We all know that a 4 core is a lot better for BNGD than a two core, and 4c/8t gets about 20% over no HT, but has anyone tried this game with a six core extreme or an (dare I hope?) 8 core Xeon? I may be able to get a 3930k if I get a graphics card refund, and I was wondering if there is much difference.
Well, yes. But I mean how much more? HT brings 20% more performance, so does the return ever decrease, or could something like a 64c Xeon machine beat the crap out of this game.
Xeons arent actually that great for gaming compared to core i5/i7 or AMD FX. Monstrous core counts (however the largest they do is 12 cores and non server editions of windows only support 2 CPU's so thats actually 24c/48t) which really do help for servers. But they aren't really aimed at all out CPU performance. Huge core counts really don't help for most games, BeamNG is one of the few exceptions. Xeon base clock speeds are often lower than i5/i7 and they don't overclock as nicely. There is little reason to use one for gaming. Probably more than powerful enough for BeamNG though before counting the multitasking they bring to the table. If they had clock speeds similar to their core i5/i7 (or heaven forbid AMD FX) brethren then they would be monsters. But most games benefit more from the higher clockspeed than the extra cores. It is rare for a game to fully utilise a quad core even. How many threads would BeamNG max out at or is it a case of it trying to make 1 per car whenever possible etc?
As far as I know there is no maximum; if you had a 32 core processor with a high enough clock speed you could probably run almost as many cars smoothly I have an i7, quad core hyperthreaded, at 3.6 Ghz, and I can run 4-6 cars at a playable framerate (>30 fps). None of us have processors with more than 8 threads to test with, so we cannot say for sure what it would scale like, but in theory you could have a lot more cars at a playable framerate if you had enough threads.
on another note, if you had an insanely high clocked cpu, how would the performance compare? let's say a 16ghz (lol) single core vs a quad core 4.00ghz?
I don't think it would help much at all. You would be able to simulate one very complex vehicle smoothly (if it existed) but not any more. Clock speed is not all that relevant these days. Remember the 3 Ghz Pentium 4s from over 10 years ago? And now clock speeds are only a bit higher than that.
Just out of curiosity, what kind of FPS do you get on puregrid with low graphics settings (so we isolate it to just the physics) with 1, 2, 3, 4, or 5 D15s?
guess-the-fps-drop I don't actually know how to spawn more than one car because it's never really been an option. I'll try it out though. The game will probably just crash lol. I don't know how applicable a p4 is to anything now though. No processors right now have a similar setup like the netburst p4's to my knowledge.
I'm curious if a 3930k would run this game alot better than my 3770k would then. It has 2 more cores. My 3770k runs this game fluidly but if just curious if anyone has a 3930k or 3960x that could shed some light on this.
Well if what gabester said is true (which is something I suspected was the case) then the extra cores would allow for more vehicles to be simulated reliably at once (but no difference to performance for a single vehicle). Performance of a piece of software is somewhat obviously dictated by the number of instructions which can be executed in 1 second, however no CPU actually executes 1 instruction from 1 clock signal. In reality a clock signal causes a certain number of "cycles" to occur within the processor, certain instructions take differing number of cycles (with alot of efforts made on each new CPU generation to decrease the number of cycles an instruction takes). Branch prediction and pipelining (and haswell has a deeper pipeline than P4) also throw spanners in the works. End result is that a pentium 4 clocked at 1ghz would actually be slower than a modern ivy bridge or haswell powered computer also clocked at 1ghz. Alot has changed under the hood to increase the speed without altering the base clockspeed over the years.
Haswell-E also support for DDR4 memory clocked up to 2133MHz as well, 20MB L3 Cache reading from some websites about that..., its due for release sometime around 2014 I am unsure of the Quarter which it will be released in
Has anyone tested this (for whatever reason) on a dual-CPU server board, if possible? I'd like to see how and if it addresses multiple physical CPUs.
Considering most software just addresses threads and then windows decides where the thread is physically executed, it should be fine. Would be nice if someone could try though
The image uploading thing on here is annoying so I used imgur http://imgur.com/a/9vISs The game did not crash.
the potential is there, I mean if someone built a Dual Xeon Processor computer and then put it up with beastly graphics cards in it and huge Ram, I would expect High FPS because its got more processing power than any current i7 computer as of yet the power difference between a i7 and a Xeon is Considerable... I hope my punctuation and use of commas has helped out here..