BANANA Benchmark Results Comparison

I'm currently looking to buy either a 7800X3D or a 13700k. I'm also considering a 7950X if it provides a meaningful uplift over the 7800X3D in Beamng. To those of you who have these CPUs - how are your scores?

 

I think 13700k will do 25-30 % on benchmark, than 7800X3D, but single core (also gaming perf), the 7800X3D will do BETTER
The photo that I send is the equation that can encounting the real performance of the CPU (more accurately, the traffic holding multiply/number of real core. One E-core (Intel) will be half core (2 E-core= 1 normal core) and I counted that the single core also affected the real result and performance. It will be not the most similar, but I hope it can help u to compare the banana benchmark results
In this equation, we have
M for multi-core (peak/max score you reached)
S for single-core (1-car score at the banana table)
B is the RAM bus (you need to divide your RAM bus into 100 in this equation)
Big C is the CAS latency
Small c is the cache amount (that count as total L1+L2+L3 cache)
Here the conclusion:
It quite similar as the real result, with tolerance counted as (core amount÷4), that it same as 85-95% of the real benchmark that I found on yt (the final result of the traffic, should be core amount× this equation result (now I call as H), it will be the recommended traffic that you should run for about 30FPS. With 60, divide it with 1,5. With simplified traffic mode, multiply it with 1,5 (AVG of simplified traffic select/normal traffic select). This result is only valid if you don't run with any hidden applications/ parking car inside the map, or the CPU got throttling when running the games

 

Hey, while i dont have any of these cpus as you know.

I seen a video with somebody on a 7800x3d. They apperently got 40 - 60 fps on 30 cars.

Another video of 7950x has similar fps but im guessing 7950x will do better when more cars are spawned in.

Anyways i think the cpu list goes like this

13900KS > 13900k > 7950x ≈ 7800x3d > 13700k(maybe) > etc

Not sure where 7950x3d goes on this list.

May be better than 13900k but nobody knows sadly. Apperently it may be worse than even 7800x3d due to memory controller.

Either way im not sure. 7950x3d is a huge gamble. Could be amazing, could suck.

 

• Where do those results come from?
• The unit of measurement of the formula does not make sense. For example, in B-C you are subtracting nanoseconds (C) from bits (B) which is like subtracting mass (kg) from speed (m/s). This entire formula is physically wrong (see below).
  
  Code:

  M [Mbeams/s]
  S [Mbeams/s]
  B [bit]
  C [ns]
  c [Byte]
  [Mbeams/s]+[Mbeams/s]+[bit]-[ns]
  --------------------------------
    [Mbeams/s]-[Mbeams/s]-[Byte]
  

• The final performance numbers are plain wrong from a test I just did.
  Normal traffic (no simplified collisions), no parked cars, cpu (i5-7600K) not throttling.
  
  Assuming the formula above made sense, I should run my car + 6.13 other normal traffic cars at 60 fps. That's not what happens as I'm in the 36-42 fps range rounding 6.13 to 6. If I rounded it to 7 (which would be the right rounding to do as it's a more conservative/cautious approach) I would barely reach 35 fps.

Please don't post random equations and percentages as people may actually buy a cpu based on those wrong assumptions.

 

So I think the thing
First, the silicon lottery is very random and, risky too. You can't expected how much performance that it give me
And I count it based on the standard benching score, and yea, it will be random, so random that made the result become very wrong
2nd, there a thing, you can have some bad luck on silicon lottery and here the table of luck:
<-3 than the result you have counted -> very bad luck at lottery
-1,5-(-3): you have bad luck
0-(-1,5): normal luck, it can count as standard
0,1->1: you have some luck
1->2: you have good luck
2->3: you have very good luck, and it is very rare
I don't mean this equation Tobe standard, but hope it will help you to encounting your LUCK at silicone lottery, not for counting how many car you can handle at 60 FPS/1080p that I said yesterday, it should only run at 30 FPS/1080p. Sorry for wrong purpose and information
Of course I will do more test about that, and I will modify tolerance for the silicone luck on your CPU
And if I counting that, I think.... abit wrong
Ex: my laptop with 10th gen i5 (U- series) give same result as 12100f (I think you need to divide 1,5 if u using laptop, cause it will give less raw perf than desktop, and it can even throttling when on mid- game)
Ex, my 1035g1 (my previous laptops CPU) give about 91,8 score at 7 car peak, and I think we need to divide it to 1,5 before apply result
And you should upgrade your CPU now, cuz I found the guy which said the i5 7th gen is weaker than i5-2500 (??)
And this equation should be meant to run the traffic at 30 FPS on 1080p (Important: the resolution also affect the FPS, so for 1440p and 2160p we need to minus the result with follow 10% and 20% than the real result at 1080p)
Edit: you have good luck at your CPU cuz I see your FPS are higher 10% than the equation result
And final word, it also counting of your silicon luck
Somehow when I found a video from the YT channel named Kiki zones is testing the 13900k, and it can run at 45-55 FPS/4K (with 3090: 45-50; with 4090: 50-55, so the traffic should be relied only 3-5% at graphic card)

 

My answer: Do not buy either 7900x3d and 7950x3d
Because both of this CPU don't have any hardware scheduling for the future proof
And don't buy 13900ks either cuz it make u spend 50-60$ more for 0,2 GHz
Only buy 13900k if you have good enough AIO or custom cooling, and good PSU too
With 7950x, it have same price as 13900k (My region), it should be abit better cuz it have 40-50% less power consumption (I see a Vietnamese guy test that and it consump 216W, while 13900k have consumption of 331W, both run at stock) and around 10% cooler (?) Than 13900k
I think go for 7800x3d is the best choice, cause 8 core now and 5 year later with the huge amount of cache is good, and also it consump 40-50% less than 13700k (13700k is consump around 250-260W and 7800x3d have consumption of 150-160W, I found on internet)

 

Laptop MSI Vector GP68HX 12VH + undervolt + overclock

12900hx @ 5.2ghz

 

Decided to update my results (i5-7600K not overclocked):

Frequency during the test


Result: 178,941 Mbeam/s

Did some maths and it's a 18.71% increase of Mbeam/s and a 16.67% increase of realtime vehicles (2 more but ~17% sounded cooler) from last time

 

It quite neat I think
But with the real test, is it correct anymore?

 

Anyway, I want to buy either 13600k and 7700x. I pair it with 4070. Which CPU is better?

 

If by "real test" you mean how many cars I can run in game, as I've already said before the Banana benchmark score does not correlate to the cars you can spawn in game. You can compare Banana benchmark results with other Banana benchmark results, but not benchmark results with the number of cars you can spawn.
Now for this:

There are those results available on the spreadsheet:

Up to you to draw your own conclusions, maybe the owners can chime in.

 

First: yeah you are correct. It depend on the CPU samples itself and also, luck
Second: in my region, The 13600k is 45-50$ CHEAPER than 7700X (I'm NOT RECOMMENDED the 14600k cuz it as expensive as 7700x with only 0,2 GHz clock boost, also 7700x need less cooling requirements, and need higher RAM bus)
Anyway, how to quote multiple times at the same message?
Of course, my guess about RAM bus at the 13600k rigs: 6400 C34 or C36. And 13600k CAN'T BE that high as over 730 Mbeams. Should be only 650 Mbeams or he just OC-ed it to 5.5GHz, my guess
So I think still prefer the 7700x cuz it still cheaper on cooling/maybe PSU/Mobo. Just respect my opinions haha
Edit: ah, the 13600k user use DDR4 3600 C18. Somehow Aider Lake got slightly more score at the DDR4 than DDR5

 

Weird how DDR4 beat out DDR5 here.

Wonder if i could just get 4000 or smth ddr4 and do better than DDR5? Maybe its just Silicon Lottery? Who knows...

 

win 11
12900kf
P cores 5.1
E cores 3.9
192w max power during test

 

Weird.

Seems like it does as good or even better than the 7950x

Would that mean a 12900K would do better than the 7950x in a real test?

Question can you spawn in like 40 cars and tell me what FPS you get? Thanks

 

i think this is not fair cuz i have less cores)
p.s. if i have some time in future, i can do 5.2/4.2 GHz test.. and i don't know potential of cpu without E cores)
Arctic Cooling Liquid Freezer II-420

 

Yeah, I also think about that
And.....I will buy the 13600k rigs if the price is under 200$

 

Edit: today I done a benchmark about my ex-brother laptop and here the spec and result:
Processor: i7-11800h
8GB RAM (he should upgrade it)
Card: 1650 Max-Q
7 AI vehicles as a test for the games
And here the result of both Banana and real gameplay (sorry he limited my using times so no photo here):
Banana: 437,239 (max, at 17 cars), 37 Cars realtime
Real test:
West coast (downtown, high settings): 25-45 FPS, 36 FPS AVG
West Coast (downtown, normal settings): 25-55 FPS, 42-43 FPS AVG
West coast (outside the downtown) and other maps: 28-76 FPS, avg 55-56 FPS
So here my conclusions:
For the real time result, I suggest that you should upgrade with more RAM (32GB is great spot) for better FPS and stability
Also, a graphic card that can keep up with the CPU is quite necessary here, especially more FPS and can play with better graphics (with traffics, I think it can increase performance up to 20% for each 50% graphic performance at normal conditions). Notice: don't upgrade to the too high-end graphic card than the CPU or it will create a massive bottleneck
Edit: the Intel CPU is sh*tty hot as 85° and over despite changed thermal paste 2 month ago, so just think twice before buying PC or laptop with Intel processor

 

Hi guys, just wanted to share my own score for the database, i was also wondering how well a 14700k (or equivalent 13700k etc) would perform if you ONLY use E-Cores for beamng. So like using a tool like process lasso and disabling the games access to P-Cores. I think e cores could outperform my 5930k here while using a fraction of the power. I just wanted you guys opinion on it cause i found a good cheap z690 motherboard and i already got 64gb of ddr4 so it seemed like a good upgrade to this aging system

 

First, E-core have ONLY IPC is equal as skylake
So about 13700k with ONLY E-core, it should be equal as the 9700k (Nearly or lower abit)
=> So 8 E-core at 13700k will have at least 5-10% more score than your 5930k
Also 14700k, with 12 E-core, will have 30-40% more score than 5930k, if you only use E-core
And I ask you 2 things:
1: How much RAM Bus do your RAM have? (DDR4>DDR5 due to lower latency, especially 3600 with Latency of under C20)
2: Which graphic card will you pair?
So...I ask the question too: What is the best Mobo now to pair with 13600k? (Abit overclock, MATX size)