BeamNG on new AMD CPU's, will Ryzen run the game well enough ?

I'm Getting my R7 1700 tomorrow I'll run some tests once i've got it running and post the results.

 

I have an R7 1700 sitting on my desk. Still waiting for NewEgg to get the MSI X370 Gaming Pro Carbon motherboards in. I'll definitely be conducting a few tests of my own.

 

Let me know the results

 

Good, beacause I was mostly thinking about the 1700 and 1700X, because those will be the mainstream ones for sure.

 

I'm disappointed by people's attitude towards this launch. We all know that Intel has had AMD by the balls. Not anymore.

This is a huge win for the PC community. We need for AMD to do well this year, not to exploit their great effort to get lower priced Intel chips.

I understand that people utlimately want the best performance for the price they pay.

The thing is... AMD did just that. They compete on IPC, but most importantly, they bring the most future potential. Just like the 2 vs 4 core choice years ago. A lot of people went to get the 2 core because it did better in games. Well, didn't take long before 2 cores weren't enough.

We already see 4 core CPUs being squeezed today. You get smoother gameplay from Ryzen. Less instability, higher minimum FPS, albeit lower average FPS. I know what I'd take. Ryzen any day.

They did it this launch, they really did. And when the time comes and they release R5 and R3, those CPUs have been polished, both on a software level and hardware level. The price, just as before, will be hard to beat.

AMD has a great platform this time. Not only the CPUs, but motherboards as well. Better price, more features, overclocking, better compatibility, no chipset lock-in nonsense. Go where the future points. The future, like always (look in the past), go towards multithreading. Next gen APIs are made for exactly that, whether it's Vulkan or DX12. And remember, you don't want DX12 to win either. Root for Vulkan. DX12 is MS's leverage to push you towards their new OSs and throw you out of old ones. They have the power here, and they will always give themselves the advantage. You can forget Windows 7, Linux, OS-X with DX12.

Bottom line, whether it's gaming or computing, Ryzen has an edge here. I'd get their 8 core any day.

 

This is getting interesting when 2 guys can share Performance Infos about that.
My Xeon1231v3 is likely good, but not enough for this game^^

I have the chance to test 1700x in 2 months, but nice to see where it will end up.

 

Don't worry, I run the game with Core i5-2400 without any significant issues.

 

With BeamNG we don't know what kind of performance is needed to run it 12 months from now, but we know current high end offerings from any manufacturer probably will be quite bit less money after 12 months, especially if you are willing to buy used.

Any high end of today will run BeamNG quite well in 12 months time too and who knows what they will bring within 12 months, that might give better performance / money spent ratio.

This new AMD release is a big thing, but certainly there is premium to pay at first, there is going to be always new things of course which will always need a premium, but what I have found out is that waiting product to mature a bit is good, then you have more information that helps make right decision and less premium to pay for.

Being at edge of development and cool factor is then another value, if that is important, then of course getting new right after release makes sense, also it allows to be first if that is important, for younger ones that usually is, so again one has to weight in what one wants.

I could get for example 7700K now to replace i7-6700 and get good boost on lowest fps. but as there are new promising products coming that might give lot more improvement for same or less amount of money and that 7700K price is going to fall, it is wiser to look what happens on market and what happens with next BeamNG update that has some optimizations.

Lowest fps with 1800X should be around same as with i7-6700 if one has enough GPU power for given resolution, of course with overclocking 1800X one can get improvements, exactly how much will be seen with time.

Autumn's Intel release will probably lower Ryzen prices, so if feeling bad about Ryzen lowering Intel prices, it probably works other way around too, but if one has to think about budget, waiting for new to become matured is best bet

I hope that Banana bench works again in 0.9, so we can get some nice somewhat standard comparable data.

 

So, what's with them bananabench results, Robin?

 

He wrote that he is waiting for motherboard and bananabench does not work in this version of BeamNG.

 

So I couldn't get banana bench to run so I improvised manged 9 D15s before I saw below 35ish at 8 is was at around 40-45.
This is a R7 1700 at 3.7ghz on all cores with a reference Rx 480 and 16Gb of DDR4 2667mhz ram.
If anyone could instruct me on how to get banana bench running I'd be happy to run it

 

I guess you have to wait few weeks to 0.9 to come out to get banana bench working, but I would be interested too how to get it to run in 0.8.


Some testing ideas:
Have MSI afterburner or something else that can show GPU usage and make sure you haven't near 100% GPU usage (have it below 90% if possible) if you want to test CPU, otherwise you are testing more of your GPU than anything else. Of course, that will tell how well CPU can deliver stuff to GPU, but usually details and resolution is set so that GPU is not reaching limit when testing CPU.
SSAO for example adds mostly GPU load, as well as other post processing, which you might like to keep off when testing CPU.

Take off road roamer, add extra spotlamps to it, so you will have 6 lamps, turn on high beams and add another roamer and another, see how much it can handle of that, probably one of the heaviest CPU stress tests currently possible.

 

So I tried to remove as much of a GPU limitation as I could which got me to 12 D15's before it dipped below 50FPS
With one D15 I was getting about 400FPS

Any further testing I'm happy to carryout

 

I wonder how many trailers one could pull, for me 5 is maximum really, with more trailers fps dies quickly. Crashing with many trailers makes quite compact ETK cars, that should be where Ryzen is at strongest I believe.


edit: Post settings too, so people who think about upgrade can try to compare for their own system.

 

13 Makes it drop to 29-30FPS
One thing I noticed is the framerate was Very stable only moving 1-4 fps either way while playing


Im at work for the next 10 hours so I can't test for a bit :/
--- Post updated ---

I've run 2 cement mixer trucks fine 75fps locked with vsync
I have tried with 4 bluebird rear engine buses and that was fine 75FPS as well

I was on an fx6300 at 4.7ghz and i could spawn 1 fine but 2 tanked BAD



PC specs for those interested
Ryzen R7 1700 3.7Ghz at 1.297 volts
16GB G.Skill Trident Z series DDR4 at 2667mhz
RX 480 8GB reference card with 60mv under volt
120GB Kingston SSD ( really not fast )
3TB seagate baracuda
Corsair VS 650
H80i V1 Water Cooler
Gigabyte GA-AB350 Gaming 3 motherboard
Corsair Spec 03 Case
All run at 2560x1080 At max ingame settings unless trying to see how many things I can spawn

 

If you can, test out the differences with the Shadows, disabling and enabling them will provide significant performance difference.

 

Hiya people, in few days i'm gonna get Ryzen 1800X, anyone care for comparison with old Fx-6300? I sure hope you do :-D

Here are current results. 6 core FX-6300@4,3Ghz with 8 GB 1866 MHz RAM (and RX480). Testing was on gridmap, 1024x768 (to negate graphic bottleneck). Adding multiple D15s or T65s, camera set to orbit, view from behind last spawned vehicle.

 

I made little video that should help to see how one can test single core performance without getting GPU limits, at least if one has 1050Ti or faster, doing those settings, that car config and location will max out even the fastest CPU if not with 1050Ti, then anything even slightly faster should be enough.


That test uses shadows to load single core of CPU, Ryzen 1800X should be about same performance according to Cinebench, so what naturally interests is if Cinebench is comparable to BeamNG performance in this.

Turning shadows off then shows both CPU and GPU maxed out, which I guess would be ideal situation and preferably with graphics settings maxed out, not sure if single Titan would be enough with Ryzen or any other decent level CPU to get that situation. Also different cars, maps etc. change how much CPU or GPU is used, it takes bit of time to get really familiar with how settings etc. affect.