It is primarily for backward compability, but also some code (e.g. Hooks) must get adapted to compile correctly under a different endianess than it was originally intended for.
For backward compability (68k and ppc) I think it can be concentrated to the following question: what happens if some program calls 0x00000004, does it get Exec base or does it not? If not it will probably crash.

Addendum:

> I'm still not sure about the true big-endian capabilities of current ARM processors.

Olaf "Olsen" Barthel seems to be rather optimistic about it:

"AmigaOS [...] is hardwired to a big-endian host platform, not just the fundamentals but also its data structures. I would say that the chances to see AmigaOS run on an ARM are much higher than to see it run on an x86 family processor."
http://obligement.free.fr/articles_traduction/itwbarthel_en.php

Edit: Some more from him:

"the sacrifices that would have to be made to get AmigaOS working on the x86 platform would be greater than by comparison with the ARM, because you can have it load/store data from/to memory in either big endian or little endian fashion, depending upon how the chip is configured."
http://amigaworld.net/modules/news/article.php?storyid=5678#76241

[ Edited by Andreas_Wolf on 2010/10/14 3:28 ]

Addendum:

>> I'm still not sure about the true big-endian capabilities of current ARM processors.

> Olaf "Olsen" Barthel seems to be rather optimistic about it [...]

Nicholas "minator" Blachford, who works for ARM Ltd., is on the same track, at least regarding ARMv7-A:

"How well will it do on a 2.5GHz A9 in big endian mode?"
http://moobunny.dreamhosters.com/cgi/mbmessage.pl/amiga/179223.shtml

Wonderful quotes as always, Andreas. Particularly, that moobunny post by Nicholas is devastating.

I'd like t point out that theres much variety in ARM processors, being several instruction sets, superscalar and not, etc. Perhaps some obvious literature here would help:

http://en.wikipedia.org/wiki/ARM_architecture#ARM_cores

Too many candies in the box, I'd say. So, it's not that easy as "port it to ARM". That "ARM" is a very, very broad thing. Not to talk about on-die peripherals that chip makers add on their own...

> Too many candies in the box, I'd say. So, it's not that easy as
> "port it to ARM". That "ARM" is a very, very broad thing.

I think it's not too complicated really. The only current ARM architecture that would be reasonable for MorphOS is ARMv7-A. There you have ARM's own implementation Cortex-A (with A8 and A9 (and soon A15)) as well as architecture licensees' implementations like Qualcomm's Scorpion and Marvell's Sheeva PJ4.

Quote:

Yes, the "cortex" bunch. Those are superscalar, Quite some favourites are there (OMAPs, Snapdragons...).

Quote:

My brother, who has experience building computerrs with ARM processors (the humble ones, I'd add), told me that Marvell's original offerings were not strictly ARM cores, but something that they bought from Intel (XScale) that happens to be compatible, although it's not an ARM license. He also told me how hard is working with Marvell, they have zero interest in small parties, and hide documentation. In the end, he had to go for a cheap and accesible Samsung chip.

> Yes, the "cortex" bunch.

Yes, but not all of them. Cortex-M and Cortex-R are not ARMv7-A. Cortex-M0 and Cortex-M1 are not even ARMv7 but ARMv6. That's why I specifically said Cortex-A.

> Those are superscalar

Yes, Cortex-A is. The other Cortex cores are not.

> Quite some favourites are there ([...] Snapdragons...).

Wikipedia is wrong there. Snapdragon uses the Scorpion core (mentioned in posting before), which is *not* an ARM Cortex but Qualcomm's own ARMv7-A implementation.

> My brother [...] told me that Marvell's original offerings were not strictly ARM cores,
> but something that they bought from Intel (XScale) that happens to be compatible,
> although it's not an ARM license.

That sounds a little confused ;-) First and foremost you have to understand the difference between licensing an ARM ISA version and licensing an actual ARM core (which implements a certain ARM ISA version). Marvell had been an ARMv5 ISA licensee even before they got XScale from Intel. That was when they had their Feroceon line of CPUs which they got together with the ARM ISA license by acquiring the company Asica in 2003. Later in 2006, they got another ARMv5 ISA implementation in the form of XScale. They merged Feroceon and XScale into the Sheeva PJ1 core, still implementing ARMv5 ISA of course. Then, Marvell licensed the ARMv6/v7 ISA and implemented their Sheeva PJ4 core from that.
In that sense, Marvell never had "strictly ARM cores" to offer. All their ARM cores to date have been own implementations of some licensed ARM ISA version, not actual cores licensed from ARM Ltd. But on the other hand, that'd be like saying that the PA6T is not "strictly Power Architecture core" because P.A.Semi created that from an architectural license as opposed to licensing an actual Power Architecture core from IBM like Applied Micro, LSI and others have done.
Like I said, the only ARMv7-A implementations I'm aware of right now are ARM Ltd's Cortex-A, Qualcomm's Scorpion and Marvell's Sheeva PJ4. If anyone knows some more I'd be glad to know :-)

Quote:

Hummm... Probably. Perhaps I remembered my brother's stories badly. I've asked him to check your answer, to see where I was wrong.

So this Ti OMAP? 4 based system might work?

http://multivu.prnewswire.com/mnr/pandaboard/39158/

How do we find out if a particular ARM processor can use big endeia addressing?

> So this Ti OMAP? 4 based system might work?

According to Olsen and minator, yes.

> How do we find out if a particular ARM processor can use big endeia addressing?

*True* big endian addressing is of relevance here. Other than that I'd like to have this question answered myself :-)

OK, that's a fair answer.

I'm not 100% convince by jcmarcos and others pointing to ARM. It just that the pricing of the hardware does make it attractive.

ARM doesn't look fast enough to handle PPC code via JIT emulation though. That would leave us with the other (less palatable) idea of having a version of MorphOS that require recompilation of program code for existing programs.

Then there's ARM @ 1Ghz vs. PPC @2Ghz and abiove to consider.

Update:

> I did some search for the "Cherrypal developed" CherryPad America's looks
> and specs and found this: [...] ...with the difference that it's said to have a
> Telechips processor instead of the CherryPad's alleged Samsung processor.

Surprise, surprise. The CherryPad's alleged Samsung processor turned into a Telechips processor by pure magic:

"Processor: Telechips ARM11 800 MHz (CPU + GPU + DSP) system-on-chip"
http://www.cherrypal.com/c515.htm

"CherryPad is powered by the fast Telechip ARM11 processor (CPU + GPU + DSP, OpenGL 2.0) 800 MHz processor"
http://www.cherrypal.com/secure/product_info.php?products_id=16

This fact combined with the "slipped" original model number led me to this:

http://www.cadof.com/item/7965696188.html (Google translation, 908 yuan is about 137 USD at the moment)
http://www.uupad.cn/thread-3108-1-1.html (Google translation)

Great looking vaporware, Andreas. A 7 inch Android tablet, cheap? Sign me up! If one ever appears.

Do we have confirmation that Cherrypal has shipped anything?

> Great looking vaporware

No, the CherryPad C515 is real, but it's just not "Cherrypal developed". That's what I'm about.

> A 7 inch Android tablet, cheap? Sign me up! If one ever appears.

The market is really infested by those cheap Chinese tablet computers running Android. For instance have a look at this 400 MHz ARM9 based $100 one:

http://www.engadget.com/2010/10/27/walgreens-now-selling-100-maylong-m-150-android-tablet/
http://www.maylonggroup.com/M-150_MID.htm
http://www.youtube.com/watch?v=fu578pZVZm0

> Do we have confirmation that Cherrypal has shipped anything?

They say they started shipping the CherryPad C515 two weeks ago. See reviews and unboxing video:
https://morph.zone/modules/newbb_plus/viewtopic.php?topic_id=6726&forum=11&post_id=77311#77311

They also have an open discussion forum on the CherryPad C515 including reviews:
http://www.cherrypal.com/support/index.php?board=2.0

And reviews in the store:
http://www.cherrypal.com/secure/product_reviews.php?products_id=16

[ Edited by Andreas_Wolf on 2010/10/30 18:34 ]

Thanks for the references. The tablets are a neat idea. Certainly better than lugging around a iPad. I'm going to look into one.

>> http://www.mobileread.com/forums/showthread.php?p=872128#post872128
>> http://www.mobileread.com/forums/showthread.php?p=872134#post872134
>> http://www.mobileread.com/forums/showthread.php?p=882112#post882112

> Ouch... Velcro, are you on a multi-site quest to destroy your credibility?

Checking back those links it seems he's trying to get his credibility back ;-)

He also did an interesting review of the Cherrypal Asia C118:

"Moving on to the OS and software, I would say those are not as nice as the hardware. The desktop feels a bit awkward and unpolished. Despite the quick boot, it is a slow environment. The programs run slow. Cherrypal says "slow and sufficient." Having used it, I disagree about the second term. This OS and software are not sufficent on this hardware, at least not yet. I guess I would overall characterize the experience as just barely functional in some regards (browsing, connectivity, sound, a version of Android market, camera), dysfunctional in others (USB connectivity, I did not test everything there may be more dysfunction), and terribly slow overall. [...] For now, the Cherrypal Asia is a low-cost curiosity of great slowness and very limited functioning."
http://www.androidforums.com/android-lounge/204857-review-cherrypal-asia-netbook-running-android.html#post1785751

> Thanks for the references.

You're welcome. And just to show that I'm not a Cherrypal fanboy I'll also link to these:

http://www.crunchgear.com/2010/10/26/cherrypad-orders-unfulfilled-customers-antsy/
http://www.crunchgear.com/2010/10/29/cherrypad-review-is-less-than-enthusiastic/
http://www.intomobile.com/2010/10/26/cherrypad-android-tablets-still-not-scam/
http://www.the-digital-reader.com/2010/10/26/and-the-cherrypal-scam-continues/
http://liliputing.com/2010/10/cherrypal-188-tablet-now-shipping-maybe-gets-hands-on-treatment.html
http://www.techeye.net/hardware/cheap-tablet-flogger-cherrypal-accused-of-upsetting-customers
http://tech.massivelinks.com/index.php/tech/article/cherrypad-review-188-slate-thats-too-good-to-be-true/

And something not directly related to the CherryPad:

http://www.the-digital-reader.com/2010/10/20/cherrypals-max-seybold-is-so-bad/

Ouch! I guess Christmas isn't coming early this year. It's frightening how easily I fell for those specs. Has the tablet actually been seen in the wild or has Cherrypal completely scammed us on this one?


And is it just my perverse sense of humor or does Cherrypal's logo look like a pair of testicles?

> Has the tablet actually been seen in the wild or has Cherrypal
> completely scammed us on this one?

That depends on whether you regard the reviews and unboxing video I linked to as "in the wild" or not, I guess.

> does Cherrypal's logo look like a pair of testicles?

That joke is as old as the Cherrypal logo ;-)

Quote:

I wasn't trying to make a joke, I was just beginning to think my perception was off. Nice to know this is a common mistake. I guess you could add a third "cherry" and it would look like a pawn brokers shop sign.

-edit: wrong thread-

[ Editiert durch Zylesea an 2011/1/7 2:24 ]

Update:

> Surprise, surprise. The CherryPad's alleged Samsung processor turned
> into a Telechips processor by pure magic:
> "Processor: Telechips ARM11 800 MHz (CPU + GPU + DSP) system-on-chip"
> http://www.cherrypal.com/c515.htm

Did the 800 MHz turn into 500 and the 2 GB Flash into 512 MB?

http://www.cherrypal.com/support/index.php?topic=1364

I have a sense of déjà vu:

https://morph.zone/modules/newbb_plus/viewtopic.php?topic_id=6908&forum=11
https://morph.zone/modules/newbb_plus/viewtopic.php?topic_id=6976&forum=11&post_id=71732#71732

"We all bought turds."

Well if that isn't enough to make you shy away from Cherrypal...

Update:

> Did the 800 MHz turn into 500 and the 2 GB Flash into 512 MB?
> http://www.cherrypal.com/support/index.php?topic=1364

What could you do best when your recent cheap product doesn't meet the claimed specs? Exactly, you plan to sell a more expensive better spec'd successor:

http://www.mobilemag.com/2011/01/18/exclusive-cherrypads-next-7-inch-android-tablet-will-have-gsm/

|||





[ Edited by Velcro_SP 20.04.2011 - 06:25 ]