The test systems. If people think the GFX card is too much different for some test, I can replicate later swapping cards for consistency.

MDD - (MD)
1.42Ghz stock CPU
DDR RAM
Radeon 9000 Pro
MorphOS 2.7 (Demo)

Quicksilver 2002 - (QS)
2.0Ghz 7448 CPU (NewerTechnology 1.8Ghz OC to 2.0Ghz)
PC133 RAM
Radeon 8500
MorphOS 2.7 (registered)


MEMTEST (no parameters, bigger is better)
MD - Writing 691MB & 693MB, verifying 387MB & 387MB
QS - Writing 313MB & 313MB, verifying 385MB & 385MB

No surprise here. The MDD has a faster RAM interface with 167Mhz bus speed to 133Mhz on the QS, and the Quicksilver always disappoints in this test. I'll be curious to see what other tests show.


SunSider-0.9.1 Javascript Benchmark (smaller is better)
MD - 4637.4ms
QS - 3328.3ms

The greater CPU speed of the 2.0Ghz Quicksilver really helps beat the 1.42Ghz MDD here!

I will be updating this original top post every time I add benchmark results. I am looking at probably downloading AmigaMark next.

[ Edited by redrumloa 10.04.2012 - 18:02 ]

I just went back and checked the X1000 results for SunSpider. The AmigaOne X1000 scored a 5,729.9ms on this benchmark. This MDD cost me almost nothing and it beats out the X1000 despite its higher clock speed (1.8Ghz).

This is not the point of the thread, just something I find curious.

Hey Red,

Glad to see this thread (even though you brought the X1000 into it, which I think is a mistake, unless you want the thread to degrade into an X1000 & Hyperion bashing free-for-all)

Since Odyssey 1.16 is not available on OS4.x, the SunSpider benchmark can't be run on the X1000 with the same browser and the comparison is not an "apples to apples" comparison, but instead it is a comparison of Odyessy 1.16 to OWB 1.9 (I think that is the version of OWB that is used for the port of OWB to OS4.x).

No big deal to me one way or the other. I just would hope that this thread can remain focused on your benchmark results and the comparison of the accelerated Quicksilver and the stock 1.42GHz MDD PowerMac's, instead of bringing anything else into the comparisons that will detract from the discussion and comparison of two MorphOS systems.

I do think that the Radeon 8500 in the Quicksilver gives it an added advantage in some tests results. Do you mind doing the tests with the same video card in each computer, or do you have another video card that is closer to having the same performance as either the 9000Pro, or the 8500 that you can install in one of the systems to even out the test results more. I think I have an extra 9000Pro I could send to you, so you could have a 9000Pro in each of your G4 PowerMac's for testing purposes.

Looking forward to more results and some more real world program and/or game test results. This thread is very useful to MorphOS users who are considering buying a Quicksilver, or an MDD G4 PowerMac system and want to know which one is the fastest and best one to buy.

Any thoughts about over-clocking your dual 1.42GHz G4 CPU card? I have read that a bump up to 1.5GHz is usually successful and not too difficult. I don't know how much that little bit of extra speed will shorten the life of your dual 1.42GHz G4 CPU module, so it might not be worth the effort and risk, for so little improvement in speed.

Are both systems maxed out with RAM? Are both systems using IDE hard drives that spin at the same speed? If you can keep as many of the components identical, or swap them from one system to the other, so the tests will only reflect the differences between the bus speed, memory speed, caches, and the CPU's, the tests will be more useful in determining how much the choice of which PowerMac and the accelerator makes in the speed of the system, instead of introducing other differences between one hard drive and another, or some other system component, which can be changed easily, instead of comparing the components that can't easily be changed, such as bus speed, memory speed, memory caches, and the faster G4 CPU's that are much harder to find for the MDD PowerMac's than they are to find for the Quicksilver.

Edit: That is one fast Quicksilver you have there. Nice MorphOS system and probably one of the fastest MorphOS systems on the planet. I hope that Jim can also contribute his test results, from his 1.6GHz G4 Quicksilver, and his dual 1.42GHz G4 MDD PowerMac.

[ Edited by amigadave 10.04.2012 - 19:18 ]

>memtest ...


Hmm odd, I get 546 MB/sec on my QS here (Sonnet 2x1.8GHz). Any chance of a recheck what the QS does with the normal 1.8 ?

And just out of curiosity, how did you overclock ??

>Sunsider

The QS checks out at around 4000 (done a few rounds), while the 1.5GHz-Mini gets 4500 not sure WHAT we are really measuring here....

[ Edited by Kronos 11.04.2012 - 12:48 ]

amigadave,
Quote:

I doubt there will be much talk about the X1000 here, not my intent.

Quote:

If I do anything obviously GFX card intensive like Quake III, I'll move over the 8500. I just really hate to be constantly unplugging and plugging the gfx card. I'd hate for you to make the expense of mailing another Radeon though.

Quote:

I may overclock eventually, but I don't trust this MDD yet. I got it as "dead" and have it working, but am seeing a few small quirks now. If/when I eliminate the quirks, I'll try OC.

Quote:

No, I have both systems at 512MB for these tests. For hard drives:
MDD - Maxtor 30GB UDMA/100 7200RPM 2MB IDE Hard Drive
QS - WD Caviar 80 GB - 100 MBps - 7200 rpm

Kronos,
Quote:

Is your Quicksilver a Quicksilver 2002? I can test later at 1.8, but I doubt it will make a difference. I got worse results with the stock 800Mhz CPU. Quicksilver seems to be pretty well known for poor memtest results.

Quote:

The NewerTechnology MaxPower has dip switches, very easy. I simply used the dip switches to increase the multiplier. So far I am rock solid stable at 2Ghz.

>QS vs. QS2000

Mmmm, not sure cos as far I can see they don't really differ except for the (stock) CPU-module, which I don't have for this unit..

1357.0ms with intel atom dual core at 1,5ghz and intel gma3150 graphics chip (tested on packard bell netbook with windows 7 starter and the latest firefox)

Quote:

Please, bench without javascript JIT compiler.

https://wiki.mozilla.org/JavaScript:TraceMonkey

Cego,
Quote:

Who the F*%k cares about comparing an Intel Atom dual core running Win7 against a single core operating system from 2001 on antique PPC hardware? This thread is about comparing different MorphOS systems against each other on a MorphOS forum site, not a Windows or general PC forum site.

> single core operating system from 2001

Transposed digits, right? :-)

@Cego

Quote:

235.5ms here on my Core-i7 with Chrome 18.

Anyway, architectural performance differences aside, there is also a major impact of JIT for JS. Odyssey on MorphOS doesn't have that.

Are there any standalone benchmark programs that actually work? I just tried a few and no go. The last one I tried was AmiGod MOS. Everything looks good except the RUN TESTS button is ghosted out. I read the guide and readme, they say nothing!

booo, hiss

Ad my AmiGOD:

Redrumloa, so called "amigod2mos" was actually my "AmiGOD 2 Reloaded" project in early stage, which was not supposed to be on Aminet. So readme says nothing, because it's just plain changelog about GUI and hardware detection - from my homepage "it doesn't test anything, it only recognizes your Pegasos hardware" (yet). It's rewritten AmiGOD from scratch, finally C language, MUI, MorphOS native, but really in early stage, development of tests was not finished, project is paused for years.

"AmiGOD 2 - BETA 15.06.2003" benchmark (2D graphics tests, CPU tests, intuition test, memory and hdd tests) is compatible with MorphOS 1.x and was tested on several AmigaOS / MorphOS machines (Pegasos I G3, II G4) - table of benchmark results:
http://lukysoft.cz/?page=benchmarks
but seems like it's not stable under MorphOS 2.7 anymore :( This version was written in PowerD language from scratch (switch from AmiGOD 1.50 in BlitzBasic), but I found some bugs in language itself, can't implement some features, development was slowed down, so later I decided to abort AmiGOD 2 written in PowerD lang and switch again to another language (AmiGOD 2 Reloaded in C lang.).

AmiGOD homepage for more details - five "from scratch" versions:
http://lukysoft.cz/?page=amigod


Ad benchmarks:

There is a lot of "real world" applications where you can compare speed of machines. F.e. MPlayer2 playing HD videos, lame mp3 convert, Blender rendering or many games with FPS counters.

Good CPU intensive benchmark is also DNetC client RC5-72 and ORG-NG, it has large database and it's results are comparable across many CPU platforms. I always test my new computer (CPU) with DNetC ;) We have official client for MorphOS - GUI only requires MUI NList class installed.

Good luck with testing.

> Are there any standalone benchmark programs that actually work?
> I just tried a few and no go.

AmigaMARK 1.1 doesn't work?

Memtest always with the option "disable", please.

redrumloa,
Quote:


Do these real-life tests, we can have better way to compare results:

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

luky-amiga,

Thanks for the info!


Andreas_Wolf,

Nope, Amigamark 1.1 doesn't work. Is there something beyond the readme that needs to be done?

Tom01,

Will re-do with disable


Miky060,

Sounds like a plan.

Just an off topic question: I've just bought a FW800 with dual G4@1,42 GHz too. Anyone knows if 2x1GB DDR modules can be used in this computer or just 512MB modules would be recognized?

Yes, but MacOS X won't see more than 2gb, no matter what you put in.

[ Edited by stephen_robinson 13.04.2012 - 19:31 ]

Thanks! Last question: if I use just two 1GB modules is there a particular order to put them in or can I use whichever slot I prefer?
I know they can seem strange and nonsense questions, but I'm just trying to fitting a fan behind the heatsink and trying to not touch the ram modules.

> Amigamark 1.1 doesn't work.

I just tried it on my Mac mini and all tests work like a charm. What error do you get?

> Is there something beyond the readme that needs to be done?

Not that I'm aware of.

Miky060,
Quote:


err, not sure, I don't think it matters, but I'm not going to be able to check, sorry.

edited to add the rather important 'not'

[ Edited by stephen_robinson 22.04.2012 - 15:34 ]

Sorry for the delay, doing more testing on the Quicksilver. I will update the original post when I finish testing the QS and MDD. This post is a raw dump of Quicksilver tests.

Re-did memtest with DISABLE switch:
Write -312MB, verify 386MB

lame aksack.wav:
17s

Looks like the memory speed is hurting this test. A good result, only beat by 1.67Ghx Powerbook with MOS 3.0, but tied with MacMini 1.5ghz.

Where is this mythical bricks_and_water_plugin.blend?

> Where is this mythical bricks_and_water_plugin.blend?

http://os4depot.net/index.php?function=showcontent&file=graphics/raytrace/blender.lha

But beware:
http://amigaworld.net/modules/news/article.php?storyid=6338&start=20#85636
http://amigaworld.net/modules/news/article.php?storyid=6338&start=20#85641

;-)