeviL Gamer

    Nice, now wait for todays AMD announcements and hopefully we get a good 5xxx CPU and maybe GPU.

    Does anybody know how good are AMD GPUs for VR compared to NVIDIA?
     
  5. sairuscz

    sairuscz Gamer

    They should only announce the CPUs today. The GPU announcement will be on the 28th of October. The actual release dates/availability are yet to be undisclosed...

    Based on current leaks, all upcoming Ryzen 5000 CPUs should exceed Intel 10900k in single core performance. If that is really the case, it will be huge. Let's see later today :).

    Regarding AMD GPUs, their currently fastest 5700XT is about 5-10% slower then 1080Ti. With my old 1080Ti, I was able to play on Rift S with 160% SS at 40FPS (ASW), so the 5700Xt should be able to do the same with slightly lower super-sampling.

    I sold rift S and 1080ti several months ago as I was really not happy with the VR experience (performance/image clarity). I am giving it a second try now. Already have RTX 3080 (under-utilized) and waiting for new Ryzen CPUs and Reverb G2. Hopefully, this huge expense will make me play ACC again :). So far it looks like hardware is finally catching up with ACC minimum HW requirements for VR :).

    PS: I don't have any hopes that mirror optimization will somehow fix the entire performance issue...
     
  6. DaReaLDeviL

    DaReaLDeviL Gamer

    I just played ACC with the valve index on a 1080ti with a i5 somewhat 5xxx and it was not realy that good. Just wait for a 3080 also and hovering over the preorder of the reverg g2 now for serveral days and I'm that undecided :(.

    Should I go 3080 with reverg g2 + ryzen 5xxx and keep my logitech g923 or should I stay at my 1080ti and go for a DD Base and Wheel :confused:
     
  7. ohjeah

    ohjeah Hardcore Simmer

    Out of curiosity, quickly tested with good old 8700k/2080 super. Same spot on the track, same conditions, settings etc

    OFFLINE RACE 29 AI
    EPIC - 80 FPS, gpu 95%
    LOW - 140 FPS, gpu 60%

    OFFLINE HOTLAP
    EPIC - 100 FPS, gpu 100%
    LOW - 240 FPS, gpu 80%

    Quite interesting actually, how much that cpu is slowing ypu down. And then again how much faster that 3080 is running on that hotlap on epic when cpu is not an issue.
     
    sairuscz likes this.
  8. sairuscz

    sairuscz Gamer

    @DaReaLDeviL: I'm pretty much in the same position where you are. But since I cannot play games on flat screen anymore due to lack of immersion, upgrade of the GPU/CPU/HMD was pretty much the only option for me. I'm quite happy with my cheap G920 as it never really distracts me from the VR immersion. Bad performance and image clarity on the other hand, made me give up on ACC, which is really something because it is the most enjoyable video game ever for me.
     
  9. rauf00

    rauf00 Simracer

    In a low resolution like 1440p (1440p is new 1080p!) Ampere series are very limited, even by the most advanced processors in heavy applications like ACC.

    RTX 3080 paired with 3900X = 38% better AVG FPS on Intel.

    [​IMG]

    And with older CPU like 4770K, 116% better AVG FPS.

    [​IMG]

    Full review, translated from Polish

    https://translate.google.pl/translate?hl=pl&sl=pl&tl=en&u=https://simrace.pl/forum/index.php/topic/3282-test-rtx-3080-w-grach-simracingowych/

    https://translate.google.pl/transla...ma-znaczenia-w-4k-czyli-rtx-3080-na-i7-4770k/
     
    DCN, anthony galmiche and Whitestar like this.
  10. sairuscz

    sairuscz Gamer

    Nice review. Pretty much confirms all our concerns. Let's see if AMD strikes back today, in 4 hours from now :). If not, I'm switching to Intel...
     
  11. Kishanp

    Kishanp Rookie

    RTX 3090 and Ryzen 3600 i was running all epic settings on a Rift S

    I think I was getting occasional dropped frames but don't know for sure as I wasn't using monitoring software.

    I'm sure the new Ryzen 5800X will help massively when the reverb G2 arrives
     
  12. DaReaLDeviL

    DaReaLDeviL Gamer

    @sairuscz not sure if a Intel System is better at the moment... they are limited to pcie 3.0 only. So you have to wait longer to get the full potential with pcie 4.0
     
  13. Whitestar

    Whitestar Hardcore Simmer

    Just as a side note, the poor resolution scaling at 1440p isn't just CPU bottlenecking but also the Ampere architecture itself, at least according to Hardware Unboxed:
    .

    Regarding the Reverb G2, keep in mind that it has to push a lot more pixels – 2x(2160x2160) (or 4320x2160) – than the other HMDs (disregarding the 8KX, which is not included in the benches below).
    But impressively, even at half resolution it looks way better (according to MRTV) than the Rift S.

    Here are some numbers from MRTV's video on benchmarking it and some of the other headsets. This is just a stress test, not gaming benchmarks.

    [​IMG]

    The number in brackets behind the G2 is the test run at half resolution. Which means that for example the Index will run 30% (1.3) better when compared against G2 at full resolution, and 30% (0.7) slower when compared against the G2 at half resolution.

    Here is a chart over common resolutions and their pixel counts. Green=monitor, blue=HMD.

    [​IMG]


    EDIT: What I'm trying to say here is that even if the G2 should run too slow in ACC at full resolution you can run it at half resolution and get a better picture that the Rift S and (probably) a significantly better framerate as well.
    So the G2 is pretty much a win-win.
     
    Last edited: Oct 8, 2020
  14. Sofa

    Sofa Gamer

    Here is a video about how the new cards perform in various setups. Spoiler: ACC performs pretty bad.

     
    tlsmikey likes this.
  15. Thomas Fun

    Thomas Fun Simracer

    Those benchmarks have flaws. Using the top from Intel and comparing it to a Ryzen 3900x, which has 1 core disabled in each CCX, while the CCX configuration and the infinity fabric is the biggest weakness of the 3000 series AMD. It was already biased towards Intel before the bench started.

    Lets watch the AMD preview, starting now :)

     
  16. DaReaLDeviL

    DaReaLDeviL Gamer

    5950X !!!!
     
  17. Kishanp

    Kishanp Rookie

    Just read in an article it's about 25% quicker than first gen ryzen for gaming.. Surely must've been a publishing error?
     
  18. Thomas Fun

    Thomas Fun Simracer

    It beats the i9 10900k
     
  19. sairuscz

    sairuscz Gamer

    It was average uplift of 26% compare to Ryzen 3000 series. So basically, AMD should slightly exceed Intel in gaming. Since their entire presentation was gaming themed, it will end up being true in real world as well.

    Good for us. Only disappointment for me is the bios update for B450 chipsets will come at the end of January 2021... I really did not want to buy new motherboard, but I will not be able to wait 3 months... Grrrr, new MOBO it is then... :)
     
  20. DaReaLDeviL

    DaReaLDeviL Gamer

    Yeah but the big thing is the * on all benchmarks... it's says that all tests where done with ryzen and amd cpu locked at 4GHz :mad: and it is only shown with 1080p - so at full throttle on both cpus and 1440p or 4k - maybe the gap is much much smaler or at worst intel at the lead... don't forget that the amd cpus are only 105w tdp where intel is at 125 tdp.
     
  21. Kishanp

    Kishanp Rookie

    So now some rough benchmark figures are out in the open, I wonder what real world performance difference this will translate to for gaming with a Reverb G2 + 3090 paired with a 5900X, rather than a 10900k whilst running on epic settings

    More so, I wonder what the fps boost will be compared to my Ryzen 3600 Lol

    As ever, I guess we'll have to wait until these new CPU's fall into reviewer hands
     
    AndyK70 likes this.

Share This Page

  1. This site uses cookies to help personalise content, tailor your experience and to keep you logged in if you register.
    By continuing to use this site, you are consenting to our use of cookies.
    Dismiss Notice