Your video card is designed for PCIe 4.0. That could be an additional choke point.(although I don't know how much) Prior to 4.0 becoming common, there were a lot of performance complaints from users with 2.0 compatible motherboards, those complaints lessened as 3.0 became the norm. So while your only having PCIe 3.0 'may' be somewhat of a limiting factor, I don't think it should be a huge fps loss.
We still miss some basic information. Press ctrl-C while in the track and a small graphic will appear in the top right corner. Please, make a capture when performance is the worst.
I am aware of that and asked before buying the GPU. Guy at the store told me that there is not much of an issue in my particular case in terms of performance. And I guess he was right becouse I see almost same results as before only in rfactor. On all other sims I have big boost. For examle in ac with csp and sol I cant managed 60fps on all scenarios and have big drops in corners with ai cars even with extra fx off. Now with all that stuf and 2k shadows, extra fx and rain, on heavy thunderstorm i can't go below 100fps against 23 ai even on Nords. I really don't know what can be, this is hardest puzzle I ever had.
I also run a 9600K, and I suffer from similar performance issues on rF2. There's some people talking complete nonsense here about that CPU, saying it is ancient or not capable. It's simply not true. At stock, it gets 71% of the performance of a 13900K. I run watercooling on mine, but typically I don't enable any overclocking for rF2. When driving, I get anything up to 120fps. Its a stock 2070, not overclocked. Usually, the CPU utilization as seen in Game Bar is about 8%. Since its 6 core, one core completely taken up is 16%. I typically run about 15 AI, and 12 visible vehicles. So at least as regards the physics of your own car, and the physics of the AI, there is plenty of headroom left. Depending on what I run, either IR18s or BTCC full field on S397 DLC tracks only, the GPU VRAM (8Gb) is from 80% to 97% full, and utilization is 75% when driving but will go to over 95% when watching replays, in busy track sections. Single screen, 1920x1200, static 60hz no VRR. However, I do notice quite bad spikes. CPU will go up to 35% or even more sometimes. In the latest build, this was going up to over 100% and sometimes for several seconds. rF2 itself was responsible for maybe half of this. Several background processes were also triggering. In my case, I had Dell software running in the background. This is part of software they call SupportAssist, and it ran background data diagnostics and data collection constantly. After a great deal of effort, I managed to disable this. And since then, I don't see those terrible slowdowns, when it becomes a slideshow. Here's what I think is happening. First, I have replays set to On, and at High resolution. These are saving to an SSD, 6GBs SATA. But there is definitely something wrong with this process. Sometimes I can be driving and watching replays, and CPU is 8% or less and everything fine. Switch to a different car, or change in time, and all of a sudden I'm seeing 35% CPU and vast variation in it. These spikes will sometimes occur when cars enter the pits/garages, exit pits/garages, or other events; but for the most part there doesn't seem to be any rhyme or reason why it comes off the low utilization, steady mode and goes into this highly variable one. I think some of this is down to AI code. I think some of this is due to the sound engine, which is now multithreading. We all know there are overlays in replay mode which when visible destroy fps by up to 30%. So I think there is interaction here with processes that are multithreaded, and causing action both on the CPU and across the PCIe bus to the graphics card, that tanks performance. When the background tasks were being triggered (and there will be some for replay saving), I think many threads are running. And the 9600K, it is a 6 core 6 thread processor. All those threads eventually overwhelm the processor. I think there is texture swapping going on, and interaction between CPU and GPU is probably high, but Game Bar doesn't give me the tools to see this. In my case, I have a terrible other problem: Dell lied about the specs of that machine, in that graphics sits in a PCIe 3.0 slot, but instead of running at full 16x it only runs at 8x. Past analysis by others on here, has shown a high interaction over that bus. I'd like S397 to take a serious look at what is going on with optimization of the game engine. In the past, I've had no problems with this rig. But ever since the new sound engine, I've been seeing issues, and I think its down to running many more threads. I think assumptions are being made, about HT or running many more threads concurrently, that is badly impacting those with processors without HT. There's nothing wrong with the physics, a 6core machine running at 3.7GHz is plenty fast enough. BTW, I run iRacing, AMS2 and ACC on this same machine without much issue. ACC is a bit stuttery, but it pushes the GPU very hard. I run frames limited to 84 on those, which I don't on rF2.
Here are first results You can see from fps one is against 9 ai, one against 15 ai and one against 23 but only 12 ai visible as you can see from settings
@burgesjl, they could rewrite and optimize the engine and there could be still issues. The first step is knowing what's happening, because fixing every possible present and future issue that may come up on any computer with an optimization is surreal. @darkojovanoski1986 I think you pressed Ctrl-F instead of Ctrl-C. To be able to draw some conclusions a more systematic approach is needed. You have changed the graphics settings and the fps have gone up. Now you should check which settings are worst for the performance. Anyway, I guess this will only tell us that the higher the graphics settings the lower the performance, which we already could guess. Unless you find a setting that really sinks performance a lot. You haven't provided information about CPU utilization and processes running while driving. There's a lot of variables and we'll have to go through all of them to find the cause.
Yeah this is ctrl+F Will test later becouse i am out. Strange thing, before used win 11 home and i think win 10 pro gives me better fps. Forgot to mention these results are from BTCC grid. Tried on Nurburgring at sunset against 23 ai visisible and get lowest 78fps which surprise me I must say.
ANY twilight condition or darkness or rain will result in decreased fps. The Sun casting long shadows, headlights, etc etc.
Also I tried same settings but on 4k resolition and get almost the same results but without msaa. I really don't know what is causing these issues. Can someone tell me that green and purple line have vertical readings, what is that? Of full verically that would cause spike in fps?
The purple bar is physics (CPU) and the green bar is graphics (GPU). I don't know what the vertical bars mean, but we all have them. The horizontal width is the utilization. So it seems your GPU is the bottleneck. Your graph is very similar to mine. Your other screenshots look pretty normal, although we can't see how the cores are used. Some hardware report with detailed hardware info might be interesting to see.
You mean my GPU is not utilised completly in rfactor 2? I increased resolition juat becouse of that and get around same fps same fps compares to 1080p
Going back to my previous post, it may be the older bus just doesn't let that monster of a video card xplode. Couple that with memory that can't take advantage of XMP, it just may be a combination of factors that are dragging your fps down.
The green bar is graphics processing in the CPU. It'll always be full unless certain frame limiting options are used. Purple is physics, if that nears full you can slip out of realtime (slow motion offline; stuttering online). Predictably that looks fine. The graphs above and below are histograms showing the frametime for each (graphics, physics). You don't want little spikes randomly spread as that would indicate something intermittently slowing it down. AFAIK the GPU itself has no relation to that Ctrl-C graph. The Ctrl-F is your only guide there.
The evidence is there in the CPU utilization of 35%. It should not be that high. That's two full cores. This is the sort of spike I have also been seeing; some sessions I see that, other times I don't and utilization is 3-7%. We need to find an explanation for that 35%. Can you run GPU-z? We need to check the PCIe bus can switch into x16 mode. You can also take a look at the sensors on it, to make sure there are no throttling issues. The little "?" next to the bus interface speed, click that and it allows you to run a render test that'll give your GPU and bus a workout. Take a screenshot and post it here.
@burgesjl are you sure ~35% CPU on a 6-core, running 23 AI and player physics, with shadows on high, is that unusual?
I checked my motherboard and it supports xmp but probably my ram doesn't. Also my processor allows ram speed only up to 2666 Hz like i have. If I change one stuff there I should invest in new mb, ram and cpu which I cannot allow right now..
Don't worry about XMP, there's nothing wrong there. Your RAM is already at full speed. Tweaking the timings only gets you maybe a few fps or nothing. @Lazza Thanks for the clarification about the bars. Then it's the cpu graphics thread. I agree 36% doesm't seem too much but we should see if there's some core at 100%. Wouldn't a performance log allow the people at S397 see what's happening?
