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.

R9 3900X performance

Discussion in 'ACC Hardware Discussions' started by pmgfrederico, Jul 15, 2019.

  1. Deatroy

    Deatroy Alien

    Sorry but these numbers are "useless" for ryzen 3000 CPU infos as you are running fully into the GPU Limit.
    Please follow this benchmark:
    https://www.assettocorsa.net/forum/index.php?threads/the-acc-cpu-benchmark-thread.53720/
    With Fraps you can see your minFPS.

    After your run we know how your 3600 can run 30 AI's^^
     
    LeDude83 and martcerv like this.
  2. Gopher04

    Gopher04 Racer

    There are some good reviews over at Guru3d on all the 3000 series..
     
  3. Deatroy

    Deatroy Alien

    You mean ACC Benches?
    The Guru Benches i see for ryzen 3000 had the same games like 80% of german magazines. Nothing special.
     
  4. Berniyh

    Berniyh Alien



    Interesting video, IMO.
    Especially since they test the 3900X with only 8 cores activated and the outcome is pretty much the same as the 3700X when run at the same clock speed (or actually a bit better in some cases).
    So for me it's decided then, I'll get the 3900X. The question is only when. ;)
    (Currently there are too many things that seem to be f..ed up still, so I'll wait 1-2 months until most of the early bugs were ironed out.)
     
  5. vortex19

    vortex19 Simracer

    Is that because you want an AMD processor or that you want the best CPU for gaming? If it's for gaming, the 9900k is still probably the best. In the video above, the 9900k came out best in almost every game - and remember that this was with all CPUs locked to 4GHz. The stock maximum clock speed of the 9900k (5GHz) is about 9% faster than the 3900x (4.6GHz) and most games still utilise clock speed more than cores. Add to that the fact that the price of a 9900k in the UK is currently about £75 less than the 3900x (on Amazon) and the motherboards are generally cheaper, if you get the latest, and the 9900k becomes even more attractive.
     
  6. Berniyh

    Berniyh Alien

    Thanks, I know all of that.

    This particular CPU will be mostly for gaming. For the desktop PC, I'm going to get a 3950X when it is available together with one of the low cost Radeons (due to the really great open source Linux drivers) that should pop up at around the end of the year, since on the desktop I can really make use of those 16 cores.

    Personally, I don't really care that much about 3 or 4 fps more, but I can need the power from time to time for stuff like cutting, rendering etc.
    And I do indeed want an AMD cpu, since I'm not happy about how Intel dealt with the whole Meltdown and Spectre disaster (and actually that they came into that situation at all, which basically indicates where their performance gain during the last ~5-10 years came from).

    So two new systems to build in the next ~6 months, exciting times. :D
    (and possibly some cursing, when I migrate the gaming system to Ryzen due to it being Windows 7, which is not officially supported by Zen 2)
     
  7. magzire

    magzire Alien

    Using AC benchmark with a R5 3600, i went from 26000 to 54000. Before i had 3770k. Performance has been amazing so far.
     
    Loeki, exited22, pmgfrederico and 2 others like this.
  8. Deatroy

    Deatroy Alien

    i wonder really that no one with ryzen 3000 reported back if 30 AI run with xy FPS with low graphics.
    Maybe he knows the ingame FPS Counter (activated via hud.json) aswell and posts O L B values...
     
  9. Berniyh

    Berniyh Alien

    May I recommend the 3900X to the Kunos devs? ;)

    https://www.techpowerup.com/review/amd-ryzen-9-3900x-smt-off-vs-intel-9900k/3.html
    [​IMG]

    Here the CPU pretty much shreds the competition.
     
    chksix and LeiF like this.
  10. chksix

    chksix Simracer

    Since Intel has run out of numbers (i9 9999 etc) they need to bring out the i10 series which need to beat the Ryzens to be a valid option. Not sure they can do it honestly. Just guessing though...
     
    Loeki likes this.
  11. Anthonyl

    Anthonyl Racer

    Just ran a full field of cars 30 at Spa all settings on their MAX or HIGH settings.. 2560 X 1440 and was averaging around 80fps with lows around 72 ish..
     
  12. Deatroy

    Deatroy Alien

    These numbers doesnt help us for the 3600! :rolleyes:
    You are just benching your GPU in 2K with high/epic graphics.
    i have 2x RTX 2070 at home, tested also a RTX 2080. They definitly run into GPU Limit with these settings.
    Please turn down your settings to Full HD or 720p and low graphics. We dont want to bench your 2080... we want to see how well the 3600 runs 30 AI with low O L B values!

    Example of my testing when i changed from Single Ram 8GB 2667 CL19 to 16GB 2667 CL15 for my Laptop with i7 8750H:
    1.0.6

    AI Spa Porsche 22 opponents + me:
    2019-07-08 21:16:44 - AC2-Win64-Shipping
    Frames: 10286 - Time: 180000ms - Avg: 57.144 - Min: 37 - Max: 66
    beginning O:49 L:10 B:21
    end O:48 L:10 B:178

    2019-07-21 11:45:17 - AC2-Win64-Shipping
    Frames: 16271 - Time: 180000ms - Avg: 90.394 - Min: 68 - Max: 103
    beginning O:57 L:6 B:5
    end O:58 L:6 B:5

    O is Occupancy
    L is Late Physic
    B is Blend i think, also some physics stuff.

    Late need to be low as possible.
    B can raise up to 3000 or more until it lags sometimes.
    These are the results from my personal testing.
     
    Last edited: Jul 24, 2019
    chksix likes this.
  13. Berniyh

    Berniyh Alien

    Of course these numbers help, because it tells you that 30 AI opponents is not a problem for a 3600 and the GPU is still limiting.

    If I'd do that with my CPU, I'd have a slide show with about 1-2 fps.
     
  14. LeDude83

    LeDude83 Alien

    If you feel that those numbers help, then that's fine for you. For me they are useless, just as any other numbers people have randomly read from the screen with little to no information on the test setup and parameters. Deatroy is exactly right in asking for a more systematic testing approach.

    Edit: and he's just also perfectly right in suggesting lower GFX settings for a CPU benchmark.
     
    Last edited: Jul 24, 2019
    Deatroy likes this.
  15. Berniyh

    Berniyh Alien

    Yeah, I know what you mean, but:
    I know that a lot of people request that for CPU benchmarks and that there are quite a few sites that benchmarks in 720p for exactly that reason.
    However: it's a test that is completely irrelevant
    What you're testing does not have a value.

    Sure CPU A might generate 5 more fps in the 720p+low settings test than CPU B, but if CPU B does not limit in any way in a normal scenario, then there is nothing to gain from that information.
    If a R5 3600 in normal gaming mode does not introduce a CPU limit even at maximum load on the CPU (hence maximum number of opponents), then a R5 3600 is enough for running ACC.
    That's it, nothing more needed, since under normal circumstances you will certainly be limited by the GPU.

    That's not saying that you may not do those benchmarks, it's just saying that the information you're acquiring doesn't have any real value.
     
  16. LeDude83

    LeDude83 Alien

    If you want to know the performance of a CPU (question of this thread, no?), it has all the value. How this translates into the performance in a realistic, mixed CPU and GPU stress situation (like the one you describe) is a different story. And there's a thread of mine for such tests, too.
     
  17. Berniyh

    Berniyh Alien

    There are better and more realistic tools to benchmark CPUs. ;)
     
  18. Deatroy

    Deatroy Alien

    Sure it has value.
    People want to buy Ryzen 3000 (like me) if they do same good as Intel CPUs here or even better, who knows.
    Also there is already a difference seen via the ingame FPS App.
    Its not about FPS only.
    If you get too much L and B values your CPU can run FPS-wise more cars, but the physics cant follow anymore. Resulting in lags, weird things.
    Look at my I7 8750H.
    Still having some L and B values and Dips down to 63 fps after Ram Upgrade.
    Probably 30 cars are the limit of this 3,9Ghz undervolted CPU.
    Take the 8700, which is exactly the same CPU just not throttled down to 45W.
    Will run higher FPS and probably L and B values of 0.

    Which CPU you will recommend to people? 8700 or 8750H?
    Just think about more cars Kunos will unlock some day for Spa 24H Endurance..
    That has some value, but you dont get it obviously.

    Also
     
  19. Loeki

    Loeki Gamer

    You both are right. If u only choose your cpu based on 720p results u are not doing it the right way.
    I play in VR, other people play 1440p or even 4k. U want to know also what gains u can expect at those resolutions. If u are gpu limited that doesn't mean u can't get any improvements when upgrading processor.
    So we do want to know also the real world performance.
     
    luxxx797 likes this.
  20. pmgfrederico

    pmgfrederico Gamer

    Thank you for your feedback.

    That is valuable information (having a confirmed improvement) to me as when I created this thread I was not after any sort of benchmark. Being in the process of upgrading my CPU just wanted to now if these new CPUs will represent some sort of bottleneck when compared to the 9900K for instance.

    Mainly, core occupancy observation can give me a hint on how physics will scale on these CPUs and pure clock speed on the game loop thread (let's say 4.4Ghz out of the announced 4.6Ghz for a not so heavy multi-threaded load - also trying to find out how does the game profiles regarding that) plus IPC improvement will tell me how they compare directly to the 5.0Ghz baseline of the Intel processors.

    If I disregard VR for now, I pretty much prefer super sampling to 1440p if I can't maitain 100fps in 1080p and have predictable results around 60fps than floating frame times all over the place... So yeah, let's see what the 3950X brings to the table and then a decision.
     
Similar Threads
Forum Title Date
ACC Hardware Discussions Discussing hardware, rigs and performance Nov 1, 2019
ACC Hardware Discussions ACC Graphics and Performance Discussion Oct 29, 2019
ACC Gameplay Question for Devs: Automatic Performance-based AI Difficulty Oct 29, 2019
ACC Virtual Reality Finally a huge performance boost in VR! Oct 14, 2019
ACC Troubleshooting Huge performance issues since 1.09/1.0.10 Sep 30, 2019
ACC Gameplay [1.0.10] Career mode (initial test) and AI performance. Sep 30, 2019
ACC Virtual Reality Questions concerning Frame TIME in VR (the key to smooth performance?!) Sep 22, 2019
ACC Troubleshooting Lost performance after save with 1.0.10 Sep 21, 2019
ACC Graphics - Tracks and Cars So wheres the mirror performance fix? Sep 17, 2019
ACC Virtual Reality Drastically reduced VR performance with 1.0.7? (2080ti / i9) Aug 15, 2019
ACC Virtual Reality Valve Index and 2080Ti Performance Aug 3, 2019
Troubleshooting - Workarounds Low performance with 2080ti,vr and 24 ai Aug 2, 2019
ACC Virtual Reality Better clarity and performance in VR - Rift Jul 17, 2019
ACC Troubleshooting What impact on performance does the settings have? Jul 14, 2019
Troubleshooting - Workarounds What impact on performance does the settings have? Jul 14, 2019

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