BANANA Benchmark Results Comparison

Do you have an industrial reflow oven? If yes, then sure with a steady hand and a hot air gun you can remove the old CPU, clean up all the old left over solder, repaste it and then reflow the new CPU on. Otherwise no. Its a BGA package soldered directly to the motherboard, not removable.

Those arent pins that clip into a socket like on a normal CPU, those are solder pads, slightly rounded ones actually hence why its a ball-grid-array package.

 

I tried this on my Gigabyte BRIX equipped with the cheapest AMD APU, running in WINE 1.7.22 (POL) on Ubuntu 14.04:

Code:

][ v4 | 0.1.1.0 (32 Bit) built Tue Jul  1 18:55:18 2014 | now: Sunday 07/20/14 01:21:08][ Microsoft Windows XP Professional (v5.1) Service Pack 3 (build 2600) | RAM: 3.01 GB
][ AMD A8-5545M APU with Radeon(tm) HD Graphics | 4 cores | 1700 MHz
fixme:ver:GetCurrentPackageId (0x33f874 (nil)): stub
Beam ANAlysis NArrator, Version: 0.2
Test vehicle:  vehicles/pickup
 1 vehicle   :   2.765 Mbeams/s,  38.75 % realtime
 2 vehicles  :   5.515 Mbeams/s,  38.65 % realtime
 3 vehicles  :   6.570 Mbeams/s,  30.70 % realtime
 4 vehicles  :   7.204 Mbeams/s,  25.25 % realtime
 5 vehicles  :   6.924 Mbeams/s,  19.41 % realtime
 6 vehicles  :   7.814 Mbeams/s,  18.25 % realtime
 7 vehicles  :   7.898 Mbeams/s,  15.82 % realtime
 8 vehicles  :   7.829 Mbeams/s,  13.72 % realtime
 9 vehicles  :   7.402 Mbeams/s,  11.53 % realtime
10 vehicles  :   7.363 Mbeams/s,  10.32 % realtime
Max Mbeams/s:   7.898 Mbeams/s


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

Not that impressive unless you consider I only built it to play HD movies and casual games. My gaming rig however will get benchmarked some other time.

Edit:
Who am I kidding, I'm too curious to wait:

Code:

[v3|0.1.1.0| Intel(R) Core(TM) i7-2600 CPU @ 3.40GHz (4 cores, 3490 MHz) | Saturday 07/19/14 23:36:47]
Beam ANAlysis NArrator, Version: 0.2
Test vehicle:  vehicles/pickup
 1 vehicle   :  12.498 Mbeams/s, 185.27 % realtime
 2 vehicles  :  20.970 Mbeams/s, 155.42 % realtime
 3 vehicles  :  24.352 Mbeams/s, 120.33 % realtime
 4 vehicles  :  23.440 Mbeams/s,  86.87 % realtime
 5 vehicles  :  22.857 Mbeams/s,  67.76 % realtime
 6 vehicles  :  26.411 Mbeams/s,  65.25 % realtime
 7 vehicles  :  27.281 Mbeams/s,  57.77 % realtime
 8 vehicles  :  25.586 Mbeams/s,  47.41 % realtime
 9 vehicles  :  24.935 Mbeams/s,  41.07 % realtime
10 vehicles  :  26.396 Mbeams/s,  39.13 % realtime
Max Mbeams/s:   27.281 Mbeams/s


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

This is only with 4 cores enabled in Xen though, I'll try with 6 cores soon.

Edit 2:
Here we go, 6 cores:

Code:

[v3|0.1.1.0| Intel(R) Core(TM) i7-2600 CPU @ 3.40GHz (6 cores, 3397 MHz) | Sunday 07/20/14 00:04:27]
Beam ANAlysis NArrator, Version: 0.2
Test vehicle:  vehicles/pickup
 1 vehicle   :  11.636 Mbeams/s, 172.48 % realtime
 2 vehicles  :  17.316 Mbeams/s, 128.34 % realtime
 3 vehicles  :  25.768 Mbeams/s, 127.32 % realtime
 4 vehicles  :  33.305 Mbeams/s, 123.43 % realtime
 5 vehicles  :  36.460 Mbeams/s, 108.09 % realtime
 6 vehicles  :  32.296 Mbeams/s,  79.79 % realtime
 7 vehicles  :  30.986 Mbeams/s,  65.62 % realtime
 8 vehicles  :  31.573 Mbeams/s,  58.50 % realtime
 9 vehicles  :  32.557 Mbeams/s,  53.62 % realtime
10 vehicles  :  32.594 Mbeams/s,  48.32 % realtime
Max Mbeams/s:   36.460 Mbeams/s


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

Ignore the apparent change in frequency, all i did was map 2 more cores in Xen to this virtual machine.

 

The 2600 seems pretty low, should be performin better than that. Is it getting enough cooling, maybe full of dust? Check the temps.

 

The temps are fine, keep in mind it's running in a virtual machine with just half of the native cores.

Edit:
Analyzing the data I realize realtime simulation seems to work fine while up to (number of cores)-1 vehicles are spawned. It makes sense for one thread per car and one render thread but should there really be such a big drop in performance?

 

totes related

 

So apparently the Xen hypervisor might not enable Turbo frequencies automatically, I forced them on and got this:

Code:

[v3|0.1.1.0| Intel(R) Core(TM) i7-2600 CPU @ 3.40GHz (6 cores, 3397 MHz) | Sunday 07/20/14 00:34:48]
Beam ANAlysis NArrator, Version: 0.2
Test vehicle:  vehicles/pickup
 1 vehicle   :  12.666 Mbeams/s, 187.75 % realtime
 2 vehicles  :  24.401 Mbeams/s, 180.86 % realtime
 3 vehicles  :  30.989 Mbeams/s, 153.12 % realtime
 4 vehicles  :  32.495 Mbeams/s, 120.42 % realtime
 5 vehicles  :  38.043 Mbeams/s, 112.79 % realtime
 6 vehicles  :  33.680 Mbeams/s,  83.21 % realtime
 7 vehicles  :  31.709 Mbeams/s,  67.15 % realtime
 8 vehicles  :  30.884 Mbeams/s,  57.23 % realtime
 9 vehicles  :  33.103 Mbeams/s,  54.52 % realtime
10 vehicles  :  35.050 Mbeams/s,  51.96 % realtime
Max Mbeams/s:   38.043 Mbeams/s


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

I've never really had a need for fiddling with CPU tweaks until now, performance has been adequate in every game but now there's a useful real world metric to use. There's still this wall after 5 vehicles though. If I enable more cores they start moving into my Debian cores and that's something I'd rather avoid (since then context switching becomes an issue and I/O performance starts to drop). For the sake of science I can try to double the number of cores and see if it's a hardware or software issue...

Edit:
Just as expected the performance is a total disaster, however it's fun to watch:

Code:

[v3|0.1.1.0| Intel(R) Core(TM) i7-2600 CPU @ 3.40GHz (12 cores, 3397 MHz) | Sunday 07/20/14 01:28:07]
Beam ANAlysis NArrator, Version: 0.2
Test vehicle:  vehicles/pickup
 1 vehicle   :  11.677 Mbeams/s, 173.10 % realtime
 2 vehicles  :  20.163 Mbeams/s, 149.45 % realtime
 3 vehicles  :   4.297 Mbeams/s,  21.23 % realtime
 4 vehicles  :  16.851 Mbeams/s,  62.45 % realtime
 5 vehicles  :  18.377 Mbeams/s,  54.48 % realtime
 6 vehicles  :   3.427 Mbeams/s,   8.47 % realtime
 7 vehicles  :   7.589 Mbeams/s,  16.07 % realtime
 8 vehicles  :   6.320 Mbeams/s,  11.71 % realtime
 9 vehicles  :   3.901 Mbeams/s,   6.42 % realtime
10 vehicles  :   2.941 Mbeams/s,   4.36 % realtime
Max Mbeams/s:   20.163 Mbeams/s


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

The performance is wildly inconsistent between runs, but at least it proves you can play around with cores in Xen. Using the desktop feels like the pre-VT-x era and Windows 7 is desperately trying to reschedule processes to even out the load. Anyway, it was fun abusing the benchmark which in turn suggests an actual hardware limitation (never above 50% of realtime with more than 5 vehicles). Absolutely do disregard this particular run from the thread statistics.

 

Gochya...so I'd better invest in a better computer then. This laptop works wonders for all small tasks, breezes through emails and general navigation. Also plays Beamng at a steady 25-30 fps with 1 - 3 cars, but then it starts to taper off, and with it OC'd, it runs hot, so I'd rather not blow this thing out before I get another computer. Thanks for the tip

 

4670k at 4.0Ghz

 

My new FX-8320:

Stock (3.5GHz)


4.5GHz

 

This is saddening. I was thinking of upgrading to FX for a brief moment but...

 

My stock i7 3770S runs better than that in my current PC.
That's hilarious.



Sent from my iPhone 5s using Tapatalk

 

I told him not to get it.
I told him so...
I told him ;_;

 

That's why in my new PC build I'm getting an Intel i7 4790K @4.0GHz instead of AMD.


Sent from my iPhone 5s using Tapatalk

 

You never said that..

Besides, $210 for a mobo/CPU as compared to $350+...

 

*4.4Ghz
I plan on getting one or waiting for skylake. (Or whatever the new one is)

I'll admit, AMD has nice prices. That near matches my 4670k at stock and it cost $309.

 

that is one good cpu

 

2009er :[


(imported from here)

 

Don't get too cocky, there's nothing to brag about . You have a $300+ CPU with a sparkling new MoBo loaded with uberfast DDR3 memory that is being tailgated by a slightly o/c'ed 4 year old $100 chip on a 7 year old board with some DDR2 RAM .

And that was my point there. I was thinking of an upgrade to FX but even @4.5 GHz it isn't any better that what I already have.

 

Definitely not cocky at all, because I can get 30 fps on grid map with high on most things, low lighting and shadow quality, no vsync (drops my fps by 10 when turned on).
Maps like dry rock island and small island I get 15 fps, almost unplayable.

ON 1024x768. Ugh.


Sent from my iPhone 5s using Tapatalk