DirectX 9 CPU Benchmark Thread

Discussion in 'General Discussion' started by Thomas Jansen, Jul 22, 2019.

  1. Sebastien Brunier

    Sebastien Brunier Active Member Beta tester

    Joined:
    Dec 17, 2017
    Ratings:
    +38 / 0 / -0
    From this CPU-Z capture, your RAM isn't running at 3666, but 2666 (1330 Mhz x2) with pretty loose timings 19-19-19-43 so I would check in your motherboard BIOS if you have any memory adjustment possible (XMP profile, memory timing adjustments)
     
  2. S. FREDRIKSSON

    S. FREDRIKSSON Member

    Joined:
    Oct 14, 2020
    Ratings:
    +11 / 0 / -0
    Hmm, I must have written it wrong (or forgot to edit the copy/paste) but probably its just 2666. I will edit my post.
    Thanks!
     
  3. rad

    rad Well-Known Member

    Joined:
    Jan 29, 2015
    Ratings:
    +441 / 0 / -0
  4. Thomas Jansen

    Thomas Jansen Sector3 Developer Beta tester

    Joined:
    Apr 5, 2018
    Ratings:
    +434 / 0 / -0
    I wasn't planning on making this as big as this thread anyway, just would like to see a couple of cpus, maybe an i9 and a zen 2 cpu :D
     
  5. mmmmbeer

    mmmmbeer Member

    Joined:
    Jun 21, 2015
    Ratings:
    +6 / 0 / -0

    Here's my 3600, still actually OC'd to 4.5 all core but seemingly only getting to 4.2 in the test. Hefty discrepancy between the two runs.

    EDIT:- this was wrong. my BIOS reset itself at some point so this run was at stock settings and no XMP enabled.
     

    Attached Files:

    • Like Like x 1
    Last edited: Nov 14, 2020
  6. mmmmbeer

    mmmmbeer Member

    Joined:
    Jun 21, 2015
    Ratings:
    +6 / 0 / -0
    Here's my results for my actual 4.5ghz OC. Not much difference to be honest!
     

    Attached Files:

    • Like Like x 1
  7. Thomas Jansen

    Thomas Jansen Sector3 Developer Beta tester

    Joined:
    Apr 5, 2018
    Ratings:
    +434 / 0 / -0
    wow and you're 100% sure you used the same settings? 720p on the low graphics preset? That would be a massive difference to my 5800X. Do you maybe have a webhud running at the same time? I believe the .exe that reads the shared memory eats up a lot of CPU.
     
    Last edited: Nov 14, 2020
  8. mmmmbeer

    mmmmbeer Member

    Joined:
    Jun 21, 2015
    Ratings:
    +6 / 0 / -0
    I use otterhud and no its not enabled I don't think. Remind me how and I can check for sure, I haven't played for a while! I checked my game launch settings in steam and its default blank.

    Game settings were identical.
     
    • Like Like x 1
  9. Thomas Jansen

    Thomas Jansen Sector3 Developer Beta tester

    Joined:
    Apr 5, 2018
    Ratings:
    +434 / 0 / -0
    You would have something in the launch options in steam indeed, crazy results then :D curious how others will compare
     
  10. mmmmbeer

    mmmmbeer Member

    Joined:
    Jun 21, 2015
    Ratings:
    +6 / 0 / -0
    Yep they are, almost double the average frames. I've got a Reverb G2 arriving in the first batch of pre orders soon. Was hoping to have my 5900x by then to drive it, sadly won't!
     
  11. rad

    rad Well-Known Member

    Joined:
    Jan 29, 2015
    Ratings:
    +441 / 0 / -0
    Ryzen 1600X @ 3,9GHz (with boosting disabled, I think I was able to get it fairly stable at 4.1GHz, SMT is enabled), Gigabyte RTX 2060 Gaming OC Pro (Stock clocks, don't matter anyway in RR, since CPU is bottleneck), 2x8GB RAM 3000MHz 16-17-17-35

    For first run I forgot to set FOV to 1.0, fixed it for second one and third one I just did out of curiosity on settings I normally use (minus G-Sync).

    BTW, you can get lower with resolution - 1024x768, just need to change aspect ratio to 4:3

    upload_2020-11-15_0-14-11.png
     
    • Like Like x 1
    Last edited: Nov 15, 2020
  12. Thomas Jansen

    Thomas Jansen Sector3 Developer Beta tester

    Joined:
    Apr 5, 2018
    Ratings:
    +434 / 0 / -0
    Ah didn't realise that about the resolution, luckily it's only a small difference to 720p, I would assume most people have a good enough GPU to not bottleneck at 720p :D
     
  13. Thomas Jansen

    Thomas Jansen Sector3 Developer Beta tester

    Joined:
    Apr 5, 2018
    Ratings:
    +434 / 0 / -0
    It seems that whatever they did with zen 3 is having an absolutely massive effect in simracing, in ACC we're seeing a similar gap of 50-100% between a 3700x and 5800x :D
     
    • Like Like x 2
  14. Balrog

    Balrog Well-Known Member

    Joined:
    Apr 10, 2015
    Ratings:
    +390 / 0 / -0
    F*** me, I wanted to skip this generation and I have other plans to burn money, but we'll see next year.:D
     
    • Like Like x 1
  15. Thomas Jansen

    Thomas Jansen Sector3 Developer Beta tester

    Joined:
    Apr 5, 2018
    Ratings:
    +434 / 0 / -0
    I do have to add I didn't gain too much performance in VR yet, because i'm now fully gpu limited by my 1080ti, but if you pair these cpus with a 3080 or 6800XT they will be insane :D
     
  16. Peter Stefani

    Peter Stefani New Member

    Joined:
    Sep 9, 2019
    Ratings:
    +2 / 0 / -0
    [​IMG]
    [​IMG]
     
    • Like Like x 1
  17. Maskerader

    Maskerader Well-Known Member

    Joined:
    Oct 6, 2019
    Ratings:
    +48 / 0 / -0
    Is ACC also CPU-limited like R3E?
     
  18. Maskerader

    Maskerader Well-Known Member

    Joined:
    Oct 6, 2019
    Ratings:
    +48 / 0 / -0
    i5-4670k stock (should be right around Bauer's score (172) because we have basically the same hardware, but somehow it's higher)
    Untitled-1.jpg

    i5-4670k at 4.2GHz and memory at a lower latency
    Untitled-2.jpg

    -------------
    Then I wanted to upload in-game results and things got... interesting. I'm experiencing FPS variations that I cannon explain.

    1) When I load the replay and do several FPS runs just by rewinding the replay to the start, I get very consistent results, less than 1% difference. But if I quit the replay mode and load the replay again, results will be different this time. Can be higher or lower, I don't know what it depends on. Again, doing several runs just by rewinding gives me basically the same results until I quit and reload the replay. The difference can go up to 15% while all the setting are still the same.

    Anyway, I picked a somewhat average result for the graph.

    2) If I go to nVidia control panel and crank up 3D settings for R3E to max quality, the FPS increases. I'd expect otherwise or at least no change, but definitely not an increase. So far I see something about 5% increase (yes, this is on top of those 15% from my previous point). I decided not include it here.

    What I also noticed is that outside cameras not only give a higher average FPS, but also make it significantly more consistent: the difference between average FPS and 0,1 percentile is not as big as in the cockpit view. What could be the reason?

    Untitled-3.jpg
     
    Last edited: Nov 16, 2020
  19. Thomas Jansen

    Thomas Jansen Sector3 Developer Beta tester

    Joined:
    Apr 5, 2018
    Ratings:
    +434 / 0 / -0
    ACC is definitely less CPU limited than RR, but still way more CPU dependent than any other modern game. After making our testing a bit more consistent we were looking at about 30-50% difference between my 5800X and a 3700X in ACC:
    upload_2020-11-16_9-10-37.png
    Also a good example of why RAM matters :D
     
    • Like Like x 1
    • Love it! Love it! x 1
  20. Thomas Jansen

    Thomas Jansen Sector3 Developer Beta tester

    Joined:
    Apr 5, 2018
    Ratings:
    +434 / 0 / -0
    @Sebastien Brunier also did the makeshift RR benchmark with his 3700X with pretty much the same RAM overclock as i have (he has dual rank vs my single rank, so slight advantage for his 3700X). Seems to confirm the results from @mmmmbeer , insane gains :D
    upload_2020-11-16_18-14-6.png
     
    • Like Like x 1