@Fab

Quote:

I'm not really sure glquake or quake2 OS4 ports work with OS4Emu. But i'll see.

If you could give it a try, I thank you, if they don't work, don't worry about it, I'm sure I'll have a look at these things at some stage (after setting up MOS on my Peg2), I'm just too occupied with other things at the moment, I didn't even have time last night to do that AVI test with the fixed IDE driver.

sorry

this is the link for my test

A1 vs Peg vs Amithlon and in some case vs 68k Amigas

http://xoomer.virgilio.it/tuxcam/bench/result.html

_________________

aha

Last edited by Wishmaster on 19-Feb-2009 at 04:24 PM.
Last edited by Wishmaster on 19-Feb-2009 at 04:23 PM.

_________________
Pegasos PPC with MorphOS

@Seiya

Wow! Your 68k machine is faaaast!!

Even faster than Core 2 Quad Q6600 2400 MHz WinUAE box of my brother!
My RC5 68k result is about 1,900,000.00 keys/sec (you have more than 3,000,000.00 on AMD Athlon XP - Core 2 should be more than 2 times faster).

@Seiya

This looks like total 'selective data.' Plus:
1. A1/OS4 wasn't tested except for a small handful of cases. WHY?
2. No links to source of the programs used, nor params, are mentioned.
3. No system configuration data is given (clean boot, no WBStartup, vs 'any old apps running,' etc )
4. dnetc certainly runs on A1/OS4, and with JIT - why nothing there?
5. Some of your tests are obviously done with JIT disabled. (or not done at all) Why?

Incomplete data with benchmarks or only 'selective results' really isn't of much *real* use.
Can you answer the above or clarify?

_________________
Are we not done with the same silly arguments and flames yet??!

@jahc

Quote:

jahc wrote: Cobra, can you download that test video clip from Fab's site http://fabportnawak.free.fr/benchmark.avi and then test Dvplayer with the fixed Peg2 ide driver?

Here it is (finally):

6.Dev:dvplayer> DvPlayer_0.65 noaudio noskip notriplebuffer verbose Media:benchmark.avi
Video: AVI, 1024 x 576, 40.00 fps
Audio: 01 [MP3] 16-bit 44100 Hz, Stereo
Total Nr of Frames: 4002
Nr of Frames played: 3937
Nr of Frames skipped: 65 (2%)
Total Playback Time: 122.357 seconds
Average Framerate: 32.708 fps
Displayed Framerate: 32.176 fps

Note that I used 'notriplebuffer' because otherwise the vsync'ing will make the results depend on your monitor's vertical frequency, without that option here I get 124 seconds.

EDIT: Some frames are skipped because the video decoder finds some frames corrupted and is unable to decode them.

Last edited by COBRA on 20-Feb-2009 at 11:00 AM.

@COBRA

Quote:

6.Dev:dvplayer> DvPlayer_0.65 noaudio ___noskip___ notriplebuffer verbose Media:benchmark.avi

Quote:

Nr of Frames skipped: 65 (2%)

hmmm ?

@rzookol

The file is a bit corrupted, so the decoder skips some frames it's unable to decode (mplayer's verbose output shows a lot of errors too)

Quote:

wegster wrote: @Seiya This looks like total 'selective data.' Plus: 1. A1/OS4 wasn't tested except for a small handful of cases. WHY? 2. No links to source of the programs used, nor params, are mentioned. 3. No system configuration data is given (clean boot, no WBStartup, vs 'any old apps running,' etc ) 4. dnetc certainly runs on A1/OS4, and with JIT - why nothing there? 5. Some of your tests are obviously done with JIT disabled. (or not done at all) Why? Incomplete data with benchmarks or only 'selective results' really isn't of much *real* use. Can you answer the above or clarify?

yes :)
this test is made some years ago when i have asked to italian amiga community to make some benchmark within AmigaOne and Pegasos.

And then i have asked to make a 68k benchmark on OS4, MorphOS to compare with Amithlon.

the benchmark file i used is here.

A1/OS4 wasn't full tested because in italy for my experience, not all amiga users like very much made a benchmark.

you have to consider these test like a first OS4 and MOS benchmark when OS4 was born and the first test against MorphOS.

we say that this bench is an amithlon bench vesersu A1 and MOS,
only few users has acepted to make these test and i thanks again these people
benchmarking with 68k test againt amithlon has bring to us a first duel within OS4 and MOS with they limit siince that time.

after, when OS4 and MOS was more mature, i hope to have more modern test, but amiga community has a strange relationship with benchmark.

The tests also try that Amitlhon was much faster than WinUAE and much faster (on 68k application) than OS4 and MOS.

i like to see now, a 68k bench with Sam440 and Efika wth my test.
it will be very interesting for me to see P2@350 with amithlon versus Sam and Efika.

_________________

@pavlor
Quote:

pavlor wrote: @Seiya Wow! Your 68k machine is faaaast!! Even faster than Core 2 Quad Q6600 2400 MHz WinUAE box of my brother! My RC5 68k result is about 1,900,000.00 keys/sec (you have more than 3,000,000.00 on AMD Athlon XP - Core 2 should be more than 2 times faster).

amithoni is much faster than winuae :)

_________________

@thread

What I`m seeing here is that [anypc] + Amithlon outclasses all MOS/OS4 systems by such a ludicrous margin that one wonders why everyone isn`t using it.

_________________
Fraggle

- insert profound text here -

@Seiya

Quote:

yes :) this test is made some years ago when i have asked to italian amiga community to make some benchmark within AmigaOne and Pegasos.

So basically those benchmarks are missing results, and are for really old versions of OS4 and MorphOS. They don't tell us anything useful because we're interested in the performance of the current systems.

@all

As misleading as benchmarks are/can-be, they can be used to find bottlenecks. I'd be interested to see Fab's tests repeated with the fixed IDE driver. It would also be nice to see tests that don't involve OpenGL (we know the issues there) and/or old compiles of games/apps that don't take advantage of new features in Amiga OS 4.1.

Regarding new features, has Amiga OS 4.1's SDL port been updated to use the new compositing features when they're available (i.e., support hardware alpha blending)? If not, that's something that needs to be done.

Hans

_________________
Join the Kea Campus - upgrade your skills; support my work; enjoy the Amiga corner.
https://keasigmadelta.com/ - see more of my work

@Hans

Quote:

Regarding new features, has Amiga OS 4.1's SDL port been updated to use the new compositing features when they're available (i.e., support hardware alpha blending)? If not, that's something that needs to be done.

Does XMAME use alpha blending functions?

@jahc

Quote:

jahc wrote: @Hans Quote: Regarding new features, has Amiga OS 4.1's SDL port been updated to use the new compositing features when they're available (i.e., support hardware alpha blending)? If not, that's something that needs to be done. Does XMAME use alpha blending functions?

No idea, but it was compiled back in 2005, so any improvements to OS4's SDL port since then won't be included.

Hans

_________________
Join the Kea Campus - upgrade your skills; support my work; enjoy the Amiga corner.
https://keasigmadelta.com/ - see more of my work

@Glames

Compositing should be faster.

_________________
ikir

@Seiya

Quote:

i like to see now, a 68k bench with Sam440 and Efika wth my test. it will be very interesting for me to see P2@350 with amithlon versus Sam and Efika.

In 2004, 68k benchmarks were useful (majority of used applications was 68k). Now, native benchmarks are needed for real comparision (WOS is not native). P2 350 MHz is as fast as 5200B in Efika, 440EP in SAM is two times faster than P2 350 MHz.

I will try your benchmarks on my brother´s PC and you can try The DOOM Benchmark, recent OGR (I don´t know why, but I think it is better for comparision than RC5) and some sort of Dhrystone 2.1 benchmark (I know it is old and outdated...).

Last edited by pavlor on 20-Feb-2009 at 08:35 AM.

@Hans

Quote:

Hans wrote: Regarding new features, has Amiga OS 4.1's SDL port been updated to use the new compositing features when they're available (i.e., support hardware alpha blending)? If not, that's something that needs to be done.

I don't think so, the latest SDL is dated 30.01.2008, before OS4.1's release date. It would be indeed great to see SDL supporting the newest 4.1 features.

Varthall

_________________
AmigaOne XE - AmigaOS 4.1 - Freescale 7457 1GHz - 1GB ram

@pavlor

Quote:

Wow! Your 68k machine is faaaast!! Even faster than Core 2 Quad Q6600 2400 MHz WinUAE box of my brother! My RC5 68k result is about 1,900,000.00 keys/sec (you have more than 3,000,000.00 on AMD Athlon XP - Core 2 should be more than 2 times faster).

There must be something wrong with your brother's WinUAE box. I just tested the latest m68k dnetc client on my WinUAE PC (Core 2 Duo E8400@4GHz) and it gave me 6,568,902 nodes/sec in OGR-NG and 6,357,450 keys/sec in RC5-72.

On the other hand, m68k Quake is more than 50% faster on that same setup when compared to the native PPC version on my Pegasos 2.

@KimmoK

Quote:

>Ambinet is open source, but depends heavily on MUI4. As long a sMUI4 is not available for OS4 it will require major changes. Okey dokey. If AOS4 had MUI4, it would have had sputnik yesteryear. So clearly MUI4 is the bigger fish to catch. I would buy AOS4 version of MUI4 for 50¤, anyone else? :)

It was said that there is the first open source version of Ambient which depends of MUI3.9 to work on MorphOS, its source it's downloadable, and after a porting of these souces to OS4 we could merge to new Ambient version with svn and substitute MUI4 parts with proper MUI3.9 parts...it's not black magic if someone want port Ambient on OS4.
So first step is to port Ambient MUI3.9 on OS4, anyone who offer himself to this port?
Noooo MUI is evil!!!!111

Last edited by ShInKurO on 20-Feb-2009 at 09:18 AM.

@Hans
Quote:

Hans wrote: @Seiya [quote] As misleading as benchmarks are/can-be, they can be used to find bottlenecks. I'd be interested to see Fab's tests repeated with the fixed IDE driver. It would also be nice to see tests that don't involve OpenGL (we know the issues there) and/or old compiles of games/apps that don't take advantage of new features in Amiga OS 4.1. Hans

Well even with the ide bug included I doubt thet it takes that much time to read ~ 100 MB from the hd, nor that the PIO mode eats up all that cpu power. The PIO modes indeed slows down things, but I seriously doubt it is the only cause for the difference.

_________________
My programs: via.bckrs.de
MorphOS user since V0.4 (2001)