Is this possible? It seems that on modded maps the gpu usage is always 90-99% but If I spawn a second car it goes down to 30-50%. I know this is because of cpu calculations but I still think it will give me a slight boost. Also it only uses 30% cpu at all times, most I've gotten is 60% with 5 cars crashing. Also horrible fps with the gavril h cargo, always at 24 capped, no more than 24, if I used the stripped version it goes back to 140-150.
Remember the GPU cant render frames if your CPU cant keep up. Shift F1 and check Realtime %. The gavril probably has the most components and is therefore the most intensive
On this note, I have an i7-2600k and a 7970 6gb @ 1100mhz, when I spawn in about 5 or 6 cars and set my visual quality to minimum, and set the BeamNG Drive process to real time my processor doesn't go above about 80% and my video card doesn't go above about 30%. However, I still receive a significant drop in FPS. Shouldn't the game be utilizing all of my CPU? Edit: after re-reading the initial post, I realize his is basically identical.. scratch this sorry. Also, real time % is always at 100% with shift F1.. not sure what that means.
I'm not quite sure about this, since obviously one some maps where I use only 1 car I get less fps because I have less fps usage, on DRI i only get 30% and on even modded map I get 99%.
Firstly your ram is slow. That may be a bottleneck. But since so many other people have a similar issue with decent ram speeds i cant say it is that. We really need a proper system where everyone just posts in one sticky thread about this and hardware can be compared to see if there is any correlation with certain motherboards and/ or other components. There are so many threads about this right now that it is impossible to tell what the issue is.
Well I never listed my ram or it's speed, but I will list all my parts now. Windows 7 x64 i7 2600k stock speeds 8gb 9-9-9-24 1600mhz DDR3 4x 2gb AMD Sapphire 7970 6gb 1100mhz core 6000mhz memory MSI z77 mpower motherboard 750w seasonic modular PSU western digital black 1tb seagate 4tb secondary 3x 1080p monitors
Another thing to consider is that lua is single threaded. Meaning that if you spawn 4+ cars(with a quad core or better) your fps will be limited by the clock speed per core. That's why someone with an i5@3.6ghz would have better fps with 5 cars then an i7@2.66(as an example).
Im an idiot, sorry. When you posted 6gb @ 1100mhz my brain decided that context is irrelevant. My bad. What 0xsergy says makes a lot of sense.
No problem lol, I think I'm going to trade in my 4x 2gb for a single 8gb stick and overclock my CPU, I'm pretty sure I have a weak memory controller since I BSOD with even a mild overclock of 4.2ghz. If and when I do that I'll post my results. Edit: I ordered the single 8gb stick and it should be here by friday I would assume.
Q6600 @ 2.4ghz 8gb of DDR2 800mhz ram Windows 7 64bit Video card is a HD7970 CPU Usage gets up to 50% I guess that explains why I get 25FPS regardless of graphical settings when running the game
It seems that most people with that AMD CPU are having low frame rates, only thing you can do is hope that the devs can optimise the game for it. I bet when you press J your FPS will rise.
Got an AMD CPU like his, but my problem is more GPU related, as I constantly get 99% on my GTX 560, while my FX-8350 gets about 30%. Oh yeah, to the OP, when GPU and CPU usage reaches 100%, you're pretty much maxing your hardware out, meaning there no "wiggle room" so to speak.
Awe yeah! I'ma rape this game when they implement OpenCL. Anywho, just got my RAM in and ran a short 5 car destruction derby at default clocks and averaged 23 FPS. I will be back in a few moments with 4.5 ghz 5 car destruction derby and post my results. Edit: ok so I decided to go up a little bit to 4.6ghz. I averaged about 27FPS or so. Even though the framerate average only went up by about 4 frames a second, the game did feel significantly smoother. I used 1 of ever car model (1 van, 1 super, 1 truck, 1 hatch, 1 full) and I used the destruction derby map in the cement barrier enclosure for these numbers. 3.8ghz = 23FPS avg 4.5ghz = 27FPS avg
That's because you are right at the magic fps threshold, any increase above 23 FPS is going to feel very significant. If it was an increase of 4 fps in higher ranges of comparison, you would be hard pressed to notice. 5 cars is the playable limit on my 8350, too. It's basically the limit for anyone running any sort of processor under 5 ghz, regardless of what they claim. It's time for everyone (except for the OP, he has a valid issue here I think) to just sit back and wait for optimizations. If I understand correctly, everything is basically a beam connected by harmonic oscillators so it should thread nicely. Also, I profiled the application, there's certainly room for reducing execution time, let alone the nice gains with multiple car that we should see from better parallel execution utilization. Right now, It's only a few small things that are taking the majority of cycles rather than most of it being relatively equal (the point of profiling is to target these inefficiencies). Anyways, Banksy, your performance with the H cargo doesn't seem to correlate with my results, I didn't notice a significant difference between the H and the rest of the vehicles, though I would need to go double check to be absolutely sure. I can say for sure I would have noticed if it dropped to 24 FPS. I think something else may be going on. Someone mentioned memory speeds. Though I doubt it's playing any significant part in the differences, that's also something I would need to double check to be sure. Perhaps beamng is different than most games in that regard. The memory controller on your CPU is very robust, so you can definitely OC the modules if you aren't fully utilizing them right now. You really don't want to see 99% CPU utilization, there aren't any games out there with enough calculations at this time to fully utilize 8 physical cores unless it's incredibly poorly optimized to the point of broken. BF3 in a 64 person battle and crysis 3 are the games that come closest. Setting 'real-time' priority merely means that those thread(s) will get priority execution over others, such as unrelated windows desktop operations. You should see little to no difference in processor utilization by setting real time priority for any game, only reduced latency in some situations with some applications. I see mention of opencl, which would be most interesting. Similarly, I wonder if the developers have considered direct compute?