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.

The Impact of RAM Size on ACC FPS Performance

Discussion in 'ACC Hardware Discussions' started by LeDude83, Jan 15, 2019.

  1. LeDude83

    LeDude83 Alien

    It's been discussed in various threads how RAM speed and size affects FPS performance in ACC. There's usually no figures but just opinions and anecdotes. So I made figures, specifically on RAM size.

    Disclaimer: I play on a monitor. In VR it might be a totally different story.

    I increased RAM from 8 GB to 16 GB by adding 2 additional modules of the same type I already had: G.Skill Ripjaws V 3600 CL17. I then tested the 3 different benchmarks we have in this forum:

    - The ACC Benchmark (system benchmark)
    - The ACC GPU Benchmark
    - The ACC CPU Benchmark

    I ran each benchmark with 8 and 16 GB with the 2 modules for 8GB in the dual channel slots of the mainboard. The RAM speed was at 3200 MHz, all the time.

    System Benchmark:

    System Bench 8 GB 3200
    2019-01-13 15:49:23 - AC2-Win64-Shipping
    Frames: 3298 - Time: 60000ms - Avg: 54.967 - Min: 41 - Max: 79

    system Bench 16 GB
    2019-01-13 16:20:55 - AC2-Win64-Shipping
    Frames: 3307 - Time: 60000ms - Avg: 55.117 - Min: 43 - Max: 86

    Frametime Comparison

    GPU Benchmark

    GPU Bench 8 GB
    2019-01-13 15:54:30 - AC2-Win64-Shipping
    Frames: 4661 - Time: 60000ms - Avg: 77.683 - Min: 57 - Max: 117

    GPU Bench 16 GB
    2019-01-13 16:12:21 - AC2-Win64-Shipping
    Frames: 4681 - Time: 60000ms - Avg: 78.017 - Min: 58 - Max: 118

    Frametime Comparison

    CPU Benchmark

    CPU Bench 8 GB
    2019-01-13 16:00:18 - AC2-Win64-Shipping
    Frames: 5993 - Time: 60000ms - Avg: 99.883 - Min: 84 - Max: 126

    CPU Bench 16 GB
    2019-01-13 16:09:35 - AC2-Win64-Shipping
    Frames: 6302 - Time: 60000ms - Avg: 105.033 - Min: 89 - Max: 135

    Frametime Comparison


    Increasing the RAM size from 8 to 16 GB gained ~5 % FPS performance in exclusively CPU bound scenarios. In GPU bound scenarios there's no such benefit. In mixed scenarios (system bench) when both GPU and CPU are close to the limit the benefit is not present, either.

    Opposite to what some article I read on the internet claimed, there is no benefit in frame time stability when going from 8 to 16 GB of RAM, either.

    My previous tests have shown that going from 2400 to 3600 MHz RAM speed with 8GB gains ~7 % in the system benchmark (54 --> 60 AVG FPS) and going from 3200 to 3600 gains ~7 % in the CPU benchmark (100 --> 107 AVG FPS).

    It becomes obvious that the speed of the RAM is much more important than the size. So anybody looking to update RAM should rather spend that money on 8GB of very fast RAM (the fastest that the CPU will handle, the spec sheet figures can be far exceeded in many cases) instead of 16 GB of mediocre RAM.

    Also, I checked for RAM usage with 8 and 16 GB during the system benchmark runs. It was ~4500 MB with 8 and ~4800 MB with 16 GB. There seems to be a slight difference in how much RAM is assigned to the applications but the usage pretty much stayed the same.
    People claiming that 8 GB isn't generally enough if you want to run ACC are wrong. According to my findings, it's perfectly fine and even leaves a good margin for opening the browser, office documents or whatever at the same time.
    Last edited: Jan 15, 2019

  2. Similar Threads
    Forum Title Date
    PlayStation 4 Discussions Impact with other cars or barriers can cause frame-stuttering Sep 8, 2016
    Graphics and Models/Tracks Bugs & Issues Time of day/lighting impact on framerate Nov 14, 2013
    ACC Hardware Discussions Screen resolution impact on FPS? Jul 31, 2020
    ACC PS4 - XB1 General Discussions Do Thrustmaster control panel settings impact console behaviour? Jun 29, 2020
    ACC PS4 - XB1 General Discussions Boost mode impact? Jun 24, 2020
    ACC Multiplayer Glitchiness in MP and impact to ratings Feb 16, 2020
    Troubleshooting - Workarounds Huge FPS Impact w/ PP Filter ON Nov 15, 2019
    ACC Hardware Discussions Biggest settings impact on GPU? Aug 4, 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
    ACC Rating & Leaderboards SP without impacting statistics (RC,SA...) May 29, 2019
    ACC General Discussions Impact of fuel weight on GT3 cars Sep 21, 2018
    Troubleshooting - Workarounds assetto corsa wall collisions: impact too large Sep 15, 2018
    ACC General Discussions Will Enabling HDR via Windows 10 Impact Performance in ACC? Aug 18, 2018
    Physics Bugs & Issues no ffb information in crash impact Apr 22, 2018

  3. boabmatic

    boabmatic Hardcore Simmer

    this video shows that the extra Ram is most beneficial when the GPU runs out of Vram which results in maxing out system Ram causing the machine to start using the Page file.
    Shows that is most affected card was mine (GTX 1060 3gb) so I recently upgraded to 16Gb and I do get a smoother experience now with ACC in VR but the card is still struggling which is why I have an eye on the new RTX 2060 cards releasing today ;)
    MTv, mms and LeDude83 like this.
  4. LeDude83

    LeDude83 Alien

    Good addition but when you say you get "a smoother experience" that is highly non-scientific, not backed by any data and can not be tested by others. Not saying it isn't true but I don't really trust any opinion or even review by seemingly trustworthy sources, anymore.

    Many people who pay money to improve their system will say that it's better afterwards no matter if it's true. They want to justify the investment.

    One very unclear factor to anyone but OS and game devs is how they're software even uses the system resources. I think even the devs themselves sometimes don't really know that.

    So general statements are tricky and IMO you can only make statements for very specific scenarios.
  5. mikendrix

    mikendrix Simracer

    Thank you for your analysis!
    So I was about to buy another ram of 1866mHz until 16Go, instead I will buy a 3200mHz and stay with 8Go for a while.
    LeDude83 likes this.
  6. boabmatic

    boabmatic Hardcore Simmer

    I was going to try and do benchmarks before and after but was impatient and just stuck the memory straight in :) I was getting dips below 45 FPS but I feel its not doing this as often now which is giving me a "smoother experience" but dont have the data to back that up, but I'm happy even if its a placebo effect :)
    mms likes this.
  7. mms

    mms Alien

    The size of the RAM does not matter much until you have enough of it. Once you run out of RAM, the impact will be pretty significant, as the system will have to use the page file as @boabmatic said in the post above. This will result in stuttering, thus less smooth experience. You can back it up with data too, you should see lots of dips in FPS.

    If you want to test it, you'll have to use up the memory by starting some RAM-eating applications (e.g. couple of instances of windows media player loaded with some videos) until the memory is almost filled, then start ACC.
    LeDude83 and Minolin like this.
  8. LeDude83

    LeDude83 Alien

    Lol, I know where you're coming from. I usually plan on doing some in depth performance tests when I buy a new component and usually don't end up doing it.
    But I started to really enjoy all the Benchmark threads and the in-depth analysis especially for ACC so I actually enjoy this.
    And TBH, I am placeboing a little bit too: I have the impression that the system boots a little quicker and that ACC does run a tiny bit smoother. But the latter I can't prove at all. So I'm facing the decision to either enjoy this little placebo or just go by the facts and return the RAM which wasn't exactly cheap at 107 €.
  9. f_deutsch

    f_deutsch Hardcore Simmer

    This was exactly happening to me. The 1060 is a good GPU but the 3GB version limits the texture size and 3D world on ACC. I could see big fps spikes when changing cameras during replay while VRAM was used at max all the time. Changing the textures from Epic to High reduces the spikes but the textures looks horrible. When switched to a 1070 GPU with 8GB the spikes disappear completely and I could see above 5GB of VRAM usage. This was while using 16GB of RAM. My guess is that texture data were being moved between VRAM and RAM which is a slow operation compared to gather all the data from the way faster VRAM.

    @LeDude83 , on the benchmarks you performed, did your CPU was overclocked or was run at stock speeds. I have seen a more bottleneck on memory speed when the CPU is pushed above the stock speeds, like a 0.3 - 0.5 Ghz increase window.
  10. Deatroy

    Deatroy Alien

    Funny how you bought a wrong 1060.
    So you know don't save money on the VRAM and still wanna RTX 2060 with again less Ram.
    Just by the way. You could nearly afford the RTX Speed 3 years ago for 400-500 bucks and 8gb VRAM.
    Don't save money on less ram.
    You have to change GPU again too early..

    Gesendet von meinem EML-L29 mit Tapatalk
  11. LeDude83

    LeDude83 Alien

    It ran in "Game Boost", so at 4.5 GHz. Normally it's at 4.2 GHz.

    And I can second the thoughts on VRAM. I experienced stutter with AC a while ago even with VSYNC enabled and at steady 60. I checked VRAM usage and it was at ~75 % of the total capacitance. I think I had a 4GB 770 back then. When reducing the settings so the VRAM usage would drop below that mark, the stutter was gone. I have no data to back that up but I'm pretty confident it wasn't placebo as it was very obvious.

    But I'd say that even when the system RAM starts being used by the GPU you'll see stutter, not only when the pagefile is being used.

    VRAM usage is pretty high in ACC, indeed - I saw 5.8 GB being used or so.
  12. martcerv

    martcerv Alien

    Ram size as others said is only an issue when you run out, speed will give some performance boost as you found but your also getting a boost from going a single channel to dual channel setup by adding a second stick. This is where your likely getting any gains from running 2x 8gb vs 1x 8gb if the same spec ram. If your not running out of ram you would likely see the same gain going from 1x 8gb to 2x 4gb of the same speed and latency ram and this 2x 4gb setup should be much the same as 2x 8gb if your not using more then 8gb.

    If you had a quad channel system then again having 4 matching sticks at the same spec should be slightly faster then single or dual channel at whatever capacity as long as you don't need more then you have.
    Novotny likes this.
  13. Akbalder

    Akbalder Gamer

    Did you notice a difference of quality with 16GB?

    For me, when racing with many cars, the quality of the textures is often degraded.
    Does switching from 8GB to 16GB of RAM improve it? Or is the quality of the textures limited by the graphic card?
  14. gt362gamer

    gt362gamer Gamer

    One question over here, having an i5 4690k at 4.3ghz, should I get 2133mhz cl9 8gb or 2400mhz cl11 16gb? Your tests show an increase with the ram speed, but I don't know if you also squeezed lower latency on lower clocks to imitate real life situations like mine. 9/2133 is lower than 11/2400 or 10/2133, but I don't want to run out of ram either. How much does ram latency matter on this game?

    Edit: seeing that increasing voltage ram appearently increases cpu temps quite a bit, I think I'll have to settle with 2133mhz, as they work at 1.6v instead of 1.65v of the 2400mhz ones. Therefore, unless latency matters very much, I'll go with the 2133mhz cl10 16gb kit. Can you test the difference how much latency matters if you increase it by a 10% at same clock speed?
    Last edited: Dec 18, 2019
  15. Serge M

    Serge M Alien

    There’s a few more complications with ram above the usual clock speed, timing and capacity. One of the interesting ones that can make running more ram then you need worthy is ranks which can give a fair performance increase, over 10% in some cases.
    Ideally you would want to run dual rank of the fastest memory, to get that you either need two dual rank sticks (chips on both sides) or 4 single rank sticks.

    Now to get those specs you would need either 2X16GB sticks that are dual rank or 4X8GB sticks.

    There isn’t really any dual rank 8GB sticks available any more so those are your only options if really wanting to maximise performance, and both setups will give you 32GB

    For those that want to read up on all the fun stuff here a good link: https://www.tomshardware.com/reviews/amd-ryzen-3000-best-memory-timings,6310-2.html
    seb 6th gear likes this.
  16. seb 6th gear

    seb 6th gear Racer

    That's in the ballpark of the overall memory bandwith gain going from SR to DR and these gains becomes apparent when your game is CPU bound, wich is the case for ACC especially in VR. Removing or reducing CPU bound scenario helps to get a better GPU usage so more avg fps.
  17. Jayk12

    Jayk12 Rookie

    Quick hardware question guys. I have i5-6600k clocked at 4.6Ghz, on the Intel site it states the CPU supports 2133mhz RAM, my motherboard supports 4133mhz. Is there any point in getting faster ram, I'm on 8GB 2400, was about to get another 8GB 2400, but unsure of the above if i should get 16GB of 3600 etc. Cheers.
  18. LeDude83

    LeDude83 Alien

    I had 3600 MHz RAM running on an i5 7600K and it ran just fine. I compared it to 2400 MHz (7600K spec) and the FPS gain in CPU bound scenarios was 5 - 10 %. Whether or not this is worth it is up to you.

    After that I build 2 new PCs and for the ACC PC I chose 3000 MHz RAM. It's pretty cheap, not too much OC compared to 3600 MHz and it already gave pretty good performance in ACC over 2400 MHz.

    To put it in a nutshell, according to my experiments, 3000 MHz was the bang/buck ratio sweetspot. I recommend going for such RAM and maybe invest a little extra money for a CL15 kit.
    Jayk12 likes this.
  19. Jayk12

    Jayk12 Rookie

    Great thanks. I'll get some 3000/3200 CL15. Next up I've just got to decide if my CPU needs an upgrade, ACC was using 65% of the processor so total usage was 75%. Wonder if changing my GTX1080 Hybrid clocked to 2025mhx to a GTX1080Ti is worth it.
  20. LeDude83

    LeDude83 Alien

    The bottleneck in your system is the CPU. Going from 4- to 6 core CPU (physical cores) makes a big difference - that's why I swapped my 7600K for an 8700. But I found out that hyperthreading isn't needed - I made an A/B test and the difference was null. So I could also have gone cheaper with a 9600 or even 8600. But AMD will give you a much, much better bang/buck ratio - my second PC has a Ryzen 5 2600 installed and that CPU performs just ~15 % worse than my 8700. It has 6 physical cores with HT and can be OC'ed without issues up to 4.5 GHz AFAIK. It was only 120 Euros so yeah...get a Ryzen. Mainboard, CPU and that new RAM will cost you just above 300 € and you will have a way better system balance with that 1080.
    And BTW - don't trust those % values you get from Afterburner etc. The bottleneck in ACC happens somewhere under the hood in the way RAM, mainboard and CPU interact. Your GPU can be bottlenecked, already even if Afterburner says: CPU load = 75 %. Bullet proof way to find out is to run uncapped FPS and observe when the GPU goes below 100 % - this is when something in your system bottlenecks the GPU.

    As for your idea to upgrade GPU - the 1080 is just fine for ACC. I have a 2060 Super which performs on par with the 1080. Both these GPUs are very well suited to run ACC with good graphics in 1080p or even 1440p. From your 1080, the next reasonable upgrade would be the 60 or 70 model of the next NVIDIA GPU generation.

    Hope this helps.
    Jayk12 likes this.
  21. Jayk12

    Jayk12 Rookie

    Thanks, really appreciate that information. I'm taking a look for some ram based on your recommendations, i found a 3600mhz 2x8gb. Its not CL15 but the latency looks the same as a 3200mhz CL15 kit and only £9 difference in price. https://www.overclockers.co.uk/patr...dual-channel-kit-pvs416g360c7k-my-104-pa.html

    CPU wise I'd probably just want to stay intel and swap out my cpu for now, I've seen some i5-8600k for reasonable price now and they should fit in my Asus Maximus VIII Impact LGA1151 socket according to https://www.game-debate.com/motherboard/index.php?mot_id=3550&cpuList=Asus MAXIMUS VIII IMPACT

    I use VR, but these changes will help regardless of VR or not.

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