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.

Test: CPU Core count and RAM scaling in ACC, AC1 and R3E

Discussion in 'ACC Hardware Discussions' started by Daniel Costa, Aug 20, 2019.

  1. Neilski

    Neilski Alien

    Yes, I was looking at those results, in the CPU test section, which shows a huge advantage for the i7 (30-40%) because more threads are doing something useful. Did you mean the actual 3Dmark scores? Not sure why they would be directly relevant here, when we have data from actual game testing (but see below).
    OK, I realised when I read your reply that I was talking about AC and you were talking about ACC! This thread covers both (+R3E) and I now have no clue which game @gt362gamer is talking about cos he never said, LOL, but it's an ACC forum so that would be a fair guess I suppose :D
    And wow, when I look at the HT on/off results from post #1, HT does make a difference for ACC (unlike AC) but things are significantly worse with HT on (moderate effect on average FPS, more dramatic effect on the outliers). Clearly you had the opposite result; no idea what we should conclude from that other than "testing results have been inconsistent"...
    Perhaps your testing was done quite differently from @Daniel Costa's?
     
    After_Midnight likes this.
  2. gt362gamer

    gt362gamer Gamer

    Yes, I was talking about ACC. :)
     
    After_Midnight and Neilski like this.
  3. Like you said it's an ACC thread so I supposed that we're talking about ACC. :p The 3Dmark scores are just a part of my answer to the request of @gt362gamer to share benchmarks of the I5-4690K/I7-4790K.

    Out of my experience with this game I can say there is a disconnection between pure FPS and smoothness. Similar to AC1.
    Even with frame rates above 120 FPS you'll notice (micro) stuttering if one thread will reach it's limit. The physics or rendering thread for example. The biggest difference between HTT (I7-4790K) on/off was not the frame rate itself but the smoothness of the game.
    It is just my experience and I like to share it to prevent guys like @gt362gamer to waste their money like I did.

    Greets After_Midnight
     
    gt362gamer and Neilski like this.
  4. gt362gamer

    gt362gamer Gamer

    Did you also buy faster ram to compensate slow spec ddr3 ram/4 threads cpu and you were dissapointed with the results? :/
     
  5. No but I'm sure the faster RAM only will not lead to a noticeable improvement in your case. Even the RAM would be fast enough your CPU will stay as limiting factor.

    To buy first a used I7-4790K instead of invest into a complete new system was a waste of money in my case.
     
    Last edited: Nov 17, 2019
  6. gt362gamer

    gt362gamer Gamer

    So, even If I had a gsync monitor+ faster ram, the stuttering would still be there with 60fps cap? According to Rivatuner, the worst frametime I remember during the run was about 22 miliseconds, which is around 45 fps. And usually stays very close to 60fps, ranging between 14-19 ms. Or cpu spikes are not counted there?
     
    Last edited: Nov 17, 2019
  7. Yes. In my opinion faster RAM will never eliminate a CPU bottleneck.

    But it's up to you to try it out. :)
     
  8. gt362gamer

    gt362gamer Gamer

    Well... if it's huge I understand it won't fix. But the thing is, most of the time, the frametime is lower than 20ms, so most of the time the game is above 50fps, so it's not an enourmous gap like, let's say 30fps frametime. The CPU is able to run the game 99% of the time with 24 AI's between 50 and 60fps (and if I eliminate the cap it can reach sometimes the seventies, although not for long), with some dips in the mid 40's (22-23ms). Also, in top of that, I hate tearing as much as stuttering, so in top of CPU frametime then it comes the vsync pausing the cpu and gpu to wait for the next refresh to do something...

    I recently checked this video, and faster ram seems to help a lot the i5 2500k, which is slower than my cpu:


    Also it seems to help with a faster i7 4790k:


    So I get it won't do miracles, the point is, do I really need to upgrade heavily my PC in order to get an acceptable gameplay? Or adaptive sync, even without changing ram, would make it fine (but not perfect) with a 60fps cap? My monitor has light leaks so I have to change it anyways. If someone else has one and a similar cpu (or a cpu that can emulate the i5 raw power disablying cores/ht), I would apreciate a lot the feedback. This way we could find out an answer to why your experience seems to be so off compared to OP tests (and save myself from being too offtopic here, as my personal interest to know the answer would match the purpose of the thread, :D).
     
    Last edited: Nov 18, 2019
    After_Midnight and Neilski like this.
  9. What you've descripted about your current gameplay in ACC is exactly what I've experienced with my i5. Just with higher frame rates because the GTX1070. The change to the I7 elimated the stuttering mostly. Without touching the RAM.
    Frame caps at 59/60/61 resulting in heavily tearing also at my current system. But anyway good luck mate.

    @Neilski Just out of interest:
    Did you noticed something like I've descriptive here?
    https://www.assettocorsa.net/forum/...formance-discussion.60924/page-4#post-1100606
     
    gt362gamer likes this.
  10. Neilski

    Neilski Alien

    Nope, hadn't seen that, and am not sure what problem you are hitting.
    It sounds a bit like (but not quite the same as) the kind of weirdness I was seeing until I turned off the temporal anti-aliasing.
     
  11. Exactly this. It's the temporal anti-aliasing and it happens at all settings (mid/high/epic). Thank you very much for pointing me in the right direction. If I turn it off the picture becomes much sharper and the "stuttering effect" disapears.
    But now the picture is shimmering and flickering all over the place. :(
     
    Neilski likes this.
  12. f_deutsch

    f_deutsch Hardcore Simmer

    It did on my particular scenario.

    Couple of days ago I updated my BIOS version and lost all my Profiles. I came close to the same FPS I had on an ACC benchmark replay I do with Afterburner and RTS. I was falling ~8 fps short from original 112fps average until I realize my memory was not set to XMP profile while using the same 3200 speed. I recover those 8fps and also ~10fps on minimum fps stat.

    I believe it will be very dependable on the processor in use. I3-8350K @ 4.2Ghz. The faster the speed and the higher number of cores the CPU has could result on heavy bandwith memory use. The UE4 seems to be CPU/core hungry on this.
     
    Neilski likes this.
  13. Really an interesting read. It's also interesting to see your I3 is able to benefits of DDR4 3200. But did you really remove the bottleneck of the CPU? Did it removed the stuttering you should have with 4 cores/threads?
    Like I've said even with higher FPS stuttering appears if one core reach its limit in ACC. What's your core usage ingame while racing?

    ...and the application in use.

    Of course it is.
     
    f_deutsch likes this.
  14. Neilski

    Neilski Alien

    Yeah, anti-aliasing is a good thing in general but that temporal version is just... bizarre. Perhaps there are combinations of settings for which it does a "nice" overall job without intolerable artefacts - I didn't look very hard - but my experience so far has me thinking "why does this even exist?".
     
    After_Midnight likes this.
  15. f_deutsch

    f_deutsch Hardcore Simmer

    For CPU bottleneck, the CPU was able to deliver more rendering requests to the GPU, so I believe it lowered the bottleneck.
    I do not see or experience stuttering. The CPU usage is around 80% percent. I do not have an exact measure of it, can go up to 90% and down to 60%. Individual core usage is well spread.
     
    After_Midnight likes this.
  16. f_deutsch

    f_deutsch Hardcore Simmer

    Couple of days after christmas I will be replace the I3-8350K with a I5-9600K which has 2 more cores and ~ 11% higher memory bandwidth. I need to think on some metrics that would be interesting to measure for comparison.
    Want to see if two more cores may reduce the % usage of the other cores evenly. The memory bandwidth should increase FPS, particulary minimum FPS. All of these while keeping the same CPU speed.
     
  17. gt362gamer

    gt362gamer Gamer

    So it really seems that the ram is really bottlenecking in my case, as an i5 4690k at 4.3ghz shouldn't be that far from that cpu... But having half of ram speed has to be a key factor to explain the difference (although having less latency so not exactly half the bandwidth).

    A way to compare it with my cpu would be downclocking your cpu + using lowest ram speed possible (2133mhz for ddr4 I guess) so in cinebench R15 you would get the same score on the multithread test: 630. Then, testing the framerate at the scenario and settings I was testing my cpu (with 60fps cap with Rivatuner) and, again, doing it at the same cpu clock but with the faster ram speed. I would really appreciate it if you don't mind.
    https://imgur.com/a/W0CSvLj
     
    Last edited: Nov 20, 2019
  18. f_deutsch

    f_deutsch Hardcore Simmer

    I can provide the replay I do for testing and simple instructions to follow, requires Afterburner and RTS.
    I could set my CPU and memory to similar running specs and try same ACC config. What is the video card and monitor you have?
     
    gt362gamer likes this.
  19. gt362gamer

    gt362gamer Gamer

    A 4GB 960, but it really doesn't matter. It only becomes a bottleneck in this scenario in some closeups in the field before starting the race. I basically care about ingame onboard fps, as now I'm having cpu induced stuttering (most of the time the gpu is under 80% usage, while the cpu have a harsh time). The monitor is a cheap tv bought in 2011, so just a very basic 1080p 60hz set. Thanks for the feedback. Also, to make the test more realistic, I will give you the bandwidth results I have in my pc right now. Give me a moment to find it.

    Edit: 62.9% (21.4GB/s - 20GB/s - 59.8ns) in userbenchmark test.
     
    Last edited: Nov 20, 2019
  20. To be honest it is hard to believe your CPU is performing that well and you never run into this issue (I5-4670K@4.1GHz / 20AI's):
    [​IMG]

    And like I've said also with an I7-4790K@4.4GHz I was running into the CPU bottleneck in ACC. Not that often like the I5 but it still happened (multiplayer and racing against the AI).

    The I7-4790K should outperform the I3-8350K by a small margin in multicore applications and gaming:
    https://www.cpu-monkey.com/en/compare_cpu-intel_core_i7_4790k-411-vs-intel_core_i3_8350k-837



    Here are the UserBenchmark results of my current system. Maybe interessting for you:
    96.9% (96% 33.6 GB/s - 91% 32 GB/s - 80% 50.2 ns)

    Greets After_Midnight
     
    gt362gamer likes this.
Similar Threads
Forum Title Date
Chit Chat Room 95% cpu occupancy test. Oct 16, 2014
ACC PS4 - XB1 General Discussions Latest patch Oct 20, 2020
ACC PS4 - XB1 General Discussions Latest on console update Aug 26, 2020
ACC Physics Suggestion: Car setup aero test allow lower ride heights Jul 30, 2020
Screens & Videos D1NZ ASB Showgrounds 2020 Round 4 Layout / GoPro Triple Screen Test Jul 8, 2020
ACC Screenshots & Videos Nürburgring GP Sprint Race / GoPro Triple Screen / Epic Video Settings Test Jul 8, 2020
ACC PS4 - XB1 General Discussions Fastest lap Jul 7, 2020
ACC PS4 - XB1 General Discussions Just bought ACC to test on PS4 Pro after awful Xbox One X experience! Jun 24, 2020
ACC Hardware Discussions FFB on latest release May 29, 2020
ACC Hardware Discussions What is the fastest spinning consumer wheel May 16, 2020
ACC General Discussions Empty Tracks for pre-season testing / time of day adjustments in minutes May 15, 2020
ACC Multiplayer Report system/Vote system on malicious behavior (continuous blocking +brake test) Apr 29, 2020
ACC Gameplay AI always pits for Wet tyres at slightest rain Apr 13, 2020
ACC Physics Which Track Temp is the fastest`? Apr 7, 2020
ACC User Interface freeze/crash since latest update 1.3.7 Feb 29, 2020

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