One of Adam's posts makes it sound like the intention is for OS4. For anything else I wouldn't expect to see him talking about going to Amiga first and then Hyperion later if/when a license from Amiga can be had.

_________________
All glory to the Hypnotoad!

The intention may be for OS4, but as Ack has repeatedly avoided answering the question of whether or not a signed contract and/or license with Amiga Inc exists, one must surmise that it does not.

No license = no Amiga OS4

The only difference between this latest hardware announcement and the announcement about the ethereal Powervixxen is that Amiga Inc are publicising Ack's claims in this instance. We know that Amiga Inc currently has a vested interest in appearing to be interested in producing hardware, and has a long track record of obstructing the licensing of any actual hardware.

I would like to believe that this hardware will appear and will run an Amiga OS, but:

1) There is no evidence this hardware exists except as a specification;
2) The track record of the people involved in terms of undelivered promises;
3) The obvious lack of any signed contract with Amiga Inc and/or license to run Amiga OS.

Without a license this is not Amiga hardware any more than the Efika is. At least we know MorphOS is under development for the Efika.

Adam may be acting in good faith (that is another question entirely), however without a license or contract with Amiga Inc he is most likely to be just a patsy for Amiga Inc's short term need - ie they need to appear to have a vested financial interest in the ownership of the Amiga OS4 IP.

EDIT - I would very much like to be proved wrong.

Last edited by Boot_WB on 10-May-2007 at 11:58 PM.

_________________
Troll - n., A disenfranchised former potential customer who remains interested enough to stay informed and express critical opinions.
opp., the vast majority who voted silently with their feet.

Well you can run eg. 32bit x86 Linux on 64 bit hardware. You can run 32 bit OS's on 64 bit hardware, but not the other way around.

Ofcourse you don't get full power out of the the hardware, but I think it's better than nothing until there is a new 64bit OS version. Atleast hardware supports 64bit OS if there will be one someday.

Last edited by miksuh on 11-May-2007 at 06:48 AM.
Last edited by miksuh on 11-May-2007 at 12:30 AM.
Last edited by miksuh on 11-May-2007 at 12:30 AM.
Last edited by miksuh on 11-May-2007 at 12:22 AM.
Last edited by miksuh on 11-May-2007 at 12:18 AM.
Last edited by miksuh on 11-May-2007 at 12:12 AM.

Honestly, if OS5 was built on QnX Neutrino, I would be tickled. Throw an Amiga-like UI on it, and we have our next generation OS. That would be a win-win without Win.

Quote:

Just out of curiosity, how do you know that the parties resolved their differences while you were waiting?

In Massachusetts it's one day or one trial every three years and no one is exempted although you can reschedule your jury service to a different date. Even a lawyer or judge can end up on a jury. A judge comes in and welcomes you to the courthouse. You're shown a video on the process and what happens if you get selected. Sometimes they throw on a movie. There's a court officer who makes sure you don't just wander off, and he or she gives you a general outline of what's going on and tells you where to get lunch if that comes up. I've frequently been called on days when juvenile cases are on the docket, so the spiel might be along the lines of, "We've got ten juvenile cases today. There's a good chance that none of them will come to trial because what happens is that the kids don't come to grips with reality until they're in the courthouse and they know there's a potential jury waiting to be impaneled to hear their case. That's when they suddenly realize they'd better be paying attention to what their lawyers and the prosecutor have been telling them and accept whatever deal is on the table. If you guys weren't here, they'd just keep stonewalling."

Then as the morning goes on the officer tells you how many cases have been scratched and whether it looks as if any will actually go to trial. This last time they were hearing civil lawsuits that day and we were in the pool reserved for those cases. All but one was scratched by about 11 AM. Technically, the last one did start, but one side lost a pretrial motion to suppress a particular line of evidence. At that point, they abandoned the suit and settled with the other side, so no jury was needed. That we were told as we were being dismissed. To get slightly relevant to the thread, this same sort of thing could happen with Amiga Inc. v Hyperion Entertainment. A judge could cut the legs out from under one side's case by ruling some aspect of the purported evidence as irrelevant and inadmissable. I suspect that cases like these frequently turn on legal technicalities that a jury would not be expected to consider in its deliberations.

I did actually get as far as the courtroom on a different occasion, but the trial would have crossed the weekend and I had to be in Texas on Friday and California on Sunday, so the judge excused me. It was at that stage that they were questioning and challenging jurors. So, although jurors "disappear" after leaving the pool, they don't necessarily get on a jury. In the old days, I know that jurors who were not impaneled were returned to the pool, but that was before one day or one trial was in effect. Nowadays, if you're challenged or otherwise excused, you're done and get to go home.

Sometimes after all the business of the day has been completed, you'll get "invited" to one of the courtrooms where a judge will review what happened and thank you for your time and patience.

As far as TV and newspapers while in the jury pool, I've brought and read newspapers while in the pool. It's different once you're actually on a panel. My number doesn't come up again until 2009. That will be the last chance for me. The next time, I'll be over age.

HB

Last edited by Blot on 11-May-2007 at 03:21 AM.

I guess i'll have to keep an eye on all this

_________________
Helge K. Leaving the Amiga in favour of a PC..

Quote:

. A 64-bit address space requires 64-bit pointers, which take up twice the memory space of our traditional 32-bit pointers. So *existing* PPC & 68k apps cannot EVER handle a 64-bit address space (without a recompile). Neither can you mix 64-bit & 32-bit apps on a 64-bit AmigaOS4, because AmigaOS's design shares (64-bit) OS pointers to every app. Therefore the only way OS4 can support 64-bit is by loosing support for 68k apps, and requiring that all existing PPC apps be recompiled - which would put it in the same situation as AROS (and no-one wants to be in THAT situation).

That's absolutely irrelevant. Say the OS will still limit the address range to 32bit (what would you need 64bit addresses for in a shared mem space where you can't stick more then 4GB in any way ?). As long as you have say max 2GB RAM and no swap enabled, that should be sufficient. Well, sure, a large data mining app under OS4 would hardly run that way. But probably a patch to the compiler could help to force the address (void*) to be 32 instead of 64, that's it (should be a default compiler option, though). I don't know if that's possible, but I could imagine it might (would probably add mask op to every 64bit address op - but I hope not?). Are 32bit addresses still available in 64 bit mode ? Anybody ?

You can even go so far, that, later a per thread pointer is always 32bit, and a new API will provide full 64 addresses (would imply, taglist will never support a void* tag, ealow eahi). But, a full 64 bit version with the upper long set to 0 would still run and take advantage of the 64 system (at least regarding all integer ops, etc).

Quote:

2. SMP (apps running on multiple cores) requires that apps can communicate with the OS (and other apps). Which would be fine, except AmigaOS is built around the sharing of (pointers to OS) data structures & the implicit assumption that only one program can access them at any time - and that is patently untrue for a multi-core system (*). I'm a little less sure about this, but I don't think there's any way to modify OS4 without loosing all backwards compatibility (and thus all existing 68k & PPC apps).

Well, ever heard about Semaphores ? They have greatly been enhanced in OS4, BTW.
(BTW: others call them also CriticalLock or mutex ... all are similar to some extend).

The Blizzard/CS PPC had nothing to do with it. This is not a coherent SMP system. Every messages requires a full context switch on both CPUs, that means, you have to make sure your cache gets flushed on one side (68k or ppc), the (e,g,) messages body (the actuall memory) get written back to RAM, then a signal gets sent to the other cpu (ppc or 68k), which then requires to load the page back into its own cache (or reads it directly from RAM) and there you go. And that all on a lamer clock speed. Without cache the 68k and/or the PPC are pretty dead.

In that sense, this is not a real context switch, e.g. the 68k machine state is not shared with the ppc and one state does not get restored on the other cpu. Hence not SMP, and cannot be between different types of CPU. This would require coherent cores, which, guess what, multicore cpus do have.

Guess what, multicores usually also share caches between the cores and run with full clockspeed (in that case 2GHz or half). A message from one core to the other would be damn fast. I am pretty sure its doable. The OS will need some adaption, sure (e.g. all global refs to thread pointers etc. need to be resolved, some thought has to go into the scheduler), but it should be possible. (and no, mem protection is IMO not required to do so). BTW, multicores can restore one cpu state from one core to another (have to), thus a real context switch between cores.

Well, we'll see. First the HW has to be here any way, and then the two parties need to resolve their issues. All that is highly theoretical anyway.

_________________
AmigaOne XE G3/933/VIA/FM801/R200 (fixed),
G3SE/600/Voodoo3/Sil680/RTL8139/SBLive! (noiseless!)
µA1-MK2/G3/933/R200/CMI8738
XE/G4 (broken 7450/800)

@Sneaky who said Quote:

Well I don't know how, but it seems that in the PC World this seems not to be a big problem, as most 64bit Processors are run with a 32bit WinXP. Don't know though if the maximum memory border gets extended then (which, BTW, is the main reason for 64bit CPUs). And as far as I know, you don't have to get a different Office Version, if you're running it under the 64bit XP, which is available. So it has to be at least technically possible.

No, you did not read what I said carefully enough: "because AmigaOS's design shares (64-bit) OS pointers to every app". Which is to say that Amiga OS4 is completely different to modern OSes like WinXP - some might say it is archaic in design. This different/out-dated design is responsible for much of AmigaOS's speed, but it also makes going 64-bit impossible.

Quote:

At what cost, has to be answerd by someone else

I've already stated the cost: Loss of binary compatibility for all existing 68k & PPC apps, with the requirement that PPC apps must be specially (re)compiled for 64-bit.

_________________
Author of the PortablE programming language.
It is pitch black. You are likely to be eaten by a grue...

@Jorge
I think you missed my main point: AmigaOS is designed around the idea of communicating by sharing pointers to data structures. No copying is done, not even in principle, unlike Windows, Linux, etc. This means that every app must (basically) be able to access all the memory of every other app. And that is simply not possible for a 32-bit app trying to communicate with a 64-bit app or a 64-bit OS.

As to implementing SMP using Semaphores/Mutex/etc. Of course I heard of them. My point there was that I don't think this can be done in a backwards compatible manner (although I am willing to be proved wrong). If you suddenly require that every access to every OS data structure requires the use of a semphore, it will break all existing apps (esp 68k ones) which were not compiled with that in mind.

The only alternative is to create a whole NEW set of 64-bit/SMP supporting interfaces for OS4. Which is fine, EXCEPT that you are effectively sandboxing the old 32-bit/single-core apps from the new 64-bit/SMP ones. The end result of which will be that you are effectively running two OSes side-by-side; the old OS3/OS4 (interfaces) that we know & love, and a new fangled 64-bit/SMP OS (interfaces) that may not bear much resemblance to the old one. Doesn't that sound of a bit like MorphOS or Mac OS X? I have nothing against MorphOS or OS X, but a lot of people here do...

_________________
Author of the PortablE programming language.
It is pitch black. You are likely to be eaten by a grue...

Quote:

I think you missed my main point: AmigaOS is designed around the idea of communicating by sharing pointers to data structures. No copying is done, not even in principle, unlike Windows, Linux, etc. This means that every app must (basically) be able to access all the memory of every other app. And that is simply not possible for a 32-bit app trying to communicate with a 64-bit app or a 64-bit OS.

Ever heard about shared memory ? (Or MEMF_SHARED in OS4 to be more specific). Message passing can be achieved (and restricted) to a 32bit boundary and 32 bit pointers. Even for a 64bit application. That said, a MEMF_SHARED allocated mem would always return a (virtual?) 32 bit address. (BTW: AmigaOS < 4.x had MEMF_PUBLIC to achieve similar results).

Thread private mem can be whatever you want. You have to use MEMF_SHARED already if you want to do process to process communication.

I could, however, imagine, a full 64bit system which is not backwards compatible, might appear eventually. But what for if it does not have any performance advantages ?

Quote:

As to implementing SMP using Semaphores/Mutex/etc. Of course I heard of them. My point there was that I don't think this can be done in a backwards compatible manner (although I am willing to be proved wrong).

Semaphores are not rocket science. Why would an ObtainSemphore() all of a sudden not work any more, when everything the application sees is the function entry. Sure this can be blackbox implemented in a backwards compatible way (and most likely already is).
Why would that require a new API and a sandbox ?

Quote:

If you suddenly require that every access to every OS data structure requires the use of a semphore, it will break all existing apps (esp 68k ones) which were not compiled with that in mind.

This has nothing to do with multicores but is already the case with _multitasking_. Read-only elements will always be available, no matter if multicore or not (the OS might need to have special precautions, but as long as fields are read only, who cares).
It still no save to read values if they are getting updated on a multitasking system without using locks, no mater if one or two or n cores. There is a pretty good chance, if it works now, it will work also with multiple cores.

Quote:

The only alternative is to create a whole NEW set of 64-bit/SMP supporting interfaces for OS4

That's impossible. You would run two OSs, and practically emulate the 32bit system in a kind of VM. Either you use an extended API or you don't. But two, one for 32 and one for 64bit seamlessly available from an APP POV is useless (IMO). Some 64bit extensions to the current system is more likely (IMO).

But, that said, what do I know. I have a limited mind. There are other things out there which I have no clue about it. Same applies to the OS internals. Its just my perception. We'll see.

(BTW: we discussed a little bit about this in an older thread already)

_________________
AmigaOne XE G3/933/VIA/FM801/R200 (fixed),
G3SE/600/Voodoo3/Sil680/RTL8139/SBLive! (noiseless!)
µA1-MK2/G3/933/R200/CMI8738
XE/G4 (broken 7450/800)

@thread:

Re: SMP, memory protection,etc.

Everyone i hear complaining about putting all those features inside AmigaOS claims that it would break compatibility with older apps, (actual native ones aswell).
Wasn't the point of developing further the OS bringing to it the more modern features we all know from other operating systems that would make it more stable/robust/uptodate?.
I always seemed to think that backwards(68k) compatibility was only a time solution to the lack of native software and not meant to be kept forever.

Of course breaking of actual PPC applications is to be expected, but these apps can get (most of the times realistically)an update to match the OS requirements.

How much of legacy support would we need once the OS is mature enough and we have all those (in a ideal world) apps everyone(EDIT: everyone as in anyone who has his computing needs fulfilled with such machine) needs?.

Last edited by Ami603 on 11-May-2007 at 06:07 PM.

_________________
Cuida tus piedras gordas.

A1200/030 32Mb
A4000D
A1-X1000.

I don't think its that easy. Maybe with something which is not OS4 ? But then, why not just use an existing OS and sandbox AmigaOS into it. Oh, wait a minute, we have that already. Its called UAE.
Well, yes, could be a little bit more seamlessly integrated, but technically not much of a difference, except it then would have to run ppc code, too.

OTHO, getting HW like this gives at least a chance to try to evolve the OS into a more modern structure. There will always be legacy, I would guess. If not, I would simply switch the OS.

Quote:

Of course breaking of actual PPC applications is to be expected,

Well, maybe that could be a future, eventually. Say if MP is introduced, and 80% of PPC apps would still run or when they crash, the system doesn't go down, but would 100% (!) intercept the crash, well, that might be a reasonable trade off. But this will take a while, in the meantime, we could probably support a SMP with some extension or so...

_________________
AmigaOne XE G3/933/VIA/FM801/R200 (fixed),
G3SE/600/Voodoo3/Sil680/RTL8139/SBLive! (noiseless!)
µA1-MK2/G3/933/R200/CMI8738
XE/G4 (broken 7450/800)

Quote:

... (BTW: we discussed a little bit about this in an older thread already) ...

Must have been somwhere around here
Excellent to read, BTW

@ Ami603

Fully agree...

We must go FORWARD !

_________________
-------------------------------
AmigaOS, the last hope...

Can OS4 Bridge those two NICs? IE- Can have one NIC to PC and the other to Internet???

Hmmm???

_________________
If you're going through hell, keep going. -Winston Churchill

@debrun:

i'm bridging the internal A1 ethernet port to a wireless pci card to share internet with my laptop with no problems.

_________________
Cuida tus piedras gordas.

A1200/030 32Mb
A4000D
A1-X1000.