To not further "derail" threads I make this one to collect benchmarks to compare 68k to other platforms

Results from Pavlor:

Now on Core i5-2500 (WinUAE):
Lame
7.8558x (440EP 2100 MHz )

dnetc
OGR-NG: 5,055,431 (440EP 525 MHz, 750GX 450 MHz )
RC5-72: 4,851,033 (440EP 1850 MHz, 750GX 1400 MHz )

Doom Benchmark (ADoom 1.3)
114 realtics (655.2 fps)
For Comparison:
Pegasos II G4 1 GHz ADoomPPC1.7, MorphOS - 97 realtics (770.0 FPS - WarpOS wrapper)
A1-XE G4 1 GHz ADoomPPC1.7, AmigaOS4 - 175 realtics (426.8 FPS - WarpOS wrapper)
A1-XE G4 1 GHz ADoom 1.3, AmigaOS4 - 486 realtics (153.7 FPS - 68k Petunia emulation)

Edit:
Speedometer 4.0 (shapeShifter Mac emulation)
CPU: 41.868 (= 68040 1.0 GHz)
Dhrystone: 104.509 (= 68040 2.6 GHz)

Lame on WinUAE (Core2 Q6600 2.4 GHz):
4.0742x

Lame on Pegasos2 G4 1131 MHz (see OP in that thread)
10.304x (AltiVec?)
4.2741x (Non AltiVec?)

Lame on SAM440flex 800MHz (post 28)
2.9335x


Core i5-2500 in newer computer of my brother is cca 2x faster than old Q6600.

Edit: Another result for G4.

On notebook I use right now (Core i5-2430M 2.4 GHz):
Lame
5.8853x

dnetc
OGR-NG: 4,051,434 (lower than 440EP 666 MHz)
RC5-72 3,867,526 (higher than 750GX 1000 MHz)

From OP in thread I listed:

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:

Piru´s tests you linked use another wav file:
http://sintonen.fi/temp/AKsack.wav

I got with this file:
7.2076x
(38s CPU time; 33s REAL time)


xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx

Links from me:

http://www.amiga-news.de/en/news/AN-2012-02-00011-EN.html

Last edited by OlafS25 on 18-Feb-2014 at 11:43 AM.

@OlafS25

http://amigaworld.net/modules/newbb/viewtopic.php?topic_id=37054&forum=45&88


and:
http://www.amiga.org/forums/showthread.php?t=60604&highlight=lame+benchmarks+%28pun+intended%29

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

BTW is there anyone who could make a test on smartphones/tablets and on amithlon PCs?

new one:
http://via.i-networx.de/bench_en.html
http://www.jonnydigital.com/amiga-benchmarks

Last edited by OlafS25 on 18-Feb-2014 at 04:47 PM.
Last edited by OlafS25 on 18-Feb-2014 at 04:46 PM.
Last edited by OlafS25 on 18-Feb-2014 at 02:37 PM.
Last edited by OlafS25 on 18-Feb-2014 at 02:22 PM.
Last edited by OlafS25 on 18-Feb-2014 at 02:21 PM.
Last edited by OlafS25 on 18-Feb-2014 at 01:21 PM.
Last edited by OlafS25 on 18-Feb-2014 at 12:54 PM.
Last edited by OlafS25 on 18-Feb-2014 at 12:20 PM.
Last edited by OlafS25 on 18-Feb-2014 at 12:01 PM.
Last edited by OlafS25 on 18-Feb-2014 at 11:59 AM.
Last edited by OlafS25 on 18-Feb-2014 at 11:51 AM.

@OlafS25

Quote:

BTW is there anyone who could make a test on smartphones/tablets and on amithlon PCs?

UAE on ARM CPUs doesn´t have JIT yet (it will be sloooooow).

For Amithlon:
http://www.amigahistory.co.uk/emulators/amithlonbench.html

In some cases it reaches 1/3 of native performance (comparable to Petunia or Trance JIT in OS4/MotphOS)!

@OlafS25

Any modern platform should support MMU for features like virtual memory addressing, memory protection or paging. Here are results of WinUAE on Core i5-2500 3.3 GHz with MMU enabled - JIT doesn´t work with MMU...

dnetc
OGR-NG: 140,652 (68040 50 MHz)
RC5-72: 60,117 (68040 66 MHz, PowerPC 601 33 MHz )

lame
0.1522x

Edit: lame finished, at last!

Last edited by pavlor on 19-Feb-2014 at 08:10 PM.

@OlafS25

Any comparisons to x86 or x64?

_________________
Amiga Developer
Amiga 500, Efika, Mac Mini and PowerBook

@itix

X86/X64 lame tests I have seen were about 6-9 seconds

AROS was 9 seconds if I remember right

Last edited by OlafS25 on 21-Feb-2014 at 02:02 PM.

@OlafS25

Grouping some results from http://www.aros-platform.de/html/benchmarks.html


>Benchmarks (68k UAE in comparation to other platforms)
>Aksack.wav + lame (different sources)

Amiga OS 4.x + native lame
>X1000 PA6T 1.8 GHz 18s (altivec used? & one core).
>Sam 460ex 1.15 GHz 47s
>Sam 440 ep 800 MHz 1:15 min

Morph OS + native lame.
>MacMini G4 1.56 GHz 33s (mit Altivec 16s)
>Powerbook 5,6 1,67GHz 16s (altivec in use)
>Dual G5 2.7 GHz 7s (altivec used & one core)??
>Efika 2min 20s

>Peg2 1131 23s (OS? g4? altivec?)

Assumingly WinUAE with JIT, single core in use.
>i5-2500 (68k UAE) 38s
>AMD Athlon II Dual Core 2.10 GHz (68k UAE) 1:09 min
>Core2Duo 3 GHz system with Xubuntu (68k UAE) 46s (via WINE)
>AMD A6-4400M 2.70 GHz (68k UAE) 45s

AROS + NATIVE LAME (SSE in use, single core, 32bit SW) ????
>AMD Athlon II Dual Core 2.10 GHz (64bit) 10s

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

Hellbillylitre AmigaTwox86x64x6000x AMD Phenom II x4 965 WinUAE (can't remember version, ca. aug. 2014)

_________________
Using: One Commodore C64 - One Commodore Amiga 500 - One Commodore Amiga 1200 with BVision and Blizzard 68060 with PPC coprocessor running Amiga DOS - One Hellbillylitre Amigatwox86x64x6000x running Windows7

@Thread How to read benchmarks?

Lesser seconds mean machine more performant?

What does 2.9335x multiplier in the last column means for example?

Last edited by Raffaele on 05-May-2015 at 06:12 PM.

_________________
"When the Amiga came out, everyone [at Apple] was scared as hell." (J.L. Gassée, former CEO of Apple France and chief of devs of Mac II-fx, interviewed by Amazing Computing, Nov 1996).

@Raffaele

the system is faster when seconds are lower and the number before x is higher (both belongs together).

Lame converts the file so less seconds means faster running

the number before x are related to the lame test. I look what it exactly means

Last edited by OlafS25 on 05-May-2015 at 06:21 PM.

@OlafS25

this thread really is an pretty example of a mess. if you want to have something informative then set up a database containing clean categories of benchmarks and systm descriptions. otherwise its no use and the whole effort may be simply skipped in favour of anything more important.

@OlafS25

Quote:

the system is faster when seconds are lower and the number before x is higher (both belongs together). Lame converts the file so less seconds means faster runnin

While running something like LAME as comparison is in generally a great idea,
you have to be very careful when selecting the right benchmark binaries.
LAME like the LAME files from AMINET can be extremely misleading.

On Aminet are many several LAME binaries.
Some of the binaries are compile with -O2 or -O3, some are compiled with -O0.

I've seen over 100% performance differences the different binaries on Aminet.

_________________
APOLLO the new 68K : www.apollo-core.com

@thread

Hi all!

For what they are worth, using SysInfo v4.0, here are numbers with three different configurations of my Amiga 1200 + Blizzard 68060:

- Without much patch and no special processor settings:
MIPS: 49.04
FPU: 36.77
- With all the processor caches and superscalar enabled using the "Processor" patch of MCP:
MIPS: 54.82
FPU: 39.25
- Finally, with all the processor caches and superscalar enabled + SystemPatch:
MIPS: 55.32
FPU: 39.63

Processor configuration and SystemPatch make a difference!

Cheers!

_________________
Tygre
Scientific Progress Goes Boing!

and we can resume native lame like here:

_________________

The tread says clearly "Benchmarks of 68k (UAE, real hardware) so here's my Amiga x86x64x6000x VS my real Amiga 1200

My real Amiga still beats my x86x64x6000x on AreaFill.

_________________
Using: One Commodore C64 - One Commodore Amiga 500 - One Commodore Amiga 1200 with BVision and Blizzard 68060 with PPC coprocessor running Amiga DOS - One Hellbillylitre Amigatwox86x64x6000x running Windows7

@Seiya

Lame on PEG2 @ 1.1 GHz is about 50% faster than Lame on MacMini @ 1.5 GHz ?

How can we explain this?

_________________
APOLLO the new 68K : www.apollo-core.com

@BigGun

Using a overclocked G3 on the PEG2 vs G4+Altivec on the mini?
Or MacMini was running Moana & AOS4 version of LAME....?
Or MacMini ran 68k version of the LAME binary?

Last edited by KimmoK on 07-May-2015 at 07:15 AM.

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

@BigGun

Quote:

How can we explain this?

Different lame versions (with and without AltiVec) I guess. That is why I ever give version numbers (and configuration) of applications I benchmarked.

Peg 2@1,1 Ghz should run on MorphOS. in don't know if use Altivec or not

_________________