Author Topic: mobile_benchmark  (Read 4105 times)

Offline backslider

  • Community Developer
  • Prof. Inline
  • ******
  • Posts: 892
    • View Profile
mobile_benchmark
« on: 2011-Jan-24 »
Hi guys,

I wrote a little benchmark for mobile (or other) devices.
You can create cubes with the plus-button and delete them with the minus-button.

The program renders all faces (12 faces per cube).

It shows you the actual framerate, the cube count and number of all faces.

Here is a little example:


cheers

[attachment deleted by admin]
« Last Edit: 2011-Jan-25 by backslider »

Offline Marmor

  • Community Developer
  • Prof. Inline
  • ******
  • Posts: 908
  • 96A285CC
    • View Profile
    • my youtube channel
Re: mobile_benchmark
« Reply #1 on: 2011-Jan-24 »
ok
iphone 1G   cubes 30 360 faces    -- 10 FPS
ipod    2G    cubes 30 360 faces    -- 15 FPS
Palm Pre     cubes 30 360 faces    --  27 FPS  standard    settings to 500 MHZ
Palm Pre     cubes 30 360 faces    --  37 FPS  uberkernel settings to    1 GHZ

we need iphone 3gs , iphone 4  &  palmpre 2   at this time
 

Offline Slydog

  • Prof. Inline
  • *****
  • Posts: 929
  • KodeSource
    • View Profile
    • KodeSource
Re: mobile_benchmark
« Reply #2 on: 2011-Jan-24 »
I think it's really drawing twice the amount of polys, 24 per cube.
12 per cube per cull face, and you're drawing both sides of each poly with 'X_CULLMODE 0'.
To draw only one side, you could try 'X_CULLMODE 1' if the vertices are drawn in the right order (or '-1' if reversed).

And, it looks like you are recreating the cube objects each frame.
Could you not speed this up quite a bit if you only recreate the dynamic cubes whenever you add or delete one?

But, interesting demo, could come in handy when testing devices to find their limits.
My current project (WIP) :: TwistedMaze <<  [Updated: 2015-11-25]

Offline backslider

  • Community Developer
  • Prof. Inline
  • ******
  • Posts: 892
    • View Profile
Re: mobile_benchmark
« Reply #3 on: 2011-Jan-25 »
Okay, now I´ve changed X_CULLMODE to 1 (renders only visible faces) and the cubes will be created only once, not each frame.
It´s MUCH faster on my virtual machine (round about 250% performance boost).

Now you could test again on the phones, Marmor :)

If you see another performance problems in the code, please tell me.

cheers

P.S. I updatet the project in the first post.

Offline Kitty Hello

  • code monkey
  • Administrator
  • Prof. Inline
  • *******
  • Posts: 10666
  • here on my island the sea says 'hello'
    • View Profile
    • http://www.glbasic.com
Re: mobile_benchmark
« Reply #4 on: 2011-Jan-25 »
Great to see how the "old" Palm Pre outperforms an  iPod 2G.

Offline trucidare

  • Administrator
  • Prof. Inline
  • *******
  • Posts: 1377
  • Bachelor of Fail
    • View Profile
Re: mobile_benchmark
« Reply #5 on: 2011-Jan-25 »
uhm i think because 128 mb ram from ipod 2g and the 256 mb ram of the palm pre and the mbx lite instead of the sbx gpu. nice values for ipod 2g.
MacBook Pro 2,2 GHz Core 2 Duo, 4 GB RAM, 160 GB HDD, 8600M GT
Core i3 - 3,07 GHz, 8 GB Ram, 2.5 TB HDD, Geforce GTX 260+ OC

Offline Marmor

  • Community Developer
  • Prof. Inline
  • ******
  • Posts: 908
  • 96A285CC
    • View Profile
    • my youtube channel
Re: mobile_benchmark
« Reply #6 on: 2011-Jan-25 »
doy you really mean that 30 cubes needs 128 mb ram ?
its the gpu i guess .
who wrote a prime number check test for cpu  without print in the check (no gpu use)??

Offline Kitty Hello

  • code monkey
  • Administrator
  • Prof. Inline
  • *******
  • Posts: 10666
  • here on my island the sea says 'hello'
    • View Profile
    • http://www.glbasic.com
Re: mobile_benchmark
« Reply #7 on: 2011-Jan-25 »
I was wrong. iPod 2G = iPhone 3G, right? Or 3Gs?

Offline Marmor

  • Community Developer
  • Prof. Inline
  • ******
  • Posts: 908
  • 96A285CC
    • View Profile
    • my youtube channel
Re: mobile_benchmark
« Reply #8 on: 2011-Jan-25 »
The IPod 2g  (known also as IPod G) is the last IPod with 8Gb ram.
The Applications Processor inside the 2nd Generation iPod Touch runs slightly faster than the processor inside the iPhone 3G, but slower than the iPhone 3GS.


Offline Marmor

  • Community Developer
  • Prof. Inline
  • ******
  • Posts: 908
  • 96A285CC
    • View Profile
    • my youtube channel
Re: mobile_benchmark
« Reply #9 on: 2011-Jan-25 »
ok first test with the newer bench ! 
never use setscreen for mobile !!   i disable the function because it crashes !
ive testet for 30 fps because this is  the minimum  for playable 3d games or ?


Code: GLBasic [Select]
                                     CUBES            POLYGONES            FPS       START-FPS

Palm Pre  with standard 500 MHz      400               4800                30              63    
Palm Pre  with u-kernal   1 Ghz      400               4800                49              63  
IPod 2G                              400               4800                21              61
IPhone 1G                            400               4800                15              61
IPOD 3G
IPOD 4G

IPHONE 3G
Iphone 3GS
IPhone 4
IPAD (1)
IPAD (2)
ANDROID (later)
 


ähm thx who ever you are
« Last Edit: 2011-Jan-25 by Marmor »

Offline trucidare

  • Administrator
  • Prof. Inline
  • *******
  • Posts: 1377
  • Bachelor of Fail
    • View Profile
Re: mobile_benchmark
« Reply #10 on: 2011-Feb-02 »
iphone 3gs - 25 fps bei 401 cubes und 17 fps bei 649 cubes
MacBook Pro 2,2 GHz Core 2 Duo, 4 GB RAM, 160 GB HDD, 8600M GT
Core i3 - 3,07 GHz, 8 GB Ram, 2.5 TB HDD, Geforce GTX 260+ OC