Actually the 8320 is a 8 core cpu with 8 threads BUT on most cpus there is a information delivery thing (I forget what they are called) and the fx cpus only have one per core. All of intel's cpus have two of them which makes it so that information is not being bottle-necked before it gets to the core unlike the fx cpu which does have some of that. This is why fx cpus have poor single core performance but decent multi-core performance do to one of their everyday cpus to be 6 or 8 cores.
Since you were doing this on a regular map, I decided to do this on a regular map... Spawnd up cars until it crashed, then restarted and spawned in one less... today I got 14. Also, I stated that my CPUs were at 3.2GHz... I could have sworn they were, but by task manager claims 2.67... although during these tests I saw it get as high as 3... so I don't know what's going on. CPU Usage... bit higher on this map then grid map, but still not full utilization. Decided to set all cars on flee as suggested. CPU with all cars running around. and again, Then I set all cars to random thinking that they could handle themselves... They could not... Not sure whats going on here. I then got one and drove away and then came back with practically no difference. I don't understand why the FPS says 12 either. It was very smooth and very playable. The rev counter was running really choppy, 12fps there I could believe, but the actual game itself was running really quite well. At least 30 fps it felt to me... either that or I just have really low standards. It felt a little slower than at 60, but not a whole lot.
I think you can get as many cars as cores your cpu has. (no virtual cores) performance starts going down after you go past that number
Must depend on CPU, mine struggles with the 2nd car and I have a quad core. I'll be happy when I replace this CPU with one from Intel.
Believe it or not, but I have no problem what so ever while running 6 cars in the Utah or even the JRI maps. And I have a stock AMD FX-9590 at 4.7GHz. Plus I'm running at 3840x2160 and could hit well above 30 FPS if I turn V-Sync off.
Due to my benchmark being quite old, and you guys reporting higher car counts than what I thought was possible on my system, I reran my benchmark on my updated system. I am now able to run 6 cars at great FPS! Much better than the 4 I was capable of on this processor in Feb. 2016. This means that my i7s cores are capable of more than one vehicle each, possibly due to HT and the great optimization by the devs. New Benchmark: https://www.beamng.com/threads/my-updated-system-benchmark.38206/ Old Benchmark: https://www.beamng.com/threads/performance-scaling-4-cars.21880/
It's very difficult to get 100% CPU or even 100% single core usage at JRI, I'm starting to see drop in GPU usage only with 6th car in JRI, I just got a thought that maybe drop is because only 4 cores and HT being used, but still with 6 cars I'm getting over 75fps at JRI, where at ECA I got around 70FPS with 5 cars and one core was hitting 100% constantly. Maps are not equal, that is why no generalization works, between maps there are much greater differences in CPU usage than there is between JRI and ECA (also graphics settings affect CPU usage, some a lots!), I don't know what exactly in some maps is causing higher CPU usage, but on some it's shadows, on some it's shaders, on some none of that, but on all reducing visual distance lowers the CPU load and allows better framerates, it is related to meshroad or similar type of objects. Maybe it is collisions, but on some maps such does not happen. That ingame performance graphing is reporting CPU frequency wrong, I think it matches to ring clock on my CPU, but not actual clock rate, it claims for me that it is 3.4Ghz while in reality it is running 3.7Ghz, even task manager gets clock wrong often. Not sure about 12FPS, but that graphing thing slows FPS down a lot, if you don't hide it with Alt-U it lies to you horribly. You have simple damage UI app enabled, that is reason why rpm counter lags, it has a bug and turning it off should give you better working UI and possible might be reason why you see 12FPS, but from that I'm not sure of. At ECA AI actually drives on road, when you start them all at highway spawn at least, they do crash a lot though, sometimes spectacular way
Actually, the entire computer came to me for just $900. I just found a really good deal on this thing and so far have been trying to find something that utilizes all its power to minimal success. Hmm... I will have to try some of those things to see of that makes the FPS count seem more reasonable.
Wish everyone could get such deals from hardware I got quite ok deal from mine too, 160 for i7-6700 and 16GB 2666 DDR4 was not too bad, another 160 for GPU, so it was not too much. Anyway about that CPU load and how performance things work in BeamNG, only common aspect with maps with higher CPU usage or low GPU usage seems to be with number of buildings / mesh objects / mesh roads in them and ECA is heavy on CPU perhaps because of that, but somehow it does not make sense to me as why. I'm thinking there has to be some kind of bug or feature in BeamNG or Torque 3D that is causing it, which is holding back potential of city maps or other more complex maps. When there is enough lot of objects it seems that even CPU load starts to come down, which for me is telltale sign of something in code causing something that should not happen, maybe something gets thrown off by too many calculations needed to be done at once, who knows? I wrote more about it to last post to this thread where I'm wondering about behavior I'm seeing which does not follow logic: https://www.beamng.com/threads/low-fps-cpu-gpu-usage.35667/ So I don't know if that is something that is already improved in next update or if it is something that gets improved in long term, my bet is on Torque 3D issue/bug, but then again I'm right about 10% of the time when tossing coin
I can run atleast 20 cars without a crash with my I7-6700k, but fps is shit and completely unplayable.
Well... this thread was from about 3 years ago... but since you brought it up... the game has had some MASSIVE improvments done to it since this thread was made... for example... In that post I managed to get 64 cars in game while before (as seen earlier in this thread) on the exact same computer, I couldn't get above 15. I would also like to point out that that massive jump in performance was seen in a timespan of only about 1 year... because that post with the 64 cars was made in 2018... If I did this test again... now in the year 2020, I might actually be able to hit the 100 car hard limit on the exact same hardware. I may even try it one of these days.
I'll put it this way in terms of performance gains. I've had the same processor, an AMD PhenomII 1090T for 11 years. Since the day BeamNG released. At first, I dipped down to 25 fps after 5 cars. Till 2 weeks ago, that number increased to 8 on the same processor. I'm now rocking a ryzen 7 2700x and 22 Covets at 30fps is a lot of fun. Improvements are MASSIVE also in terms of the engine's efficiency.
Oh shit, didn't notice that this was so old thread Didn't get any bump warning since there was one fresh post and didn't pay attention to the dates
I am not really a bump warning guy... especially if the post actually pertains to the original topic and the topic wasn't some mod that the modder stopped working on 2 years ago. So if its relevant information to the original thread, its ok in my book
My plan is to buy amd ryzen 9 3900x, idk if i spawn 30 cars a fps is going to 35fps? --- Post updated --- Because my brain always ask for "beams", each vehicle have "beams" about 4000 - 6000 beams, so my i7 4790k run 1 core=5500 beams?? --- Post updated --- --- Post updated --- Because my brain always ask for "beams", each vehicle have "beams" about 4000 - 6000 beams, so my i7 4790k run 1 core=5500 beams??[/QUOTE] And i have wld locomotive train Maximum i can 4 trains only cause this train have about 9800 "beams", So should i upgrade my i7 4790k to amd ryzen 9 3900x?? With 32gb ddr4