@Niolator

It looks like in everyday usage SAM 800MHz is equal to AOne SE 600MHz...not bad.

@DAX

I'll do the same tests with 4.1.1 then we compare

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

@KimmoK

Quote:

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

I sincerely hope so! The RAM Disk on my system (Sam/OS4.1-debug kernel) is incredibly slow. I tried a program that frequently accesses datafiles on disk. Guess what - although I have a VERY slow 2.5" harddisk, running the program from Ram Disk is noticeably slower than when run from HD!

_________________
The Rear Window blog

AmigaOne X5000/020 @ 2GHz / 4GB RAM / Radeon RX 560 / ESI Juli@ / AmigaOS 4.1 Final Edition
SAM440ep-flex @ 667MHz / 1GB RAM / Radeon 9250 / AmigaOS 4.1 Final Edition

@ChrisH

>I'm afraid I only half understood you.

Wy do nobody the lame benchmark with SAM ?

I happily get some values from a friendly SAM user.Here can see when do some calculation with 68 k Program SAM have really bad performance when measure performance at same theory clockspeed.

Ok SAM have no 2. Level Cache, but the 604 on the cyberstorm PPC have too not a 2 .Level cache and cyverstorm PPC have slow RAM.

So it can compare the 604 and the SAM good ,here are the values.Please do a G3 user too the test.

http://amigaworld.net/modules/newbb/viewtopic.php?topic_id=29964&forum=25&13

when use FPU performance then a 604/G4 run at 500 MHZ have same speed as a SAM run on 700 MHZ.also on other tests the 604 or G4 is faster much faster

PPC 604 150 MHZ OS4

Speed test for FFT + iFFT: (float)
time needed 19449ms for 413696 samples, => .241165652871131x speed
@44100Hz/stereo
------------ (5840ms at 500 MHZ)
Speed test for FFT + iFFT: (integer)
time needed 16485ms for 413696 samples, => .28452718257904x speed
@44100Hz/stereo
------------ (4950ms at 500 MHZ)
Speed test for FFT + iFFT: (integer handopmized 68K ASM)
time needed 7108ms for 413696 samples, => .659880518913269x speed
@44100Hz/stereo
------------ (2134ms at 500 MHZ)

----------------------


Sam 667 MHZ OS4

2.WORKBENCH:> "RAM Disk:FFTDemo"
Speed test for FFT + iFFT: (float)
time needed 6136ms for 413696 samples, => .764411807060241x speed
@44100Hz/stereo
------------ (8185ms at 500 MHZ)
Speed test for FFT + iFFT: (integer)
time needed 4735ms for 413696 samples, => .990587294101715x speed
@44100Hz/stereo
------------ (6316 ms at 500 MHZ)
Speed test for FFT + iFFT: (integer handoptimized 68K ASM)
time needed 2510ms for 413696 samples, => 1.86869752407073x speed
@44100Hz/stereo
------------ (3348ms at 500 MHZ)
2.WORKBENCH:>

Last edited by bernd_afa on 13-Nov-2009 at 03:52 PM.
Last edited by bernd_afa on 13-Nov-2009 at 03:52 PM.
Last edited by bernd_afa on 13-Nov-2009 at 03:51 PM.

@bernd_afa
Ain't that the same 68K emulation benchmark you proposed in another thread?

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

@Thread

Here are my results with my AmigaOne XE/G4 (Acube) and a Radeon 9200 pro on the AGP port :

FFMpeg test :

5mn20

Lame test :

8.Tests:> lame WhereDoAllTheJunkiesComeFrom.mp3
ID3v2 found. Be aware that the ID3 tag is currently lost when transcoding.
LAME 3.98.2 32bits (http://www.mp3dev.org/)
Using polyphase lowpass filter, transition band: 16537 Hz - 17071 Hz
Encoding WhereDoAllTheJunkiesComeFrom.mp3
to WhereDoAllTheJunkiesComeFrom.mp3.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
9468/9473 (100%)| 0:34/ 0:34| 0:34/ 0:34| 7.0908x| 0:00
-------------------------------------------------------------------------------
kbps LR MS % long switch short %
128.0 3.6 96.4 88.5 6.9 4.6
Writing LAME Tag...done
ReplayGain: -5.4dB

Cube test :

10936 frames drawn in 14.0 seconds
1223 fps (0.817 milliseconds per frame)
785 fps (1.274 milliseconds per frame with swap)

I did not do the HD test yet.

_________________
PowerMac G5 2,7Ghz - 2GB - Radeon 9650 - MorphOS 3.14
AmigaONE X1000, 2GB, Sapphire Radeon HD 7700
FPGA Replay + DB 68060 at 85Mhz

@DAX

>Ain't that the same 68K emulation benchmark you proposed in another thread?

yes, but there are some systems miss for completeness.

aone with G3
efika
pegasus 2 with G4 and do test on MOS and OS4 to see if the JIT speed is maybe diffrent.

currently i get values of ATOM X86 systems and a P3 650 need calc the 500 MHZ value and post later.

wy does nobody post the lame values of SAM ?

Last edited by bernd_afa on 13-Nov-2009 at 05:45 PM.
Last edited by bernd_afa on 13-Nov-2009 at 05:44 PM.
Last edited by bernd_afa on 13-Nov-2009 at 05:43 PM.

@bernd_afa
It would be nice to have a suite of benchmarks that entails several common usage PPC programs (as that comprises the biggest portion of the overall experience a final user will have with the HW).

20 different programs (like they do on Toms hardware Guide for Windows HW) should suffice I think.

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

@all

lame test SAM440flex 800MHz

2.RAM Disk:> lame WhereDoAllTheJunkiesComeFrom.mp3
ID3v2 found. Be aware that the ID3 tag is currently lost when transcoding.
LAME 3.98.2 32bits (http://www.mp3dev.org/)
Using polyphase lowpass filter, transition band: 16537 Hz - 17071 Hz
Encoding WhereDoAllTheJunkiesComeFrom.mp3
to WhereDoAllTheJunkiesComeFrom.mp3.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
4584/4588 (100%)| 0:40/ 0:40| 0:41/ 0:41| 2.9335x| 0:00
-------------------------------------------------------------------------------
kbps LR MS % long switch short %
128.0 4.7 95.3 88.0 7.2 4.8
Writing LAME Tag...done
ReplayGain: -5.2dB

Last edited by mr2 on 13-Nov-2009 at 05:52 PM.

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

@mr2

mmm...
your lame test was a bit strange...
Why got 40 secs but 2,9x?
I got 58 secs but 4,2x...
Maybe you hanvent donwloaded the entire file(maybe truncated?)
and for better comparison pls beefore encoding decode it to wav with:

lame --decode

Thank you! :)

_________________
Simone"Tuxedo"Monsignori, Perugia, ITALY.

Has anyone tried to optimize code for PPC440 DSP instructions?
Are the instructions useful for any desktop use?
Could they speed up TCP/IP stack or something like that (PP440 is a Telecom chip after all).

update:
btw. Do not stop.
http://aminet.net/search?query=benchmark

Last edited by KimmoK on 14-Nov-2009 at 12:54 PM.

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

@KimmoK Quote:

wb and DOS copy. Both were slow

(compared to DOpus4)

Sounds like a buffering issue. Presumably the drives you use do not have large buffers (are you using FFS?), so copy routines without large copy buffer's themselves may be slowed. And presumably DOpus4 allocates it's own large buffers.

@bernd_afa
Sorry that I'm not benchmark obsessed like you. Maybe when OS4.1.1 is out it might be worth doing some (and less hassle as I won't have to stop using Debug Kernel every time I want to do a test).

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

@ChrisH

>Sorry that I'm not benchmark obsessed like you.

thats not good, if you do a programming language its important that you do benchmarks to see the weak point of a CPU and you can change the optimizer to produce faster code.

maybe the amblitz benchmark is not for you usefull because you do portable E, but i think when you do that in amiga E is also usefull and help you to get faster executables from amiga e when you change the optimizer.

@KimmoK

Here my tests made with the only 2 other tests available for AmigaOS4 on the list:

SDLBench:


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): 4.39319 166.667 0.481435 46.7836
Fast points (frames/sec): 168.865 101.628 38.2147 25.3717
Rect fill (rects/sec): 13255.7 113778 3346.41 65015.9
32x32 blits (blits/sec): 42226.8 87148.9 39384.6 117029

CGXBench:

Raw transfer speed
------------------
Screen | Register to | FAST RAM to
depth | video RAM | video RAM
-------+-------------+-------------
15 | 164.4 MB/s | 60.7 MB/s
16 | 161.9 MB/s | 74.5 MB/s
24 | 164.4 MB/s | 60.7 MB/s
32 | 164.4 MB/s | 60.7 MB/s


WritePixelArray() 320x240
-------------------------
Screen | Source: LUT8 | Source: ARGB
depth | secs | fps | MB/s | secs | fps | MB/s
-------+-------+-------+------+-------+-------+------
8 | 0.19 | 1078.0 | 79.0 | ----- | ----- | ----
15 | 0.78 | 257.0 | 75.3 | 0.49 | 203.6 | 59.6
16 | 0.36 | 550.3 | 80.6 | 0.32 | 310.8 | 45.5
24 | 0.79 | 254.4 | 74.5 | 0.50 | 198.0 | 58.0
32 | 0.78 | 255.9 | 75.0 | 0.51 | 194.9 | 57.1


ScalePixelArray() 320x240 -> 640x480
------------------------------------
Screen | Source: LUT8 | Source: ARGB
depth | secs | fps | MB/s | secs | fps | MB/s
-------+-------+-------+------+-------+-------+------
8 | ----- | ----- | ---- | ----- | ----- | ----
15 | ----- | ----- | ---- | 0.96 | 103.7 | 30.4
16 | ----- | ----- | ---- | 1.12 | 89.6 | 13.1
24 | ----- | ----- | ---- | 0.99 | 101.2 | 29.7

_________________
Simone"Tuxedo"Monsignori, Perugia, ITALY.

@bernd_afa
Well, by completely changing topics twice, you just proved that you are a nasty troll, who attacks whatever they think they can, rather than actually being interested in being helpful.

I am not interested in arguing with trolls, nor helping them by posting benchmarks (if it does not serve your purpose then you will ignore it, but if it helps you nasty purpose then you will use it).

(Edit: And yes, I could answer your implied criticism(s) - but what would be the point? You would just change topics again.)

Last edited by ChrisH on 15-Nov-2009 at 07:12 PM.
Last edited by ChrisH on 15-Nov-2009 at 07:04 PM.

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

ehm...
can we remain in topic plz...
I dont want flames here plz...

_________________
Simone"Tuxedo"Monsignori, Perugia, ITALY.

@Tuxedo

Here my test on SAM 440 Flex 800MHz, 1 GB RAM

mkfile hd:file 200000000 test

SFS0->RAM: 10s cpu 67-77%
SFS0->SFS0 13s cpu 39-42%

Video conversion test

16 min 50 s

Lame test

LAME 3.98.2 32bits (http://www.mp3dev.org/)
Using polyphase lowpass filter, transition band: 16537 Hz - 17071 Hz

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
12948/12948 (100%)| 1:59/ 1:59| 2:00/ 2:00| 2.8233x| 0:00
-------------------------------------------------------------------------------
kbps LR MS % long switch short %
128.0 1.9 98.1 99.1 0.6 0.3
Writing LAME Tag...done
ReplayGain: -7.9dB

No time to test Quake

_________________
BACK FOR THE FUTURE

http://www.betatesting.it/backforthefuture

Sam440ep Flex 800 Mhz 1 GB Ram + AmigaOS 4.1 Update 6
AmigaOne XE G3 800 Mhz - 640 MB Ram - Radeon 9200 SE + AmigaOS 4.1 Update 6

@ChrisH
>Well, by completely changing topics twice, you just proved that you are a nasty troll, >who attacks whatever they think they can, rather than actually being interested in >being helpful.

I have done a abuse report, that you call me a troll about this is really bad.

And my arguments that compiler decelopers are intrestet about benchmark is no troll.How they should get out of the optimizer the best speed without benchmarks ?

Also you see, that in X86 land when a new system come to market, there are benchmarks and speed values all around.

So please explain me, wy there are so much Benchmarks on many sites for X86 when you call a user that is intrestet on speed on SAM as a troll ?

How is that possible that the lame output values on SAM diffrent to the values from aone ?

replay gain mean the loudness of signal and -5.4db to -7.9db say the SAM output signal is near 20% more silence.(6 db is double loudness).

aone

""""
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

Sam

kbps LR MS % long switch short %
128.0 1.9 98.1 99.1 0.6 0.3
Writing LAME Tag...done
ReplayGain: -7.9dB
""""

Last edited by bernd_afa on 17-Nov-2009 at 08:55 AM.
Last edited by bernd_afa on 17-Nov-2009 at 08:54 AM.
Last edited by bernd_afa on 17-Nov-2009 at 08:54 AM.
Last edited by bernd_afa on 17-Nov-2009 at 08:46 AM.

@bernd_afa Quote:

that you call me a troll about this is really bad.

It was my honest assessment, for the following reasons given below:

Quote:

And my arguments that compiler decelopers are intrestet about benchmark is no troll.How they should get out of the optimizer the best speed without benchmarks ?

Oh it is a troll, because:
(1) You have never contacted me (privately) with any concerns about performance, so clearly you are not interested in help me, but only wish to bash me in public by implying there are problems.
(2) The issues was benchmarking OS4.1 on various hardware, it was NOT benchmarking compilers on unspecified OSes/hardware.
(3) You changed the topic (and went off-topic), because you wanted to bash me for not doing what you wanted.

Quote:

you call a user that is intrestet on speed on SAM as a troll ?

I don't call asking for SAM benchmarks a troll. I call REPEATEDLY publically HARASSING me personally for benchmarks as trolling. You don't know when to drop it, nor how to be polite, and frankly I am fed-up with you.

@Tuxedo Quote:

can we remain in topic plz... I dont want flames here plz...

I'm sorry, but when bernd_afa addresses me personally with accusations, I have to respond in some fashion. (I tried to avoid it continuing, by not answering his accusations, but that doesn't seem to have helped too much.)

Last edited by ChrisH on 17-Nov-2009 at 09:45 AM.
Last edited by ChrisH on 17-Nov-2009 at 09:44 AM.

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

@ChrisH

>I don't call asking for SAM benchmarks a troll. I call REPEATEDLY publically >HARASSING me personally for benchmarks as trolling. You don't know when to >drop it, nor how to be polite, and frankly I am fed-up with you.

I have not repeatly asking you for benchmark.read here.

http://amigaworld.net/modules/newbb/viewtopic.php?mode=viewtopic&topic_id=30008&forum=2&start=20&viewmode=flat&order=0#520235

""""
>I'm afraid I only half understood you.

Wy do nobody the lame benchmark with SAM ?
""""

i write nobody and thats diffrent to you...

>(1) You have never contacted me (privately) with any concerns about performance, >so clearly you are not interested in help me, but only wish to bash me in public by >implying there are problems.

In one of your answers you use the word obsessed.thats really no nice word, in this term, so wy i should think when i send you a private mail that you tell me how fast SAM is ?

and that other not think i am benchmark obsessed, i explain wy benchmarks are usefull.

This post you do was btw offtopic.

""""
Your Post
http://amigaworld.net/modules/newbb/viewtopic.php?mode=viewtopic&topic_id=30008&forum=2&start=20&viewmode=flat&order=0#520462

"""
Sorry that I'm not benchmark obsessed like you
"""
""""

Just to place the benchs in real world perspective:

On an old (2/3 years old) iMac with core2duo@1.8ghz it takes 8 secs. With the very same mp3 file and lame version.

_________________
http://www.warpdesign.fr/