Petunia web site updated

[Opi-Poi]

The Petunia web site has been updated with new info and benchmarks.

http://amigos.amiga.hu/rachy/petunia.html

Source:AMIGAplus/GFX-BASE

[Elektro]

in-te-gra-te

thats an order...

 :-D

[System]

On my A1200 603e/240 060/50 it goes a litle bit faster then a 060/50.
So on a G3 it will fly like amithlon, i guess.

[ikir]

@Dirk-B

Yeah! :-)

[Elektro]

Can anyone tell whats the speed penalty because of the contex switching now?  :-o

[KennyR]

Can anyone tell whats the speed penalty because of the contex switching now?

Virtually none. Context switching on dual CPUs was due to the two CPUs fighting over the address space. With one CPU that's gone. It'll be just like amithlon, but of course better. :-)

Ok, software context switching slows it down a little too - about a ten-thousandth of the slowdown hardware switching does...maybe. How much slowdown there will be I don't think anyone knows. A truly tiny amount. It's a bone of contention - MOS say their way is faster, OS4 say their way (using the MMU) is faster. Only the benchmarks will prove it, and no amount of trolling and flaming will change that (I thought I'd add that before the FUD throwing started).

[Elektro]

That's what I meant. If you run this benchmark on a CS or Blizzard as compared to a1?

[HyperionMP]

We never claimed our MMU approach was faster, in fact it most likely is slower.

(Although the net-effect will not be very important.)

That's not the point however.

The idea behind the MMU approach is to leave the ABI/API unpolluted by the emulation technology.

Once the system is fully PPC native and all your apps and games are PPC native, you'd still be stuck with your legacy  baggage in the ABI/API.

This is called thinking ahead. We're not going to go for short-term fixes.

Another reason is that it obviates the need to insert emulation traps into the source-code.

[blubbe]

We never claimed our MMU approach was faster, in fact it most likely is slower.

(Although the net-effect will not be very important.)

That's not the point however.

The idea behind the MMU approach is to leave the ABI/API unpolluted by the emulation technology.

Once the system is fully PPC native and all your apps and games are PPC native, you'd still be stuck with your legacy baggage in the ABI/API.

This is called thinking ahead. We're not going to go for short-term fixes.

Another reason is that it obviates the need to insert emulation traps into the source-code.

Sorry, but the  BS-o-meter is about to burst..

What legacy baggage are you talking about ?
The emulation-traps ? Why would they still exist
(as baggage) in a system that is fully ppc-native ?
As you have to recompile to change an exe, the traps
would be gone anyway.

Instead you have legacy-supporting baggage *in the system* to support this. Somehow I think having baggage in transition-apps is better than baggage in the new system.

About MMU-method beeing faster.. I cannot remember anyone official saying this, but Im
not sure, as that has been something floating around as an argument to use it for quite some time.
Well, anyway, it wasnt true.

[Kronos]

@blubbe
"BS-o-meter" :roflmao:

@Ben

Do you really think that you will ever be able to replace all the
legacy-stuff  with PPC-native SW ?

How long did it take until someone could finally live without AA
after the GFX-cards were introduced ? How many developers were
active back than, and how much are now ? Why did Apple need years
for it, when they are 100 times bigger than anything "Amiga" ?

Ever thought about a career as stand-up-comedian ?  :ranting:  :flame:  :crazy:  :boohoo:

[Elektro]

As far as i can see most if not all components of OS are already PPC native.

And what have AGA games and proggs. got to do with the above? Seems to me you're confusing things all over. Modifying games and apps for use with RTG has nothing to do with emulation technique. AFA apple is concerned macos is macos & amigaos is amigaos. The comparison (except for the PPC) is invalid.

[Kronos]

@elektro

Ben has been writing about the elimination of the need for a 68k-emu,
which would only be possible if all SW that one needs is rewritten for PPC.
People who didn't want to use AA (or didn't have like Draco) faced the same
problem when they tried to find RTG-replacements for their AA-SW, and
Apple still has the 68k-emu in the system.

[Elektro]

What elimination of 68k emu? Thats why there is petunia.

And again I see no relevance with regards to AGA software you keep mentioning.

[Kronos]

Ben wrote:

Once the system is fully PPC native and all your apps and games are PPC native,

And I said that this will NEVER happen as their won't be enough new
SW by the few developers left to replace all the old 68k-SW.

The transition from AA (or OCS) dependant SW to RTG-compatible SW was
similar as it also needed lot of replacements for old SW. The years
this took (or Apples journey to PPC) serve as an perfect example of the
problems you will encounter before you could come to the point Ben
has been writing about.

[HyperionMP]

>Instead you have legacy-supporting baggage *in the >system* to support this. Somehow I think having >baggage in transition-apps is better than baggage in >the new system.

It's clear that neither you nor Kronos understand what I'm trying to say here because you lack the necessary background information.

The idea is exactly to avoid baggage in the system.

I'll leave it at that.