BANANA Benchmark Results Comparison

Msi Stealth 15m
i9 13900H 6P 8 E Cores


--- Post updated ---
ROG Ally AMD Z1 Extreme (basically a 7940HS)



Interesting its almost as fast as a 5800X3d posted earlier in this thread despite being heavily TDP limited, 3Dcache isn't doing much here

 

Here is my new 7800X3D system, 100% stock DDR5 running at 4800 until I get everything fully tweaked

 

Wtf happened between 25 and 36 vehicles on the i9?

 

Good question, no idea, I had tweaked some TDP limits and this was my highest score on the laptop to date....but something went haywire for sure

 

I found a youtube video where a guy with 14900k tests cars in increments up to 40.

I did the same test but at 1440p since otherwise i was always GPU limited because of 4080.


Results:
0 cars - GPU limit, irrelevant
10 cars - 7800X3D wins by 30-60 fps
20 cars - 7800X3D wins by ~20 fps
30 cars - draw, similar fps
40 cars - again very similar fps

So in conclusion the 3D cache makes a massive difference ingame (700Mhz lower frequency but higher fps) and fps is very different from banana benchmark results.
But maybe the resolution had an effect, would be great if someone with a 14900k could repeat this test at 1440p, ideally with a 4080 and same 10 car steps.
I did have ~10 fps less in the east coast 40 car scenario you posted. 12700k - 30 fps, 14900k - 50fps, 7800X3D - 40fps. Despite the fact 7800X3D had same benchmark score as 12700k it got higher fps which still proves benchmark isn't accurate.

 

I don't know what the hell is going on, everytime I post, then try to add a reply it just edits my last post....so I lost my entire reply............anyways I have tested out almost every single combination between my 2 systems and recorded and posted them all on Youtube, I'm not sure how long I have to wait between replies so that it doesn't keep editing the existing post, but when that time expires i'll post all the videos

 

You could try to use a Upscale to mimic playing at 4k and you can get a similar Benchmark going

Also i do honestly think the fps your getting is due to 3d since somebody tested a 7900x3d and reported 35 fps with 40 cars which is honestly surprising since it has more cores, guess less Cache isn't helping it.

So i think your benchmark is correct here

 

It would be logical to use the same location and 1080p! And also the traffic in view of the same cars if it is possible

 

I just upgraded to a ryzen 7800X3D, it seems fine and I get this 560 score quite consistently. I've so far only tested traffic with the automatic amount setting, and I didn't manage to count how many there were

edit: bananabench doesn't know I have win11?

 

The build number is what matters: if it starts with a 2 it's Windows 11.

 

It doesn't get any more closed and consistent testing than this. I'm only going to put a few videos to I don't get this thread too far off topic.

14900K West Coast Map 4K Ultra RTX 4090



7800X3D West Coast Map 4K Ultra RTX 4090



In this case the 7800X3D is 10-15 FPS faster

Now switch to a 3090 for both systems

14900K



7800X3D



You are now GPU limited so there is no difference. Unless you have a very high end GPU you aren't going to notice much difference if any at all, with low numbers of cars and on maps that are GPU limited.

Moving on to the automation map. Both systems running RTX 4090

14900K



7800X3D



14900K in this case is 10-15 FPS faster

Go back to a RTX 3090

14900K



7800X3D



Now GPU limited again.

I'm not running tests at 1080p because no one spending good money on a graphics card is going to run at 1080p in this game.

So depending on the map, the CPU's are trading blows with which one is faster or not. My best post I'll start adding traffic, and that's where things start to change .

 

If you are on a map that is CPU limited, run vulkan, you might be surprised.

Example, loading up City of Los injurus to the Beverely Hill starting point, its very CPU heavy.

look what is getting 30fps greater? Not the 7800X3D

 

Hmm now test vulkan which is faster w cars

 

I have......check out my channel i've done tons of comparisons,



Oh and what's super fun, I just built an i7 4790K system......barely playable with a 4090 using DX11.......but in vulkan....totally playable

 

60 Cars... My god, my 5900x melts even at like what, 16 cars? not to mention the hard to run map. Now a test with the 7800x3d AND possibly a 5800x3d in vulkan could settle, does 14900k ALWAYS win vs the 7800x3d in beamng or the other way around?
Edit: nevermind ima just check your channel lmao

gosh BeamNG sure is weird... So 7800x3d does a little better in Vulkan vs the 14900K but 14900K destroys the 7800x3d while not in vulkan... But at the same time looses while in Vulkan itself? Jeez this game sure is something sometimes. But in the end it seems like the 14900K MAY still have the edge over the 7800x3d? Atleast in west coast))

 

The game IS INSANE. Some maps 7800x3d is better, others the 14900K is.....however the big thing is most people would not notice these differences unless they have a top end GPU like a 4090, You can also check out when I tested both the 7800X3D and 14900K with an RTX 3090, now admittedly I didn't do a lot of Vulkan testing with the 3090 (which I have now sold) but yeah weird game, I love tons of AI so I use the 14900K system more. I might grab a 7950X3D to see how that compares to the 14900K since it has 8 more cores vs the 14900k but I do think (well I know) the E cores really do help in Beam

 

yeah i wish the game was more reliable, so much stuff like 5800x3d does it do better than 5900x or 5950x? How about 7700x vs 5900x? All these types of CPUs and beamNG just feels like a complete dice roll to see whats best with no real way to see if X is better than Y, used to be more cores = better but with X3D now its all over the place

 

Optimized PBO Ryzen 9 3900X

 

My new PC build

i7-14700K, no OC
Peerless Assassin 120 air cooler
32GB DDR5 6000MT/s CL30
Windows 11

 

Hello again,

I'm back with my 14700K and now 100% definitive proof that something weird is going on with the windows scheduling when it's about banana bench. I recently had a usb drive with a live system of kali linux i was using to pentest some wps networks when I had the idea: "Hey! I remember Beamng having a native linux version how about I benchmark Beamng in linux?"

So I unplugged the usb from my laptop, plopped it into my pc and booted into kali linux. Everything worked seamlessly so I went to the Beamng folder on my main ssd and ran the Bananabench benchmark and BOOM! A HUGE SCORE OF: 1120MBeams/s

That is crazy high and you can see the score rising almost linearly with the amount of cores used. Of course the single car/core score is only 60mbeams/s because linux scheduled that to an ecore as I saw in the top right cpu usage meter thingy but I certainly wasn't expecting this.

I don't know much about linux so to just show you that i was actually using a 14700k i just did lscpu because i don't know about a hwinfo alternative but the score itself isn't actually the main topic just the scheduling that windows messes up for bananabench so that's my quick post.
bye