Amiga1 speed specs

[poweramiga2002]

gday can anyone let us poor people who are still awaiting our A1 g4/800 m/boards how fast the g3 ones are compared too the 603 boards please?

[KennyR]

As a VERY rough guide, a G3 at 600MHz is about 5-6 times faster than a 603e at 200MHz. But that's not all. Taking away old Amiga bottlenecks and adding a much faster memory bus, as well as eliminating context switches and having a PPC native OS, I'd *estimate* that an A1SE is about 10 times faster than a 200MHz BlizzardPPC board.

An A1XE with G4 at 800MHz could be expected to be 2-3 times faster than that.

[DaveP]

@Moderator

Can you move to hardware section pls?

@poweramiga2002
There are lots of online chip to chip comparisons for the PPC
comparing family increment to family increment.

[DaveP]

603 processor
   SPECint95 - 3.94
RS/6000 at 600Mhz
   SPECInt2000 - 310

Thats as good as it gets.

[DaveP]

Or you might like to try this very pretty graph  :-D

[Piru]

@ KennyR & DaveP:

So, I take it you didn't actually benchmark A1 at all, just pulling these values form a hat?

[DaveP]

@Piru

Read it properly please. I **said** that you can just go and compare the chips ratings for an idea. The figures are out of specbenches hat, not mine.

[Piru]

DaveP wrote:
603 processor
   SPECint95 - 3.94

What clockspeed? bus speed?

RS/6000 at 600Mhz
   SPECInt2000 - 310

Thats as good as it gets.

What is the point here? You can't compare SPECInt95 and SPECInt2000 results.

[DaveP]

1. Answered you question about where the figures were sourced from. The reason I put both SPEC *marks* in there is because that is as good as it gets for something as ancient as a 603 - people are not running 2000 against such old kit in lab conditions anymore. That is the point. Why did you think I kept what the source marks were in the post? I had assumed someone would ask the intelligent question "how do you compare them" but then if they are intelligent enough to ask that question they are intelligent enough to go find out precisely what they said above.

2. Ballpark figures are all that is required. In fact if you followed my advice and went to the spec site you would see exactly what the kit was. ;-)

The point being, dear Piru, that you cannot compare apples with apple strudel! The point being what exactly I think you were about to start berating me for! :-)

Now we could turn this into a navel gazing exercise but you know and I know that benchmarks are useless for anything other than ball park estimates and USENET rows with Steve G.

Next time I wont keep the unstated sarcasm out of my posts. Maybe then you will get what I was going on about with the apples ( 95 ) and oranges ( 2000 ) comparison and the "as good as it gets".

[Piru]

DaveP wrote:
I had assumed someone would ask the intelligent question "how do you compare them" but then if they are intelligent enough to ask that question they are intelligent enough to go find out precisely what they said above.

Great. What about the people who are not that intelligent, like mr Average who has no knowlege of SPECint? They deserve to get the false impression? You could have just been clear from the beginning you know...

And please, lets not get personal here, we're both adults (I suppose).

[DaveP]

@Piru

Thats why I added the pretty chart link because I realised it wasn't clear. Clearly even that was insufficiently clear.

I didn't mean to get personal, reading back through what I wrote does sound more aggressive than intent - trying to knock off a post like that just before a rushing off to a meeting - repent at leisure they say!

[Darth_X]

How does the MAI bug affect the results here?

[DaveP]

Darth_X wrote:
How does the MAI bug affect the results here?

It doesnt.

Firstly because the Articia-S "bug" is "fixed" in all production available boards ( or worked around )  

Secondly the measurements are mostly done in volatile parts of the system ( allthough spec-appserver intends to incorporate more persistence in the factor ).

Finally it does not effect any of the results I have posted because none of them are "MAI bug" based hardware results ( e.g. not Pegasos, Teron CX, PX etc. ).

[KennyR]

So, I take it you didn't actually benchmark A1 at all, just pulling these values form a hat?

Pretty much so, yes - didn't I make that clear enough by saying "VERY rough guide" and *estimate*? If not, then let me re-iterate: they are quasi-educated guesses. If you have better figures by all means share them with us.

[lempkee]

hmmm so ... with so many brain guys like its in here atm... is this the only answer you can give us??

i aint a tekkie at all, not within this anyway and i am really curious on how it would perform,
mainly because i have seen 600mzh G3 mac's in action (if thats the correct word) and i wasnt very impressed,
but hey i wasnt impressed when i saw em back at the 68k range either.

there must be someone who can give us abit more facts or?

and as we all know a 603 200mzh vs a 603 240mzh aint really much compared to what i expected anyway, but then again thats why i am so glad the 68k is finnaly removed from the boards, but hmmm i love 68k still...AAAAAAARGH

anyway what can we expect out of a 800mzh G4? (i have ordered one) , most of all i wonder if its good enough and i say this because i have never been impressed over the 601,602,603 series AT ALL , the 604 helped abit but still nothing impressive.

dont get me wrong, i love AMIGA and i like that it will come in a G4 form finally?) ..

cheers