Ok guys to start a comparison what is the speed diff with a Pegasos 2 and Powermac G4 1.25ghz box?

Well, it depends on the exact model, because the single Powermac G4 versions have different FSB frequencies, different RAM setups (DDR-RAM, SDR-RAM), and nevertheless different G4 versions, with or without L3 cache, which can overcome some deficits regarding the FSB. My experience is that even a default G4 400Mhz setup can be as fast as a Peg2, regarding memspeed, with the possibility of using Altivec.

> the single Powermac G4 versions have different FSB frequencies

All stock 1.25 GHz G4 PowerMacs have 167 MHz FSB, no matter if single or dual G4.

> different RAM setups (DDR-RAM, SDR-RAM)

All stock 1.25 GHz G4 PowerMacs have DDR-SDRAM (PC2700, 333 MHz), no matter if single or dual G4.

> with or without L3 cache

All stock 1.25 GHz G4 PowerMacs have L3 cache, no matter if single or dual G4. MDD (2002) has 2 MiB, FW800 has 1 MiB and MDD 2003 has 1 or 2 MiB (depending on information source).

> a default G4 400Mhz setup can be as fast as a Peg2, regarding memspeed,
> with the possibility of using Altivec.

Do you mean a 1000 MHz G4 Peg2 or a 600 MHz G3 Peg2?

@Andreas_Wolf:

>All stock 1.25 GHz G4 PowerMacs have 167 MHz FSB, no matter if single or dual G4.

Well, i thought that there maybe is a 3rd-party card with 1,25ghz for the older Powermacs.

>All stock 1.25 GHz G4 PowerMacs have DDR-SDRAM (PC2700, 333 MHz), no matter if single or dual G4.

Same answer.

>All stock 1.25 GHz G4 PowerMacs have L3 cache, no matter if single or dual G4. MDD (2002) has 2 MiB, FW800 has 1 MiB and MDD 2003 has 1 or 2 MiB (depending on information source).


OK, that makes sense, because there is no 1,25ghz 3rd party cpu card which comes without L3 level cache.

>Do you mean a 1000 MHz G4 Peg2 or a 600 MHz G3 Peg2?

Both in the end, but I was speaking of a Peg2 with G4. I know that the CPU frequency is obviously higher, but memspeed is higher anyway, ironically with a lower FSB frequency (133 <-> 100Mhz).

> i thought that there maybe is a 3rd-party card with 1,25ghz for the older Powermacs.

I bet there is. But as magnetic didn't mention any CPU upgrade I took it he meant only stock 1.25 GHz G4 PowerMacs. I guess he'll clarify soon.

> Peg2 with G4. I know that the CPU frequency is obviously higher, but memspeed
> is higher anyway, ironically with a lower FSB frequency (133 <-> 100Mhz).

Interesting. Seems the Marvell Discovery II is the culprit. At least I don't have any other explanation for a CPU/FSB MHz combination of 1000/133 resulting in lower memspeed than 400/100.

thx for the responses guys.. yes i'm referring to the peg2 g4 vs other pmac g4s. Real world results would be cool..

>> as magnetic didn't mention any CPU upgrade I took it he meant
>> only stock 1.25 GHz G4 PowerMacs. I guess he'll clarify soon.

> yes i'm referring to the peg2 g4 vs other pmac g4s.

That's obvious. What's not obvious and thus prone to misunderstanding is if you referred only to stock 1.25 GHz G4 PowerMacs (as I assumed) or also to older PowerMacs with 1.25 GHz G4 3rd party upgrade cards (as analogkid assumed).

i refer to both and all.. some results would be cool

>> What's not obvious and thus prone to misunderstanding is if you referred
>> only to stock 1.25 GHz G4 PowerMacs (as I assumed) or also to older
>> PowerMacs with 1.25 GHz G4 3rd party upgrade cards (as analogkid assumed).

> i refer to both and all..

Thanks for clarification. Seems I misunderstood your question to not include 3rd party CPU upgrades.

In memtest, my MDD write slower than my Pegasos ( 320MB/sec VS 449 ).

Tiki

tiki
thanks posting. this is an example ofthe kind of real world examples we are looking for

@Tiki: my Gigabit gives 438MB/s

"Verifying" ( that mean reading ? ) give me 387MB/sec for the MDD and 208 for the Peg.
Maybe I use a bad ram quality.

Youtube run really smoothly on the MDD. I think that the sonnet G4@1.8 help. C'est un pur bonheur.
MDD's boot time is faster than Peg's boot.
When Morphos Freeze, Pegasos reboot faster because there is no reset button on MDD

Tiki

> Youtube run really smoothly on the MDD. I think that the sonnet G4@1.8 help.

How does it run with 720p videos in fullwindow mode? What's the CPU workload?

I had "benchmark" a 720p trailer. Mplayer play in 67sec this movie. Normalement, he 's during 95sec.
The CPU workload is between 60 and 80 % in normal play.
So, in full windows, it's very smooth.

For benchmarking, use this command : mplayer -benchmark -nosound -ao null tonfichier


Tiki

>>> Youtube run really smoothly on the MDD. I think that the sonnet G4@1.8 help.

>> How does it run with 720p videos in fullwindow mode? What's the CPU workload?

> I had "benchmark" a 720p trailer. [...] in full windows, it's very smooth.

Thanks, but I specifically referred to your mention of Youtube. So again, to make it more clear: How does Youtube run with 720p videos in fullwindow mode in OWB? Is the CPU workload between 60 and 80% there as well?

Sorry for the misunderstood. I tested somes trailer HD in Youtube. Non HTML5 doesn't run very good. HTML5 are correct in FullScreen mode ( cpu 100% ), and smooth in non fullscreen mode ( cpu 100 % ). It's possible that compatibility is broke again. I will see that with Fab.

Tiki

>> How does Youtube run with 720p videos in fullwindow mode in OWB?
>> Is the CPU workload between 60 and 80% there as well?

> Non HTML5 doesn't run very good.

That was expected ;-)

> HTML5 are correct in FullScreen mode ( cpu 100% ) [...]. It's possible
> that compatibility is broke again.

Tried http://fabportnawak.free.fr/owb/scripts/YouTube_Fullscreen_Fix.js (from October 1st)?

tiki

Thanks for your reports.. i wish others would post results here..

Real world results of Peg2 g4 vs powermac g4 1ghz +

1. OWB internet page load times

2. Opening drawer contents with tons of icons

3. Image Fx or Art Effect render tests

4. Lightwave or Blender render tests

etc

Test 1 is quite pointless at least. It's impossible to guarantee identical conditions for this test, unless if you load some page from RAM disk.

Test 2 requires quite a bit of preparations and as identical as possible configuration to make sense.

piru

Yeah, it would be nice to have comprehensive tests. I'm trying to decide if i should keep my peg2 or go powermac!

The main factor for me is driver speed like ethernet, ram, bus speed etc. Is it worth the change?

@tiki

You must choose a mp4 format. The flv one adds quite some overhead (no idea why, though).

And 360p vids take about 30% cpu in overlay mode on my Peg2, for reference.

Sorry to be one of the Bigest " Boulet " of the world. In fullscreen 720p, youtube work perfectly (approx 90%CPU) IF I use the right button and select Fullscreen (menu contextuel), otherwise, that don't work.

Really sorry for my bad english

Tiki

[ Edited by Tiki on 2010/10/18 19:29 ]

> In fullscreen 720p, youtube work perfectly (approx 90%CPU)

Thanks. This means that at least about 1.6 GHz are necessary for a G4 to play it realtime, which explains why my 1.5 GHz G4 cannot do it.

@Andreas_Wolf

You might want to play with H264 loopfilter option if you happen to be a bit too short.