Hi ALL!

Here the test updated with my "new" PEgasos2@1131 and Radeond9250:

******************************************************************************************************
*NEW* : added in parenthesis () tests relative to my new gfx card a Radeon9000pro *NEW*
* RageMEM test was realtinve to Radeon9000pro card
******************************************************************************************************

nbench(http://amigadev.free.fr/powerpc/nbench.html) test:

BYTEmark* Native Mode Benchmark ver. 2 (10/95)
Index-split by Andrew D. Balsa (11/97)
Linux/Unix* port by Uwe F. Mayer (12/96,11/97)

TEST : Iterations/sec. : Old Index : New Index
: : Pentium 90* : AMD K6/233*
--------------------:------------------:-------------:------------
NUMERIC SORT : 590.05 : 15.13 : 4.97
STRING SORT : 18.371 : 8.21 : 1.27
BITFIELD : 1.6372e+08 : 28.08 : 5.87
FP EMULATION : 70.537 : 33.85 : 7.81
FOURIER : 6054.3 : 6.89 : 3.87
ASSIGNMENT : 13.728 : 52.24 : 13.55
IDEA : 2500 : 38.24 : 11.35
HUFFMAN : 1119 : 31.03 : 9.91
NEURAL NET : 9.3971 : 15.10 : 6.35
LU DECOMPOSITION : 322.4 : 16.70 : 12.06
==========================ORIGINAL BYTEMARK RESULTS==========================
INTEGER INDEX : 25.657
FLOATING-POINT INDEX: 12.018
Baseline (MSDOS*) : Pentium* 90, 256 KB L2-cache, Watcom* compiler 10.0
==============================LINUX DATA BELOW===============================
CPU :
L2 Cache :
OS : AmigaOS 53.8
C compiler : gcc version 4.0.2 (AmigaOS build 20051012)
libc :
MEMORY INDEX : 4.657
INTEGER INDEX : 8.129
FLOATING-POINT INDEX: 6.666
Baseline (LINUX) : AMD K6/233*, 512 KB L2-cache, gcc 2.7.2.3, libc-5.4.38
* Trademarks are property of their respective holder.

RageMem(http://os4depot.net/share/utility/benchmark/ragemem.lha) test:

RAGEMEM v0.37 - compiled 11/06/2010

CPU: Motorola MPC 7447/7457 Apollo 1.1 @ 1133 Mhz
Caches Sizes: L1: 32 KB - L2: 512 KB - L3: none
Cache Line: 32

* CPU*
MAX MIPS: 3398

*L1*
READ32: 4318 MB/Sec
READ64: 8634 MB/Sec
WRITE32: 3701 MB/Sec
WRITE64: 2880 MB/Sec

* L2 *
READ32: 1942 MB/Sec
READ64: 2529 MB/Sec
WRITE32: 1745 MB/Sec
WRITE64: 2193 MB/Sec

*RAM*
READ32: 144 MB/Sec
READ64: 144 MB/Sec
WRITE32: 197 MB/Sec
WRITE64: 425 MB/Sec
WRITE: 795 MB/Sec (Tricky)

* VIDEO BUS *
READ: 51 MB/Sec
WRITE: 221 MB/Sec

c-ray test(http://os4depot.net/share/utility/benchmark/c-ray.lha):

8.RAM Disk:c-ray-1.1> c-ray-f -i scene -o foo.ppm
Rendering took: 3 seconds (3960 milliseconds)
8.RAM Disk:c-ray-1.1> c-ray-f -i sphfract -o foo.ppm
Rendering took: 81 seconds (81147 milliseconds)

LAME test:

2.RAM Disk:> lame WhereDoAllTheJunkiesComeFrom.mp3.wav
LAME 3.98.2 32bits (http://www.mp3dev.org/)
Using polyphase lowpass filter, transition band: 16537 Hz - 17071 Hz
Encoding WhereDoAllTheJunkiesComeFrom.mp3.wav
to WhereDoAllTheJunkiesComeFrom.mp3.wav.mp3
Encoding as 44.1 kHz j-stereo MPEG-1 Layer III (11x) 128 kbps qval=3
Frame | CPU time/estim | REAL time/estim | play/CPU | ETA
9467/9467 (100%)| 0:24/ 0:24| 0:24/ 0:24| 10.304x| 0:00
-------------------------------------------------------------------------------
kbps LR MS % long switch short %
128.0 3.6 96.4 88.6 6.9 4.5
Writing LAME Tag...done
ReplayGain: -5.4dB


ffmpeg TEST:

4 minutes 29 seconds

Quake1 TEST:

SOFTWARE:

Demo1 39.4 FPS (40.9)
Demo2 38.2 FPS (39.7)
Demo3 38.1 FPS (39.5)

OpenGL:

Demo1 84.2 FPS (114)
Demo2 94.7 FPS (130.4)
Demo3 67.0 FPS (90.3)

Cube TEST(slowest than previous I think that my new GFX board was worst than previous...):

7286 frames drawn in 14.8 seconds
849 fps (1.177 milliseconds per frame)
496 fps (2.012 milliseconds per frame with swap)

***** *NEW* Cube TEST with Radeon9000pro:

13758 frames drawn in 15.0 seconds
1177 fps (0.849 milliseconds per frame)
923 fps (1.083 milliseconds per frame with swap)

File Copy:

from sfs0 to ram: got about 8 seconds 70% cpu
from sfs0 to sfs0 (same hdd different partition) got about 24 secs and 13% cpu


SDL TEST(from http://code.google.com/p/os4sdl/downloads/list ):

Mode = 320x240, software
Pitch = 320
Hardware surfaces avail = 1
Window manager avail = 1
Blitter hardware = 1
Colorkey blit hardware = 0
Alpha blit hardware = 0
Software->Hardware accel = 0
Video memory = 0

Slow points test
Fast points test
Rect fill test
32x32 Blitter test
Mode = 320x240, hardware
Slow points test
Fast points test
Rect fill test
32x32 Blitter test
Mode = 640x480, software
Slow points test
Fast points test
Rect fill test
32x32 Blitter test
Mode = 640x480, hardware
Slow points test
Fast points test
Rect fill test
32x32 Blitter test
320x240 320x240 640x480 640x480
software hardware software hardware
Slow points (frames/sec): 12.0846 148.148 1.59968 51.6129
Fast points (frames/sec): 755.162 118.849 194.973 29.8786
Rect fill (rects/sec): 21005.1 113778 6511.92 55351.4
32x32 blits (blits/sec): 55351.4 141241 53194.8 141241


Still waiting to try OpenArena but dunno how to make a bench with it...
I found on the NET that I have to download a specify map for bench but none found(dead links)...
Anyone can help there?



**********************
OLD Topic text:
**********************

I've dome some nice benchmarks on my A1SE@666 with 512MB RAM and Radeon9250 gfx card runing AmigaOS4.1...

EDIT: Also have Sil680 and wb set to 1680x1050 32bit composite on.

I post the results here so maybe someone can do the same tests and compare the tests(from every OS and hw...):

I executed:

1 - video conversion test:

got the video http://www.youtube.com/watch?v=nUT5Y5LtiIs with GetVideo:

GetVideo "http://www.youtube.com/watch?v=nUT5Y5LtiIs" save

got ffmpeggui(and ffmpeg) from os4depot)
choosed the predefined "dvd (pal)" conversion

command for ffmpeg line was(copied from ffmpeggui output):

"ffmpeg" -i "Mast:BenchMarks/Sigla_completa_giapponese_Sailor_Moon,.mp4" -target pal-dvd -vcodec mpeg2video -b 6000k -r 25 -s 720x576 -acodec ac3 -ar 48000 -ab 448k -ac 2 "RAM Disk:Test.mpg"

the mast partition where the video was is a SFS0 partition.
elapsed time about 15 minuti e 15 seconds...calculated with a clock-cronograph.

2 - lame 3.98-2 test:

got the public mp3 file(the first I found):

http://www.maninblack.org/demos/WhereDoAllTheJunkiesComeFrom.mp3

decompessed with:

lame --decode to obtain wav file for the test.

than executed lame without args(simply "lame filename") and got:

11.RAM Disk:> lame WhereDoAllTheJunkiesComeFrom.mp3.wav
LAME 3.98.2 32bits (http://www.mp3dev.org/)
Using polyphase lowpass filter, transition band: 16537 Hz - 17071 Hz
Encoding WhereDoAllTheJunkiesComeFrom.mp3.wav
to WhereDoAllTheJunkiesComeFrom.mp3.wav.mp3
Encoding as 44.1 kHz j-stereo MPEG-1 Layer III (11x) 128 kbps qval=3
Frame | CPU time/estim | REAL time/estim | play/CPU | ETA
9467/9467 (100%)| 0:57/ 0:57| 0:58/ 0:58| 4.2741x| 0:00
-------------------------------------------------------------------------------
kbps LR MS % long switch short %
128.0 3.6 96.4 88.6 6.9 4.5
Writing LAME Tag...done
ReplayGain: -5.4dB

3 - file copy test:

create a file with mkfile(from os4depot):

mkfile hd:foo 200000000

I got:

copying file from sfs0 to ram: about 14 seconds cpu load(from cpu.docky) about 74%
copying from sfs0 to sfs0 (same hd different partitions): about 28 cpu load about 28%


4 - Quake test:

I got the packages from OS4Depot().
For the software version I set the res to 800x600
For OpenGL I leave the default res choosen from the program.

I got:

SOFTWARE:

Demo1 25,6 FPS
Demo2 24,8 FPS
Demo3 24,7 FPS

OpenGL:

Demo1 73,8 FPS
Demo2 82,6 FPS
Demo3 56,1 FPS


EDIT2:

New test(for OS4 users and maybe OS4Emu too?)
Download cube bencmark from OS4depot:

I got:

8693 frames drawn in 15.5 seconds
923 fps (1.083 milliseconds per frame)
567 fps (1.763 milliseconds per frame with swap)

That's ALL!

Hope that few and simple tests was nice for someone

PS I've also OpenArena installe dbut dunno the way to get timedemo test if someoneknow it plz tell me

Last edited by Tuxedo on 25-Apr-2017 at 12:17 PM.
Last edited by Tuxedo on 01-Feb-2012 at 12:56 AM.
Last edited by Tuxedo on 01-Feb-2012 at 12:53 AM.
Last edited by Tuxedo on 01-Feb-2012 at 12:35 AM.
Last edited by Tuxedo on 01-Feb-2012 at 12:29 AM.
Last edited by Tuxedo on 29-May-2011 at 11:11 AM.
Last edited by Tuxedo on 23-Mar-2011 at 11:08 PM.
Last edited by Tuxedo on 13-Nov-2009 at 07:01 PM.
Last edited by Tuxedo on 12-Nov-2009 at 09:55 PM.
Last edited by Tuxedo on 12-Nov-2009 at 09:42 PM.

_________________
Simone"Tuxedo"Monsignori, Perugia, ITALY.

@Tuxedo
I don't know what you are doing right, or I am doing wrong, but on my Sam GLQuake is really bad: Looks bad, graphical glitches & often slows right down. For demo2 I only get 24 fps (at 1280x1024 but resolution hardly seemed to make any difference).

Where-as with normal Quake (800x600 like you), I got 22.5 fps (which is almost the same as your 667MHz A1SE).

BTW, I am running the Debug Kernel, so performance will be worse than normal (so normal Quake may perform identically or even better than your A1SE!). Looking forward to OS4.1.1. where Sam won't need debug kernel.

_________________
Author of the PortablE programming language.
It is pitch black. You are likely to be eaten by a grue...

@ChrisH

I'm waiting here to test again with update :)
I'm really curious of the results :)

However the benchmarks I propose was intended to test every Amiga and non hw just for fun and curiosity :)

Good test to everyone!

_________________
Simone"Tuxedo"Monsignori, Perugia, ITALY.

Can anyone show me a benchmark of these two setups?

# CD32 with SX-32 Mk2 expansion and 8MB RAM
# A1200 with E-matrix '030 card and 32MB RAM

_________________
SEX SUX

Sex doesn't prove to me that sex doesn't suck.

@ChrisH
>BTW, I am running the Debug Kernel, so performance will be worse than normal (so >normal Quake may perform identically or even better than your A1SE!). Looking >forward to OS4.1.1. where Sam won't need debug kernel.

this apps do not much in AOS funcs so please try the lame Benchmark.
here is same too, there are much calculation but very few amiga OS func calls.

@bernd_afa
I'm afraid I only half understood you.

Last edited by ChrisH on 12-Nov-2009 at 05:31 PM.
Last edited by ChrisH on 12-Nov-2009 at 05:30 PM.

_________________
Author of the PortablE programming language.
It is pitch black. You are likely to be eaten by a grue...

@Tuxedo

SAM 440flex 800MHz, RAM-1GB, HDD-320GB 7200rpm 2,5''(WB 1920x1200 32Bit):

mkfile hd:file 200000000 test

SFS0->RAM: 14s cpu 98-100%
SFS0->SFS0 13s cpu 65-71%
JXF->RAM: 15s cpu 95-100%

WB 1280x1024 32Bit
Pretty same results

video conversion test

16 min 50 s

Have no time for the other benchmarks

Last edited by mr2 on 12-Nov-2009 at 09:00 PM.
Last edited by mr2 on 12-Nov-2009 at 05:36 PM.

_________________
Sam440ep-flex 800MHz 1GB RAM R9250 128MB SB Live!

@mr2

It's interesting how slow RAM disk access is. Is there any known speed ups coming from Hyperion for RAM disk handling?

other than that ... from your numbers ... sfs to sfs transfer delivers 15M+15M simultaneously. Not too bad.
I wonder how much faster it happens if the whole read operation is buffered before starting the writing process? (would we close up to HW speed limits?)

I test too.... SAM440ep + Seagate 2,5" 7200rpm 80GB + DEBUG kernel (1920x1080x16+compositing off))
(OWB and Dopus4 running as well as a test docky anim on the bar, no CPU meter)
mkfile ... ok, that was fast. Less than one second?
sfs0 to RAM - 5 seconds with Dopus4
sfs0 to sfs0 - 6...7 seconds again with dopus4
copy foo to RAM: in shell - 5 seconds
copy foo to DH0:tst/ 13 seconds ??

Dos prefs file handle buf size 8kB->64kB
no change? Do I need to reboot after changes?...

workbench copybuffersize is 256kb and it copies same speed than dos.
I tried with 10Mb copy buffer and workbench copy process slows down...??

better use good old Dopus.

Last edited by KimmoK on 13-Nov-2009 at 07:52 AM.
Last edited by KimmoK on 13-Nov-2009 at 07:43 AM.
Last edited by KimmoK on 12-Nov-2009 at 08:13 PM.
Last edited by KimmoK on 12-Nov-2009 at 08:05 PM.
Last edited by KimmoK on 12-Nov-2009 at 08:03 PM.
Last edited by KimmoK on 12-Nov-2009 at 08:01 PM.
Last edited by KimmoK on 12-Nov-2009 at 07:55 PM.

_________________
- KimmoK
// For freedom, for honor, for AMIGA
//
// Thing that I should find more time for: CC64 - 64bit Community Computer?

ooops... OWB back button crashed the whole computer...
(and copying from dos is not any faster after reboot)

Yet another speed test:
Cube benchmark & default window size...
(Cube Benchmark by Alex Carmona - Based on the MiniGL cube demo.)
13824 frames in 16.1 seconds
1054 fps
871 fps with swap (what swap)


I wish that was Q3 speed...

Last edited by KimmoK on 12-Nov-2009 at 08:16 PM.

_________________
- KimmoK
// For freedom, for honor, for AMIGA
//
// Thing that I should find more time for: CC64 - 64bit Community Computer?

@KimmoK

Nice testto add :)
(Edited and updated first post)

With cube test I got:

8693 frames drawn in 15.5 seconds
923 fps (1.083 milliseconds per frame)
567 fps (1.763 milliseconds per frame with swap)


so a bit faster really little difference...so system was rather equal in that way...
I wonder someone with Sam733(with faster fsb) to see what changes...

_________________
Simone"Tuxedo"Monsignori, Perugia, ITALY.

@Tuxedo

cube:

20584 frames drawn in 24.5 seconds
1136 fps (0.880 milliseconds per frame)
849 fps (1.177 milliseconds per frame with swap)


sam440ep 667mhz

Last edited by Antique on 12-Nov-2009 at 10:18 PM.

_________________
I'm an antique. Don't light my fuse

@Antique

plz try to test with about 16 secs runtime...
I noticed different fps with different times...

Thank you!

_________________
Simone"Tuxedo"Monsignori, Perugia, ITALY.

@Tuxedo

cube test

WB 1280x1024 32bit:

7757 frames drawn in 15.7 seconds
758 fps (1.318 milliseconds per frame)
497 fps (2.009 milliseconds per frame with swap)

WB 1920x120032bit:

7776 frames drawn in 16.0 seconds
764 fps (1.308 milliseconds per frame)
489 fps (2.041 milliseconds per frame with swap)

_________________
Sam440ep-flex 800MHz 1GB RAM R9250 128MB SB Live!

@all

Anyone knows why AOS4 disk to disk copying is so slow when compared to Dopus4?

_________________
- KimmoK
// For freedom, for honor, for AMIGA
//
// Thing that I should find more time for: CC64 - 64bit Community Computer?

@KimmoK
For "AOS4" do you mean "Workbench"?

_________________
Author of the PortablE programming language.
It is pitch black. You are likely to be eaten by a grue...

@ChrisH

wb and DOS copy. Both were slow.

Quote:

I test too.... SAM440ep + Seagate 2,5" 7200rpm 80GB + DEBUG kernel (1920x1080x16+compositing off)) (OWB and Dopus4 running as well as a test docky anim on the bar, no CPU meter) mkfile ... ok, that was fast. Less than one second? sfs0 to RAM - 5 seconds with Dopus4 sfs0 to sfs0 - 6...7 seconds again with dopus4 copy foo to RAM: in shell - 5 seconds copy foo to DH0:tst/ 13 seconds ?? Dos prefs file handle buf size 8kB->64kB no change? Do I need to reboot after changes?... workbench copybuffersize is 256kb and it copies same speed than dos. I tried with 10Mb copy buffer and workbench copy process slows down...??

Last edited by KimmoK on 13-Nov-2009 at 09:37 AM.

_________________
- KimmoK
// For freedom, for honor, for AMIGA
//
// Thing that I should find more time for: CC64 - 64bit Community Computer?

@Thread
One thing that should be stated is that there are Tests that could be hindered by Debug Kernel, and unoptimized/half buggy routines in the OS and others which only measure raw CPU power (such as a 3D rendering for example).

That said I would suggest to stop any further testing of AOS4.1 on SAM hardware until the new update will be out.

Contrary to what have been stated here by some "other camp" supporters (which stated that aside from 4/5 new features the update won't produce any boost in performance since it's just a bunch of Bug Fix), the new version is indeed better optimized, and I received confirmation in various forms, and even the latest issue of Amiga Future (#81 just out) confirms that (QUOTE FROM PAGE 4): "4.1.1......will update almost ALL system components...".

Considering that helping the Sam version is Acube's Massimiliano Tretene which is also porting to Sam several 440 optimized C functions, I believe that with 4.1.1 (which as ChrisH already stated will also allow us to run AOS without the Debug Kernel), we will have a better understanding of overall system performance.

Why start benchmarks at the very eve of a major update?

_________________
SamFlex Complete 800Mhz System + AmigaOS 4.1 Update 4
Amiga 2000 DKB 2MB ChipRam GVP G-Force040 Picasso 2 OS3.9 BB2
AmigaCD 32

@DAX

"Why start benchmarks at the very eve of a major update?"

Because it is fun!

And because we want to prepare to analyze improvements when 4.1.1 arrives.

_________________
- KimmoK
// For freedom, for honor, for AMIGA
//
// Thing that I should find more time for: CC64 - 64bit Community Computer?

@KimmoK
Quote:

And because we want to prepare to analyze improvements when 4.1.1 arrives.

That makes sense

_________________
SamFlex Complete 800Mhz System + AmigaOS 4.1 Update 4
Amiga 2000 DKB 2MB ChipRam GVP G-Force040 Picasso 2 OS3.9 BB2
AmigaCD 32

Very interesting results. So a SAM at 800 MHz is almost as fast as a G3 at 666 MHz? It would be interesting to see which scores A1XE G4 and Pegasos produce.