Your performance testing results

0.9 is close, so final chance to do some benchmarks to compare 0.8 to 0.9.

I just keep logging data here, maybe it helps someone some day.

Latest Update *** West Coast USA*** ~~ Updated instructions so it is lot easier to participate ~~

\\\\\\\\\\\\\\\\\\\\\ TOOLS AND HOW TO ////////////////////////
Use Fraps <Link> to get fps logged. It is enough to have Fraps running.

Use HwInfo64 <Link> to log GPU usage, framerate and CPU usage (among lot of other performance information) during benchmark. It is simple, just start HwInfo64 after you have started fraps, press that button shown in image and choose place for the log file where you can find it later

After done, press that button again and send log file to me or post in this thread, I can make graphs for you, doesn't even cost anything and you can do this while you would be playing BeamNG anyway, so it takes only few seconds.

Other tools work too, but this is what I used and to minimize variables everything should be kept as same as possible for testing to serve any purpose to new comers and who knows maybe older members too.

------------------------------------------------------------------------------

Test should be done with following settings so it is easy to compare systems, which might help people to decide which hardware upgrade might work for them.

-Default view behind the car, driving such pace that not crashing or tires smoking. See videos for routes (where available).

VSYNC OFF
FPS LIMITER OFF
RESOLUTION:1080P (1920 x 1080) full screen


-Vehicle used was default D15 V8 4WD Auto
With T-series cement mixer there is around 5fps less.
--------------------------------------


<Info> How things work, short and long version, long version contains some performance improving tips too:
https://www.beamng.com/threads/your-performance-testing-results.37866/page-2#post-568949

/////////// My results \\\\\\\\\\

System specs:
CPU: i7 6700 @ 3.4Ghz (3.7-4Ghz Turbo, depending from active cores)
GPU: Gigabyte GTX 1050 Ti OC 4GB 1341Mhz/1455Mhz (boost) 3504Mhz memory
RAM: 16GB DDR4 2666Mhz Kingston HyperX
Motherboard: Gigabyte Z170X-UD3


East Coast USA, MIN 62, AVG 72, MAX 98
Starting from highway spawn, driving around the loop from town back to highway spawn
GPU limited all the time, GPU usage 99-100%

gtx1080 (game v0.14) Min 73 Average 95.39 Max 126
West Coast USA, MIN 38, AVG 51.3, MAX 63

Spoiler: FPS chart

View attachment 370444

Gridmap, MIN 106 AVG 141 MAX 158
Really hard to describe with any accuracy, but straight from spawn, then right clockwise after long tube things, maybe 3/4 circle inside fences, after that fence crossing and drove under the oval loop road.
GPU limited all the time, GPU usage 99-100%

Hirochi raceway, MIN 66, AVG 75, MAX 92
Around short version of track
GPU limited all the time, GPU usage 98-100%, 98% might of been just a hiccup.

Spoiler: VIDEO

Jungle Rock Island, MIN 60, AVG 73, MAX 103
Starting from Observatory spawn, going loop around the island and stopping to front of the tunnel.
GPU limited all the time, GPU usage 99-100%

Small Island, MIN 70, AVG 93, MAX 111
Around the island loop
GPU limited all the time, GPU usage 99-100%

Utah, MIN 69, AVG 83, MAX 102
Starting from industrial spawn right from intersection, right from highway over the bridge and driving the loop back to industrial spawn, there is one left turn at some point I think?
GPU limited all the time, GPU usage 99-100%

------- Mod maps below -------

Advanced Island Redux 2.11, MIN 72 AVG 92 MAX 113
From spawn, tarmac road over the bridge, end of the road intersection to left, end of the road intersection to right, around full loop, then back to spawn.
GPU limited whole time, good performance, pretty map with nice roads.

Altitude 1.5, MIN 42, AVG 67, MAX 99
Default spawn, drive to jump, then right turn to dirt road, at end of the road turn left and continue as far as road goes, took around 14 minutes.
GPU limited whole time, but noticed that on full screen GFX thread caused some 40% higher load to core running GFX thread than windowed mode, peaking to 100% at times, needs investigating as can cause lower min FPS for maps that have GFX thread limits.

Beam Mountains MIN 63, MAX 117, AVG 86
Drove around 12 minutes and got hopelessly lost, GPU limited whole time, two hiccups at beginning, which I don't count as lowest FPS.

Spoiler: FPS Chart

Desert Highway UPDATED Version 13 13.01, MIN 73, AVG 116, MAX 144
Driving full loop, when ending up to T-intersection turning right and ending up to spawn location.
CPU limited when camera facing spawn area about half the distance(?), GPU usage 73-78% when CPU limited by CPU's GFX thread.

Spoiler: FPS chart

Spoiler: VIDEO

Glass View Mountains Map V 0.8 Min 37, AVG 60, MAX 84
Driving full loop of paved roads, starting from gas station and returning to gas station after full loop.
When camera facing towards town, CPU's GFX thread was limiting factor, GPU load dropped to around 70% FPS really low when that did happen, turns often bring view towards town so constantly dipping under 60FPS

Russian Roads NP 0.24, MIN 73, AVG 88, MAX 103
Starting from town gas station, drove to Stop sign, turned right and drove full loop back to gas station.
GPU limited whole time, heavy CPU load also, so slower CPU would not be able to feed GPU on this version.

Spoiler: FPS Chart

Spoiler: Russian roads old version

Russian Roads NP 0.24, MIN 76, AVG 91, MAX 104
Going where road goes without turning from intersections until finding road blocked by red/white barriers after bit over 5 minutes of driving.
GPU limited whole time, really good performance, even buildings seem not to cause GFX thread overload.

Seaside Roads 0.81 MIN 44, AVG 88, MAX 121
Attempted to drive around the island, got lost, took a long time to eventually find back to spawn, there is lot of roads with very steep hills.
GPU limited all the time, one location quite beginning of journey, some turns and lot of trees at the spot, had very low framerates, rest did run above 60fps, make sure to drive few circles before starting benchmark

So-Cal Interstate 0.8 MIN 35, AVG 90.6, MAX 137
Very long drive, around 23 minutes at high speed and still covering only small part of the roads, map is vast.
Min FPS is bit unfair on this map as there is possibility to change visual distance which helps a lot, also low FPS issue comes from some issue in Torque 3D where really lot of objects cause heavy CPU usage and/or drop in GPU usage.
Mostly GPU limited, only when camera facing certain direction with lot of mesh objects, map becomes heavily CPU limited for short time only. GPU load 51-100% with median of 99%

Spoiler: FPS chart

Chart is rather condensed as logging time was so long, very difficult to read if I stretch it to full length

Tennessee USA Roane County Beta 5, MIN 39, AVG 77, MAX 103
Starting from Burger world spawn where CPU limited, right, then right, soon became GPU limited with higher FPS then after some time road ends and drove bit towards right again where again CPU limited
At spawn area heavily CPU limited by CPU's GFX thread, GPU usage 60% at lowest, 70-80% typical and 100% away from build areas.
<Info / Tips>

Spoiler: How to double FPS

Press F11, type info to box as shown, find visual distance and type in 2000, click play button from toolbar Video from So-Cal interstate that shows effect of visual distance

(beginning and end CPU limited, rest GPU limited):

Spoiler: VIDEO

Trial Mountain <Link> MIN 64, AVG 82.8, MAX 100
Driving one full lap.
GPU limited all the time, performance bit better than East Coast USA.

Spoiler: FPS charts

Spoiler: VIDEO

//////////// Results from @BombBoy4 \\\\\\\\\\\
Hardware: I3-6100, RX460 2Gb, 8Gb DDR4, 1Tb 7200RPM

ECUSA: Min 40, Max 81, Avg 63.814, Usage 98-100%
Small Island USA: Min 36, Max 115, Avg 95.082, Usage 98-100%
JRI: Min 43, Max 124, Avg 93.262, Usage 97-100%
Hirochi Raceway: Min 73, Max 97, Avg 83.225, Usage 93-100%
Industrial: Min 52, Max 106, Avg 77.807, Usage 99-10%
Utah: Min 66, Max 102, Avg 85.857, Usage 99-100%
Gridmap: Min 89, Max 175, Avg 127.68, Usage 89-100%
Grid Small Pure: Min 169, Max 216, Avg 194.202, Usage 76-100%
More results using different graphics settings in his post https://www.beamng.com/threads/your-performance-testing-results.37866/#post-568144

///////// Results form user @ForZac
Hardware: Ryzen 5 1600 (OC 3.875Ghz), gtx 1060 6GB, 2x8 gbs of ddr4 2400mhz and the 2x1 tb hdd 7200k rpm + Boot SSD beeing crucial mx500

ECA, Min 22, AVG 82.584, Max 117, GPU min usage 70-80%
Small island Min 72, AVG 105.831, max 127
JRI, Min 69, AVG 105.854, max 147, GPU usage up to 85%
Hirochi raceway, Min 105, AVG 124.397, Max 145, GPU min usage 93%
Gridmap, Min 110, AVG 160, Max 224, GPU usage average 81%
WCUSA Min 50, AVG 80.577, Max 146, GPU usage 74% avg
UTAH Min 91, AVG 129.545, Max 169, GPU usage 91% average

 

You must have gotten a bad card because my RX 460 (which is slightly worse than a 1050 non Ti) can run those maps/cars at those settings and FPS. When I'm not lazy I'll update this post with proof.

Max w/ half updates per second (dynamic reflections) on JRI makes about 55 FPS

 

You can still copy the game files and store on a hard drive or your desktop and then test later if needed.

 

You need to do exact same settings with exact same methods to really be able to compare anything. There are all kind of generalizations, assumptions etc. at the end though, it seems all of those are wrong, have just one setting differently and FPS can be a lot different, so test should be done methodically with at least some precision.

 

I can run a test like this on my computer copying your stuff to the letter.... it looks like you spent a while doing it though, so it will take some time.

 

Actually it is quite quick to do, many maps it is about 5 minutes to drive slowly around, after putting settings and having software ready and set. Interested to see your results, I will edit them to first post so we can have some kind of table to refer, hopefully

Most challenging part is to keep on eye of GPU usage and not to hit tree, I have no idea how to get MSI Afterburner logs to excel, that would be very useful though.

 

Can you take a random screenshots on one the maps ? I Want to see the difference between your pc and my laptop

 

Alright, yea, I can do all of that. Also, my computer has dual graphics cards, but I think they are kind of old and I am not sure crossfire is working properly, so when I do them, I will only be counting one of the graphics cards as working for the performance I get. I think my computer will be completely graphics limited for that reason.

 

Which map would you like? My intention has been to make video too, but haven't got around of that yet.

It would be nice to have this kind of overlay in video and marker pointing where current position is, but I have no idea how to make that happen, but maybe graphs could be useful too. X-axis is seconds, but it is bit hard to know which location each second was of course. When recording video there is then some kind drop in FPS, but maybe screenshots that show FPS might help a bit too at least easier to do than more detailed stuff.

Need to experiment with that part, graphs are at least easy for everyone to make as Fraps saves values to file you can open in Excel and insert line graph.

 

Tennessee USA Roane County
Advanced Island Redux
Desert Highway

 

My computer can't handle maps over 1.6 sq mi lol

 

Hardware: I3-6100, RX460 2Gb, 8Gb DDR4, 1Tb 7200RPM
Conditions: Stock truck, default map loading points, orbit cam, no prerun (to load map chunks), cleared cache, 1080p
Replace D-Series w/ T-Series = WIP

Spoiler: Vanilla Maps

Settings:

Benchmarks:
ECUSA: Min 40, Max 81, Avg 63.814, Usage 98-100%
Small Island USA: Min 36, Max 115, Avg 95.082, Usage 98-100%
JRI: Min 43, Max 124, Avg 93.262, Usage 97-100%
Hiroshi Raceway: Min 73, Max 97, Avg 83.225, Usage 93-100%
Industrial: Min 52, Max 106, Avg 77.807, Usage 99-10%
Utah: Min 66, Max 102, Avg 85.857, Usage 99-100%
Gridmap: Min 89, Max 175, Avg 127.68, Usage 89-100%
Grid Small Pure: Min 169, Max 216, Avg 194.202, Usage 76-100%
Cliff: Min 45, Max 157, Avg 128.933 Usage 97-100%
Derby Arenas: Min 68, Max 129, Avg 93.933 Usage 98-100%
ETK Driver Experience Centre: Min 5, Max 102, Avg 72.417 (1 Major Lag Spike) Usage 99-100%
Dry Rock Island: Min 8, Max 79, Avg 62.317 (1 Major Lag Spike) Usage 99-100%
Port: Min 52, Max 58, Avg 71.983 Usage 99-100%
Small Island USA: Min 63, Max 121, Avg 95.583 Usage 98-100%
Template: Min 115, Max 160, Avg 136 Usage 89-100%

Spoiler: Modded Maps

My personal settings, and what I'll be using for the next tests:

Brutal Slope: Min 80, Max 101, Avg 93.467
Crimson Valley: Min 72, Max 88, Avg 79.617
Endless Tunnel: Min 85, Max 115, Avg 96.229
Antigravity: Min 51, Max 138, Avg 106.267
Frozen Lake: Min 53, Max 73, Avg 61.533
Indianapolis: Min 26, Max 96, Avg 62
Lunar Surface: Min 67, Max 145, Avg 125.433
Dunsfold: Min 91, Max 103, Avg 96.85
VR City: Min 70, Max 124, Avg 102.75
Ultimate Test Terrain: Min 54, Max 99, Avg 73
Crash Hard: Min 46, Max 94, Avg 77.733
Mt Shell: Min 48, Max 72, Avg 58.183
Ghost's River: Min 16, Max 57, Avg 26.75
SSRB: Min 85, Max 122, Avg 92.583
Wasteland: Min 30, Max 81, Avg 55.367
Lime Rock Park: Min 54, Max 81, Avg 70
Beam Mountains: Min 32, Max 68, Avg 54.05
Crash Mountain: Min 40, Max 77, Avg 56.167
Desert Highway: Min 31, Max 61, Avg 51.867 (Forgot to turn off Vsync)
Rally Forest: Min 41, Max 60, Avg 55.067 (Forgot to turn off Vsync)
Tennessee: Min 0, Max 26, Avg 13.783 (Forgot to turn off Vsync)
Bird Rock Quarry: Min 24, Max 59, Avg 41.567 (Forgot to turn off Vsync)
Rolling Hills Rally: Min 44, Max 61, Avg 56.533 (Forgot to turn off Vsync)
Trial Mountain: Min 52, Max 86, Avg 69.667
Fir Rock: Min 51, Max 79, Avg 67.1

I will keep adding more

 

I did Advanced Island first as it was easiest, lazy me

Each of those drops is screenshot and screenshots are in numbered order (except in post they seemed to become bit random order, there is not even way to tell which is which, so compare FPS or something if want to figure out graph position for individual screenshot), now that is quite a lot of screenshots, might be better take video and do the math for fps difference for other ones or just take few shots


Ton more pics

Spoiler

--- Post updated ---

I spotted you have borderless checked, I have seen huge differences in CPU load between maps when running windowed or full screen for some maps windowed is heavier, for some full screen is heavier, it is bit odd thing.

RX460 is clearly not slow in BeamNG, no matter what benchmarks would claim. I'll add your results to first post, somehow in somewhat useful way.

 

Lol my laptop runs the game on 25-40 FPS with all shadows disabled and the quailty on low xD , damn

 

Laptops are bit challenging, BeamNG really seems to like single core computing power and vanilla maps @1080p need some GPU performance too.

Video came bit choppy, as when recording video, it is very important that fps stays exactly 60, however this shows how FPS changes trough Hirochi raceway, FPS is bit slower than when not recording video, I guess mostly around 5fps difference, but I did not watch whole clip, there might be bigger differences at places.


Might do tomorrow videos from some other tracks.

 

I think the problem with this test for BeamNG specifically is you are not taking into account how the game runs with many vehicles on screen at a time, since that is the main performance difference between BeamNG and other games. About 9 months ago I made this video, while limited to 60fps you can see how the game performs spawning more and more cars.
This was a while ago but the game still performs basically the same, maybe a little smoother now.

 

That is what Banana bench is for, I believe. Also from single vehicle test you can then start to calculate multiple vehicles quite well with most CPU's.

Most important thing is to test with exact same settings to have some form of standard measurement which would help to see what kind of GPU and what kind of CPU would be worthwhile upgrade to current system.

Surely I can show video like this, but it does not contribute much, it just shows that 4 cores with HT can run many vehicles:


Problem is figuring out what part GPU and what part CPU is needed to run most of the maps, running many vehicles is much easier to figure out after that.

For example @BombBoy4's i3 system can run single vehicle just as well as my i7 system, only if you want to put that many trailers behind a truck you might want to consider more expensive CPU.

Problem is, you have to know what you want to find out, different kind of testing is for different purpose, there is point with multiple vehicle testing too, especially boosting self esteem of those having more cores, but for most of the time, at least I find using no more than 2-3 vehicles at maximum and more than 50% of time 1 vehicle.

What BeamNG is different at is that you really have to understand some of how it works and way you use it in order to understand which hardware upgrade or detail setting is giving you best performance, many have no idea. Generalizations don't really fit in at all, so best is to have as many as possible different systems tested with same methods and settings to have better understanding.

That said, I have had though about some heavier vehicles, like T75 with trailer to be added, but it might be better to be own separate testing as it is more test of CPU than GPU.

GPU load in this test gives indication of how well CPU single core performance copes with what GPU needs, so that is helpful too. Many don't realize CPU limit, they up the resolution and don't see difference in FPS, which misleads quite quickly to what is really happeing.

Things get more challenging at low budget level, where it is not to be excepted to be running many vehicles, but to get BeamNG to run well at lowest possible investment. Everyone knows that 1000 dollars buys CPU and GPU that will run BeamNG just fine, with multiple vehicles, but what about 100 or 200 dollars upgrades to current system? So testing is needed of many various systems to find out that.

Also 0.9 optimizations will be interesting to see, how it affects systems, using multiple vehicles there is not too big performance issues when one has enough cores, but how about many maps that are quite problem with single or multiple vehicles, will there be change?

Perhaps this helps to understand better what purpose this kind of testing can serve vs the typical I can run tons of vehicles, which of course gives bragging rights, but I would say for larger part of BeamNG users is not helpful, only to video makers and other smaller group of users who runs a lot of vehicles.

 

Tennessee is kinda hard even for my system, video recording took out surprisingly lot of high end FPS too. Pressing F11 and changing visual distance from over 30 000 to 2000 helps a lot in it, but for testing purposes I left everything as they come.


GPU usage is 64% in this picture, for anyone not familiar with MSI afterburner, when ever that percentage is below 99-100% GPU is waiting CPU and that is how you can tell when you are CPU limited. CPU limit comes in two forms, limited by physics engine and limited by GFX processing. At Tennessee video limit is by CPU GFX processing at the beginning and at the end of video.

 

Yep, I'm so crazy that I 24/7 using planes on beamng x)
Extreme landings and bad weather and low vis which makes everything awesome because I love challenging