BANANA Benchmark Results Comparison

I take it that the number of CPU cores is only beneficial if there is more traffic? 1 flow = one car? Or how does it work? Personally I noticed that disabling hypertrading on my 13600kf has a positive effect on FPS, but in banana test the result is 15-20% worse

 

5900X, Standard everything, Win10

5900X, Standard everything, HT disabled, Win10

5900X, Some bios/ryzen master tinkering, 4.5ghz, HT enabled, Win10


5600H, Lenovo legion laptop, Out-of-box settings, held 3.9ghz under full load, Win11


5900X, Standard settings i guess (btop says 4.2ghz?), Debian12, BeamNG trough Proton, up to 65% overall cpu load


5900X, Same as above, Linux Native BeamNG, ingame Bananabench does not start. I guess its bugged anyway, i dont see that huge of a performance difference ingame, load was up to 85%


Yeah, some funny business going on here

 

9900x first run, havent played around with OCing and undervolting yet.

 

old ass i7

 

I have access to a Ryzen 7 laptop, so here's how the Ryzen 7 8845HS runs BeamNG

 

9950X3D, unoptimised, PBO + EXPO

841.160 Mbeams/s

Benchmark seems wonky, what's up with these results?

Code:

][  ____                       _   _  _____      _      _
][ |  _ \                     | \ | |/ ____|    | |    (_)
][ | |_) | ___  __ _ _ __ ___ |  \| | |  __   __| |_ __ ___   _____
][ |  _ < / _ \/ _` | '_ ` _ \| . ` | | |_ | / _` | '__| \ \ / / _ \
][ | |_) |  __/ (_| | | | | | | |\  | |__| || (_| | |  | |\ V /  __/
][ |____/ \___|\__,_|_| |_| |_|_| \_|\_____(_)__,_|_|  |_| \_/ \___|
][
][ ----------------------------------------------------------------------------
][ BeamNG.drive 0.34.2.0 (64 Bit) built Sat Nov 30 16:06:24 2024
][ Microsoft Windows 11 (build 26100) | RAM: 63.58 GiB | Toolset: 19.29.30154
][ AMD Ryzen 9 9950X3D 16-Core Processor | 16 physical cores, 32 logical cores | 4292 MHz
][ current time: Sunday Sun Mar 16 19:21:58 2025
][ console documentation: https://go.beamng.com/consoleDocs
   +----------+----------+
   | MBeams/s | % Realt  |
---+----------+----------+
 1 |  148.249 |  1528.02 |
 2 |  256.158 |  1320.13 |
 3 |  289.249 |   993.78 |
 4 |  358.477 |   923.72 |
 5 |  433.659 |   893.96 |
 6 |  505.839 |   868.96 |
 7 |  568.356 |   836.88 |
 8 |  600.817 |   774.09 |
 9 |  598.684 |   685.64 |
10 |  602.564 |   621.07 |
11 |  629.970 |   590.29 |
12 |  634.567 |   545.05 |
13 |  669.551 |   530.86 |
14 |  681.341 |   501.62 |
15 |  677.729 |   465.70 |
16 |  701.303 |   451.78 |
17 |  724.602 |   439.33 |
18 |  741.825 |   424.78 |
19 |  767.020 |   416.09 |
20 |  778.996 |   401.46 |
21 |  809.676 |   397.40 |
22 |  822.742 |   385.46 |
23 |  807.447 |   361.85 |
24 |  802.620 |   344.70 |
25 |  808.026 |   333.14 |
26 |  813.228 |   322.39 |
27 |  819.944 |   313.01 |
28 |  828.568 |   305.01 |
29 |  820.918 |   291.77 |
30 |  841.160 |   289.00 |
31 |  767.700 |   255.25 |
32 |  751.401 |   242.03 |
33 |  727.630 |   227.27 |
34 |  733.875 |   222.48 |
35 |  760.471 |   223.95 |
36 |  766.122 |   219.35 |
37 |  779.636 |   217.18 |
38 |  787.046 |   213.48 |
39 |  797.919 |   210.88 |
40 |  805.085 |   207.45 |
Max Mbeams/s:   841.160 Mbeams/s

 BANANAA!!!
 .______,#
 \ -----'/
  `-----'

EDIT:

Results with defaultVehicleCount = 150

Max Mbeams/s: 922.792 Mbeams/s
90 realtime

Code:

][  ____                       _   _  _____      _      _
][ |  _ \                     | \ | |/ ____|    | |    (_)
][ | |_) | ___  __ _ _ __ ___ |  \| | |  __   __| |_ __ ___   _____
][ |  _ < / _ \/ _` | '_ ` _ \| . ` | | |_ | / _` | '__| \ \ / / _ \
][ | |_) |  __/ (_| | | | | | | |\  | |__| || (_| | |  | |\ V /  __/
][ |____/ \___|\__,_|_| |_| |_|_| \_|\_____(_)__,_|_|  |_| \_/ \___|
][
][ ----------------------------------------------------------------------------
][ BeamNG.drive 0.34.2.0 (64 Bit) built Sat Nov 30 16:06:24 2024
][ Microsoft Windows 11 (build 26100) | RAM: 63.58 GiB | Toolset: 19.29.30154
][ AMD Ryzen 9 9950X3D 16-Core Processor | 16 physical cores, 32 logical cores | 4292 MHz
][ current time: Sunday Sun Mar 16 19:33:36 2025
][ console documentation: https://go.beamng.com/consoleDocs
   +----------+----------+
   | MBeams/s | % Realt  |
---+----------+----------+
 1 |  147.474 |  1520.03 |
 2 |  240.381 |  1238.82 |
 3 |  299.870 |  1030.27 |
 4 |  364.778 |   939.96 |
 5 |  438.933 |   904.83 |
 6 |  500.120 |   859.14 |
 7 |  571.898 |   842.09 |
 8 |  600.752 |   774.01 |
 9 |  589.540 |   675.16 |
10 |  615.188 |   634.08 |
11 |  638.022 |   597.84 |
12 |  645.566 |   554.50 |
13 |  665.734 |   527.83 |
14 |  687.194 |   505.93 |
15 |  686.158 |   471.49 |
16 |  722.924 |   465.71 |
17 |  734.339 |   445.23 |
18 |  742.740 |   425.31 |
19 |  757.944 |   411.17 |
20 |  791.522 |   407.92 |
21 |  816.667 |   400.83 |
22 |  824.216 |   386.15 |
23 |  840.709 |   376.75 |
24 |  832.171 |   357.39 |
25 |  824.254 |   339.83 |
26 |  825.643 |   327.31 |
27 |  835.121 |   318.80 |
28 |  841.692 |   309.84 |
29 |  845.711 |   300.58 |
30 |  864.367 |   296.97 |
31 |  880.544 |   292.77 |
32 |  811.323 |   261.33 |
33 |  789.387 |   246.56 |
34 |  765.011 |   231.91 |
35 |  779.377 |   229.52 |
36 |  775.078 |   221.91 |
37 |  779.802 |   217.23 |
38 |  792.708 |   215.01 |
39 |  804.160 |   212.53 |
40 |  808.642 |   208.37 |
41 |  821.806 |   206.60 |
42 |  836.252 |   205.22 |
43 |  848.408 |   203.36 |
44 |  853.960 |   200.04 |
45 |  861.124 |   197.24 |
46 |  866.859 |   194.24 |
47 |  868.917 |   190.55 |
48 |  864.722 |   185.68 |
49 |  863.804 |   181.70 |
50 |  859.564 |   177.19 |
51 |  861.028 |   174.01 |
52 |  875.203 |   173.48 |
53 |  886.003 |   172.31 |
54 |  887.252 |   169.35 |
55 |  885.677 |   165.98 |
56 |  898.889 |   165.45 |
57 |  906.759 |   163.97 |
58 |  910.909 |   161.88 |
59 |  908.710 |   158.75 |
60 |  914.130 |   157.03 |
61 |  909.656 |   153.70 |
62 |  912.827 |   151.75 |
63 |  922.792 |   150.97 |
64 |  886.450 |   142.76 |
65 |  861.726 |   136.65 |
66 |  857.866 |   133.97 |
67 |  862.487 |   132.68 |
68 |  865.246 |   131.15 |
69 |  864.063 |   129.07 |
70 |  870.259 |   128.14 |
71 |  861.748 |   125.10 |
72 |  863.647 |   123.64 |
73 |  864.594 |   122.08 |
74 |  871.726 |   121.42 |
75 |  867.161 |   119.17 |
76 |  867.342 |   117.63 |
77 |  866.248 |   115.96 |
78 |  865.341 |   114.35 |
79 |  860.201 |   112.23 |
80 |  912.395 |   117.55 |
81 |  905.630 |   115.24 |
82 |  902.776 |   113.48 |
83 |  901.326 |   111.93 |
84 |  899.440 |   110.37 |
85 |  889.349 |   107.84 |
86 |  905.674 |   108.55 |
87 |  887.046 |   105.09 |
88 |  889.302 |   104.16 |
89 |  883.068 |   102.27 |
90 |  877.751 |   100.52 |
91 |  894.603 |   101.33 |
92 |  880.062 |    98.60 |
93 |  876.751 |    97.17 |
94 |  867.606 |    95.13 |
95 |  872.677 |    94.68 |
96 |  867.965 |    93.19 |
97 |  870.959 |    92.55 |
98 |  871.549 |    91.67 |
99 |  877.063 |    91.31 |
100 |  878.747 |    90.57 |
101 |  880.162 |    89.82 |
102 |  873.841 |    88.30 |
103 |  882.329 |    88.29 |
104 |  880.770 |    87.29 |
105 |  879.981 |    86.38 |
106 |  886.677 |    86.22 |
107 |  888.123 |    85.55 |
108 |  893.550 |    85.28 |
109 |  889.445 |    84.11 |
110 |  893.641 |    83.74 |
111 |  893.311 |    82.95 |
112 |  894.917 |    82.36 |
113 |  886.146 |    80.83 |
114 |  889.880 |    80.46 |
115 |  884.914 |    79.31 |
116 |  882.721 |    78.43 |
117 |  881.286 |    77.64 |
118 |  875.743 |    76.50 |
119 |  880.342 |    76.25 |
120 |  877.345 |    75.36 |
121 |  887.349 |    75.59 |
122 |  898.692 |    75.93 |
123 |  875.341 |    73.35 |
124 |  870.985 |    72.40 |
125 |  865.523 |    71.37 |
126 |  864.792 |    70.74 |
127 |  856.329 |    69.50 |
128 |  851.920 |    68.60 |
129 |  838.665 |    67.01 |
130 |  832.259 |    65.99 |
131 |  811.779 |    63.87 |
132 |  836.160 |    65.29 |
133 |  822.333 |    63.73 |
134 |  823.585 |    63.35 |
135 |  827.033 |    63.14 |
136 |  821.588 |    62.27 |
137 |  827.807 |    62.28 |
138 |  826.610 |    61.74 |
139 |  821.350 |    60.90 |
140 |  829.361 |    61.06 |
141 |  825.393 |    60.34 |
142 |  823.083 |    59.74 |
143 |  818.185 |    58.97 |
144 |  824.656 |    59.03 |
145 |  813.031 |    57.79 |
146 |  814.913 |    57.53 |
147 |  809.102 |    56.73 |
148 |  808.080 |    56.28 |
149 |  802.875 |    55.54 |
150 |  807.352 |    55.48 |
Max Mbeams/s:   922.792 Mbeams/s

 BANANAA!!!
 .______,#
 \ -----'/
  `-----'
Press any key to continue . . .

 

Intel Core i7-3537U, 2.89 GHz throughout the test.
Compared to the last time I benched this cpu I have observed a 28.8% increase in maximum mbeam/s (from 54,096 to 69.678) and a 50% increase in the number of cars above 100% realtime (from 4 to 6), so good job devs in optimizing the engine.

 

I come here every few years when i get an upgrade, here are the results for my new 9800X3D.

 

I got 580 with my 7800X3D so while the 9800x3d is faster than the 7800x3d in lots of other games, it doesn't seem like a big increase in beam.

 

May I add the worst result of 3 runs using
"gamemoderun ./banana-bench.x64.sh"
for my casually-tuned r7 7700 (( using curve-optimizer set to minus_39, factor_10; capped to a 105W max_PPT limit))
lightly-tuned Hynix 2x16GiByte single-rank "A-die" RAM running @6400MT/S (UCLK=MCLK, FCLK=2133MHz)
System in use: current Tuxedo-OS (Ubuntu-LTS 24.04.1 derivative with customized KDE-Plasma) with a "Xanmod.org" mod-Kernel: "6.14.4-x64v3-xanmod1"

best run (out of 3) to my untrained eyes seems to be the third run with

So when comparing to the results of the last 9800x3d entry above, should I rejoice that I did not pay twice as much for the latest arch and some chunky extra bolt-on cache or would I be best-served to humbly accept that I ran the wrong test somehow? And yes, I do still consider myself lucky having scored a CPU that seems happy at those settings (built this new-to-me platform less than a month ago) and I am unsure if this is easily achieved by any and all 7700 non-X copies out there. 5260MHz seems in-line with my cpu_multicore Cinebench R24 runs in Windows using hwinfo64 to monitor clocks and temps. Only prime95 and similarly taxing loads would make it clock below ~4.8 GHz for the majority of the stress-runs.

Thanks for making this game run natively on Linux, by the way!

 

Core Ultra 5 225 10 Cores, 6P 4E, doesn't score that great, but the damn thing handles multiple AI cars better than my 7800X3D system

 

Something very odd with Arrow Lake and beam Banana Bench.

Much worse scores than my 14900K system (which was around 980M/beams



Only 658, however this arrow lake system will easily do 60 AI cars spawned without really breaking a sweat, quite interesting

 

I'm sorry I have to ask this: how do you have access to such a variety of modern components?

 

It's a hobby, I go to the store (or order online) and then build a system, test it out in Beam.....then usually sell it or keep it.

 

R9 7945HX3D 16 Core 32 Thread with 3D Cache

Not the best results



Also had some time to investigate why I was getting such poor results in Banana (I know it really doesn't matter) despite having the best performance I've ever had in Beam, with my arrow lake cpu's more specifically the 285K

From my previous post.....I got 658m/beams from a 285K

So I pulled up HW monitor to see what was going on.



When I ran the benchmark only the 16 E-cores were being used and the P-Cores were basically sitting idle.

When I ran the benchmark again (but from inside the game using Vulkan) the results were better (but still less than my 14900K



In the end it really doesn't matter to me because it's the real in game performance that matters, and arrow lake is absolutely killing it right now.

 

This is a problem with 3D Cache CPUs. It has been investigated by an user I know on Discord; you have to first bump the number of cars up on bananabench to like 50/55 in your case, so that we know the number of cars (n) when the % realtime drops below 100%, then modify the code of bananabench to just single shot those n cars, without the buildup from 1 to n. This will give you a far better result, but I cannot really count it for the spreadsheed since it's a heavily modified benchmark scenario...
If you want, do it just for testing and post the results.