BANANA Benchmark Results Comparison

I7-8086K
16GB HyperX 2666Mhz set to use XMP
274.445 Mbeams/s


I just set turbo clocks to 5Ghz for 1 core, 4.8Ghz for 4 core and 4.6Ghz for 6 core, did not touch other settings.

Automatic overclocking wanted to put all cores 5.2Ghz, but 'surprisingly' that did not work out too well, froze at 10th car, but it did show 56 Mbeams/s for 1st car, maybe it is possible to get it stable.

I'm not sure if this would do over 300 Mbeams/s, maybe I try some day.


This was 286.480 Mbeams/s but single car result is bit lower, I did set 2 cores to 5Ghz and rest to 4.8Ghz, other settings stock as seems stable with this.

 

I7-8086K @ 5Ghz (all cores, stock voltages, stable)
301.477 Mbeams/s

 

I've read somewhere that 8086k is just 8700k renamed, but you got much better scores than the posts #873 and #874 here on this thread

How is that even possible?

 

It is binned 8700K, mostly, some rumours are that TIM is better and same as used in workstation CPUs, but don't know about that, it does get quite hot quite soon with 4.8Ghz or above, partly because of motherboard which I had to also RMA and might see if I get something better after that comes back.

Now that score you quoted was 5Ghz (just changed turbo multiplier and set to run on turbo constantly) on all 6 cores, running Windows 8.1 without all the background junk which most people have running, especially with Windows 10.

Have a browser open and it is already affecting the score.

Notice how this run of mine with stock clock I7-6700 is getting higher score than some others with I7-6700K:


That is keeping Windows quick, not running all heavy junk etc. but also having good settings in BIOS, no overclocking. Also HyperX with XMP is not that horribly slow, not the fastest, but not slow either.

Having windows defender for example causes more than 20% lower results on that machine, with AVG score is much faster, but AVG is crap in other ways, however it has low impact on system performance.

This with 8086K is though that without delidding on that motherboard, running at 4.8Ghz is max for rendering etc. After banana bench run with 5Ghz temp was already close to 90C even Banana bench is not generating very high heat. So sadly it is not quite possible to use those clocks without delidding and I'm not really keen to do that.

Cooler would easily handle 200W of heat, but it really is not much over 100W than 8086K can transmit to cooler without a delid, with better motherboard it would be possible to get more Ghz for that 100W or so limit.

Now sad part is this, with stock clocks 8086K get 43-45Mbeams/s for single car, where my 6700 gets that 40Mbeams/s, Turbo clocks when stock are just a joke in 8th gen, completely useless as it drops turbo clocks so fast with any sign of possible CPU load, set all cores to 4.6Ghz and it is getting much better result.

For me 1 car result is pretty close to 10Mbeams/s for 1Ghz with 6th gen and also with 8th gen. so 5Ghz gives close to 50Mbeams/s, 8th gen has slightly better IPC so it is bit higher, but not really far from that. If someone gets lower than that, there is some crap running on background or if using stock clocks turbo is dropping off. Also can be thermal / power limits etc. but you should get 10Mbeams/s for 1Ghz with fast 4th gen to 8th gen.

8 core intel is going to be probably even worse without overclocking as base clocks are lower and turbo is probably even worse joke.

In BeamNG currently single core is what really matters a lot, for that it is about same if you have 4th gen overclocked or 8th gen stock CPU.

 

I don't see my CPU in the google file thread, so here are the results for the i5-7500 (win10, base clock : 3.6 GHz during the test, undervolted -140mV) :

 

So in the end, this is relying on single core performance, rather than the multicore capabilities ?

 

Currently graphics rely on single core, mostly, but physics are very much using all cores, however as graphics are bottlenecking CPU single core, there is not as much gained from threading of physics.

However developers are working hard to improve the situation as it is kinda limiting maps and number of vehicles on some maps.

So it can be that in a future importance of single core disappears.

Ryzen, Threadripper, 8th gen intel and newer are not being used to their full potential at the moment, but they will be, I'm quite certain of that.

If you reduce graphics enough, disable shadows, run on low graphics etc. threading of physics can be used much better with current version of BeamNG, GPU usage can actually increase with lower graphics.

 

Good info, thanks. Only if there would be easy way to disable vegetation on certain maps and when using multiple cars. Trees are very prone to slow down things quite badly.

 

Only if you have a weaker GPU than say gtx1070, trees are very easy on CPU as they get batched up, so they are not that heavy, however if you have shadows set to full, then they might cause heavy CPU usage. I'm usually keeping shadows at partial as it is not very visible difference, however CPU load can be lot less.

 

A $400 card, that's still always in the top 10 fastest GPUs around, and is at worst 1/4th slower than the fastest ($1,200) card out today?

 

BeamNG likes fast GPUs as well as CPUs

1060 at WCUSA can be bit slow, depending what you do, 1070 might be bit better, what I know is that my gtx1080 is fine, is not bottleneck, unless doing doughnuts, then I kinda hope for RTX2080Ti, playing at 1080p

This game is heaviest game I know of really, especially in graphical department. Final Fantasy XV is also kinda heavy, but maybe they don't quite compare, I heard they had 2 euro cents bigger budget when making FF

Devs probably do their magic on gfx department before Italy is released, so it should be bit better then, I hope. Running out of options where to upgrade to :-/

 

It was just the ‘only if’ that got me, considering most everyone fits into the other side (<1070) of that if statement.

I agree though. Run a 1070 on my desktop and it is usually fine. Run a slow 1050 in my laptop and that poor thing is basically always in pain.

 

I have GTX 1070 and R5 1600 and those run the game just fine with one car with somewhat high settings, but a soon as you start to implement higher graphical settings, effects or more cars, it plummets pretty fastly and there's loads of angles around the maps where the performance really sinks deep, even thou the CPU/GPU usage % stays pretty much the same all the time. So clearly it's also a lot about the optimization as well, since the game isn't like AAA-title in terms of graphics imo.

 

i7-8700k clocked at 4.7GHz on all 6 cores, ROG Strix 1070 slightly overclocked - 253.032 Mbeam/s

 

Got a new Laptop with i7 8750H + 1070 as it is cheaper as a new GPU (Turing is overpriced completly).
Turns out it is nearly equally to my R5 1600 3,8Ghz.
Didnt any optimisation yet, will post later with undervolting and more stable higher clocks.

 

my i3-6100 and gtx1050ti 4 gig run the game pretty much perfectly, except for excessive smoke and WCUSA is very laggy in spots, but altogether for my 400 computer i built 2 years ago it is still superb lol and I might just go for an i5 or i7 skylake upgrade. I havent dont the banana bench yet on this computer actually, the last time I did (my results are in the excel document somewhere in the beginning of this thread) was on a samsung 15" laptop with an AMD A6-3400 or something at 1.5 Ghz and then overclocked to 2.4 lol i still have that laptop somewhere. I might consider doing that cooler master small water cooler for my pc too with fresh thermal paste.

 

Don't go with anything else than K models, 6700K for example is providing pretty much same performance with mild overclock than 8086K as right now there is not so much use for more cores on high graphics as one is getting overworked, but that becomes more evident with faster GPU.

I had i3-6100, then bought 1050Ti and then i7-6700 non K, which did exactly same in this game as i3 with 1 car, then gtx 1080 and also after that 8086K, which did not really do that much improvement, but with overclocking is much better, 1080 does help a lot over 1050Ti, but 1060 6GB would probably do pretty much same as 1080.

So, with 6700K and small overclock and 1060 6GB you can be running game pretty much best it will run. gtx 1070 would be tiny bit better, but no point going for anything more expensive. That might change in future of course, but trying to help you get better performance without spending too much, non K variants of those CPUs you listed are not so much of ugprade as K variants.

 

well i didnt list any actual CPU's as an upgrade, just that I would like to go from an i3 to an i5 or i7, but I will def keep all that in mind. I def want to go for a K tho for some light overclocking with watercooling. Nothing fancy, just that little watercooler for the cpu only. my 1050TI has been treating me well, but yea I def want to upgrade that and I have been thinking selling mine for, i dunno, 100 bucks or less and getting a 6g 1060, as that seems to be a great amount of performance boost for what I use it for (pretty much BNG and some other games, nothing crazy)

 

heres my pc

 

Let´s say .. I am pleased with a stock clocked 4770k .. Compared to my old FX´s 40~ it´s not bad