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.
    Dismiss Notice

The ACC CPU Benchmark Thread

Discussion in 'ACC General Discussions' started by Deatroy, Dec 7, 2018.

  1. Ole77

    Ole77 Rookie

    2018-12-31 14:44:12 - AC2-Win64-Shipping
    Frames: 7293 - Time: 60000ms - Avg: 121.550 - Min: 101 - Max: 152

    i5 8400
    16GB RAM
     
  2. greedo86

    greedo86 Rookie

    2018-12-31 22:19:46 - AC2-Win64-Shipping
    Frames: 4145 - Time: 60000ms - Avg: 69.083 - Min: 59 - Max: 80

    i5 2500K (stock)
    16GB DDR3 Memory
     
    Deatroy likes this.
  3. Deatroy

    Deatroy Alien

    Can you recheck if you have done this test in 720p?
    You beat easily the i7 8700k here.

    But on the other side, R5 2600 is also crazy fast.

    Table updated.
     
    After_Midnight likes this.
  4. baakstaff

    baakstaff Gamer

    Ran the test in both 720p and 1080p on version 0.4.2 (didnt restart the game when I switched to 1080p).

    720p:
    2019-01-01 11:03:20 - AC2-Win64-Shipping
    Frames: 5667 - Time: 60000ms - Avg: 94.450 - Min: 80 - Max: 118

    1080p:
    2019-01-01 11:04:58 - AC2-Win64-Shipping
    Frames: 5707 - Time: 60000ms - Avg: 95.117 - Min: 78 - Max: 124

    It is possible to abuse the benchmark though, dropping back a little and turning in late to the corners so the other cars were out of view improved my 1080p results like this:

    2019-01-01 11:12:40 - AC2-Win64-Shipping
    Frames: 6267 - Time: 60000ms - Avg: 104.450 - Min: 83 - Max: 130

    So to take the player out of the equation i tried doing the formation lap, then going into the replay and riding onboard with the AI car that was directly in front of me on the grid. I ended up with this:

    2019-01-01 11:15:03 - AC2-Win64-Shipping
    Frames: 5713 - Time: 60000ms - Avg: 95.217 - Min: 78 - Max: 115

    Considering how close it is to my live benchmark runs, it's safe to say the replay stresses your system in the same way. It could be worth seeing if using the replay onboard an AI car can take away some of the uncertainty from our results due to differences in how each of us may drive the formation lap
     
  5. cobrettee

    cobrettee Gamer

    2019-01-01 13:02:10 - AC2-Win64-Shipping
    Frames: 6662 - Time: 60000ms - Avg: 111.033 - Min: 93 - Max: 135

    R7 2700x/16GB
     
    Deatroy likes this.
  6. Ole77

    Ole77 Rookie

    Rechecked: tested in 720p and results are reproducible.
     
    Deatroy likes this.
  7. Deatroy

    Deatroy Alien

    @Ole77 thanks mate.
    Impressive performance.

    @luxxx797 can you recheck, your results are too low then.
    You should perform better than i5 8400 and the older i7 5820k.

    @cobrettee
    Nice, RyZen 2000 Series vs 1000 Series performs way better in ACC than usual in other games.
     
    Ole77, cobrettee and After_Midnight like this.
  8. luxxx797

    luxxx797 Racer

    even the 7700K(in 1080) got better results but i'm in line with Atak_kat (8700K too)....so the i5 8400 is 21% more performing, guessing why....
    will test again next days

    Edit: redone the benchmark just now, forgot to change FOV... was set to 35 instead of 45...anyway:

    2019-01-01 21:31:08 - AC2-Win64-Shipping
    Frames: 6707 - Time: 60000ms - Avg: 111.783 - Min: 95 - Max: 138
     
    Last edited: Jan 2, 2019
  9. Deatroy

    Deatroy Alien

    Fov don't have to be 45 here.
    Most of us play with default Fov.

    Seems more reasonable now.
    But still slower than expected.
    Check again with default Fov.

    Gesendet von meinem EML-L29 mit Tapatalk
     
  10. Ole77

    Ole77 Rookie

    I expected that all Coffeelake i5/i7 CPUs would perform quite similar in this benchmark. As my results were way too good and rather implausible i started to (try to) investigate the matter.

    As first step i checked and updated all drivers and BIOS. Mainboard BIOS got updated from 0416 2017/10/25 to 1410 2018/09/29. Intel Chipset Driver had to be updated. Nvidia driver already was 417.35 and Windows already was version 1809. I checked all BIOS settings and set Nvidia Control Panel to QUALITY settings.

    Since these updates i consistently get these results:

    2019-01-02 11:57:26 - AC2-Win64-Shipping
    Frames: 6971 - Time: 60000ms - Avg: 116.183 - Min: 102 - Max: 137

    Results now seem much more plausible. i5 8400 still shows fantastic performance imo!
     
    Deatroy likes this.
  11. Deatroy

    Deatroy Alien

    Well when we do benchmarking you will do the most performing setting
    My results are always optimized for best runs. So quality is on performance and everything is updated to newest driver/bios.
    But yes seems on line now.

    And 0.4.x brought optimisation on CPU Side.
    Intel and newer Ryzen seems to like it on CPU Limit vs AI.
    My 1600 got way slower.
    My laptop i7 is also way faster now.
    But Release is still far away.

    Gesendet von meinem EML-L29 mit Tapatalk
     
    Ole77 likes this.
  12. baakstaff

    baakstaff Gamer

    I disabled every other thread by using task managers core affinity (afaik basically disabling SMT), ended up with a small but consistent gain.

    2019-01-02 11:20:30 - AC2-Win64-Shipping
    Frames: 6034 - Time: 60000ms - Avg: 100.567 - Min: 84 - Max: 123
     
    Deatroy likes this.
  13. Deatroy

    Deatroy Alien

    Yea Full Formation Lap behind AI Cars is wanted.
    Surely you are right, you can gain FPS due not driving slowly behind them.
    Well i hope and think everyone will do it right.

    Anyways my CPU is way way slower than yours.
    Dont know why it is dropping so hard in 720p.
    You have like 20 FPS more.
    Maybe i should test it aswell, turning SMT off.
     
  14. NeRo2

    NeRo2 Gamer

    1280x720:
    2019-01-02 17:41:36 - AC2-Win64-Shipping
    Frames: 4067 - Time: 60000ms - Avg: 67.783 - Min: 56 - Max: 83

    1920x1080:
    2019-01-02 17:44:09 - AC2-Win64-Shipping
    Frames: 4078 - Time: 60000ms - Avg: 67.967 - Min: 57 - Max: 83

    FX8350 @ 4,5Ghz
    GTX 980ti
    16GB Ram DDR3
     
    Deatroy likes this.
  15. LeDude83

    LeDude83 Alien

    I'm starting to wonder how much RAM (size, speed and usage) plays a role in this equation.

    @f_deutsch sent me a PM. He made a custom benchmark, one for GPU and one for the entire system. He ran both with stock and manually reduced RAM speed. There was a small difference in the GPU heavy test and a significant one in the system (read: CPU heavy) test. He had changed RAM speed by "only" 300 MHz and lost about 10 % FPS IIRW. I'll go get his exact numbers and report back...
     
  16. LeDude83

    LeDude83 Alien

    Here's what @f_deutsch wrote me in PM. I guess it's fine to share with everyone :)

    Hi,

    Here are the test I did. Two scenarios:
    • Quick Race 20 cars @ Nurburgring, back of the grid all the time, close to the cars in front
    • HotLap @ Nurburgring
    Both at sunny weather 14:00 hours. No frame limit and no vsync.

    Results:

    Memory at 1336Mhz
    Quick Race
    Frames: 3488 - Time: 60000ms - Avg: 58.133 - Min: 47 - Max: 73
    HotLap
    Frames: 6877 - Time: 60000ms - Avg: 114.617 - Min: 103 - Max: 129

    Memory at 1600Mhz
    Quick Race
    Frames: 3806 - Time: 60000ms - Avg: 63.433 - Min: 48 - Max: 80
    HotLap
    Frames: 6946 - Time: 60000ms - Avg: 115.767 - Min: 102 - Max: 129

    Only difference on memory is the speed. Memory timings remained the same.

    IMO, the more complex the scenario (more 3d models on the quick race to render), the more dependent of memory speed. We need to keep in mind memory speed delta on the test is not that much (300Mhz). Comparing these DDR3 speeds to newer DDR4 memory technologies, which ranges from 2200Mhz up to 3200Mhz and beyond, could result in a bigger difference. The higher the CPU speed is, the more memory bandwidth limited can be (my processor 2500K is clocked @ 4.4Ghz). That is probably a reason why latest CPU generations are faster on certain applications as the memory bandwidth is wider.

    Hope this could help in understand how this resource affects the performance of ACC and other applications.
     
    Deatroy likes this.
  17. Atak_Kat

    Atak_Kat Racer

    I discovered today that my resolution was changing when I was going from windowed to full screen (I was going back and forth from a chrome window so I could see the forum post and related game settings, and apply them in ACC). Whenever I go from windowed to full screen, it somehow changes the resolution, even if the settings screen still shows something different. And I suspect that somehow my first tests might not have had the correct resolution. For sure this happened on the 720p test. Also, since prior runs, I'm on new version of ACC. Interesting how different the results are now, compared to 0.3.5. For sure my original 720p result was messed up. But I am not sure about the 1080p one.... I think that one was originally ok....

    Anyway, updated results here:
    i7 8700k, 1080ti, 16gbram, no overclocks

    1080p
    2019-01-06 11:54:19 - AC2-Win64-Shipping
    Frames: 6812 - Time: 60000ms - Avg: 113.533 - Min: 96 - Max: 141

    720p
    2019-01-06 11:48:41 - AC2-Win64-Shipping
    Frames: 7033 - Time: 60000ms - Avg: 117.217 - Min: 99 - Max: 144
     
    Deatroy likes this.
  18. baakstaff

    baakstaff Gamer

    I tried using ryzen master to drop my core frequency to 3825 mhz to see how much core speed impacted my performance; kinda surprised really, it didnt do much at all. I've also improved my memory performance, so we can only compare the first of the 3 tests here to my previous result to see how memory improves performance.

    Cores at 3875 mhz, 3400 mhz ram

    2019-01-06 11:32:23 - AC2-Win64-Shipping
    Frames: 5919 - Time: 60000ms - Avg: 98.650 - Min: 85 - Max: 119


    dropping core frequency to 3825 mhz, 3400 mhz ram

    2019-01-06 11:46:46 - AC2-Win64-Shipping
    Frames: 5833 - Time: 60000ms - Avg: 97.217 - Min: 83 - Max: 122

    Compared to my previous results (95 fps with 3875 core and 3266 memory), at least on ryzen theres more to be gained from memory tuning than raw core frequency. While intel systems arent affected to the same degree (cause ryzen is ryzen), itd be worth it to see if they scale better with memory rather than core frequency in a similar way.

    How much have you put into tuning subtimings on your kit?
     
    Deatroy likes this.
  19. Deatroy

    Deatroy Alien

    Subtimings arent tuned yet. I wanted to start but couldnt find a free weekend for it yet.
    Also having Hynix Ram doesnt make it easier for me to tune something as first ryzen gen isnt that open to all rams hehe.
    First i have to figure why my performance dropped so hard. Your CPU Performance is my goal. So around 95 FPS Avg would be lovely.

    Overclocking core speed will be interesting with Ryzen 3000. You are going the right way with memory.

    Actually i need to restart my topic again due Ram Speed.
    LeDude and f_deutsch posted their observations aswell.
    Ram is important, atleast in this Game vs AI.
     
  20. LeDude83

    LeDude83 Alien

    f_deutsch was using a 2500k and GDDR3 for his RAM test. I can also run the test later this week. I think have decent RAM-G.Skill Ripjaws 3600 CL 19 GDDR4. So there's quite some margin for reducing frequency.

    One thing about RAM: I recently checked usage during my system benchmark (15 AI) and it was at ~4.5 GB. I have 8 GB RAM so I assume that's enough, right?
     
Similar Threads
Forum Title Date
ACC General Discussions ACC - 0.5.2 - Performance Review by GameGPU (29 GPUS & 27 CPUS) Jan 22, 2019
ACC Troubleshooting gamebreaking stuttering after massive ACC crash CPU Oct 14, 2018
ACC Troubleshooting I have CPU Thermal Throttling and BSOD in ACC Oct 12, 2018
Chit Chat Room AC/ACC & CPU's / Threads & Cores Sep 22, 2018
ACC Graphics - Tracks and Cars ACC - CPU Bottlenecking Sep 22, 2018
ACC Controls & Peripherals ACC keeps changing my Fanatec wheel properties - Please stop. Yesterday at 6:14 PM
ACC Troubleshooting No connection to ACC Saturday at 6:18 PM
ACC Virtual Reality ACC launch options for various rendering modes Friday at 11:15 AM
ACC Troubleshooting Crash D3D error on ACC 0.6.5 Wednesday at 6:54 PM
ACC General Discussions Google Stadia could be great platform for ACC Wednesday at 6:17 AM
ACC Virtual Reality WMR accidentally works perfectly and I don't know why. Tuesday at 2:24 PM
ACC General Discussions ACC - Dedicated server Requirements Tuesday at 1:09 PM
ACC General Discussions ACC online server status Tuesday at 9:23 AM
ACC General Discussions ACC Release Tuesday at 1:19 AM
ACC General Discussions ACC Virtual Reality Mar 16, 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