One unified OS for the future?

[spirantho]

Spirantho is wrong. If they are selling beta OS to consumers that is the product then they are selling beta OS to consumer. Beta OS is what you get.

Besides, there is no hidden information about Pegasos 2. Its hardware is documented and it is available to developers.

My point was that comparing a mature OS of several years with a beta version of another OS isn't a fair benchmark of speed. I don't think I was wrong.

Plus there are some very low level bits of documentation that the AmigaOS devs did not have access to, which the MorphOS devs did.

Anyway, this is now off-topic, this thread isn't a "my OS can beat up your OS" thread.

[spirantho]

Yes, the Sam460 will be a much fairer comparison than the Peg 2.

I believe that bPlan kept some of the really low level stuff to themselves. Just did a quick Google and found this, for example:

I think RAM timing is kept solid for biggest compatibility, so there
will be room for improvement, but I don't know how, and bplan
won't tell us...

http://morph.zone/modules/newbb_plus/viewtopic.php?forum=11&topic_id=4548&post_id=42844&viewmode=flat&sortorder=0&showonepost=1

That thread is just talking about twiddling the timings, nothing to do with AmigaOS - but the important thing is that it illustrates that bPlan did keep that sort of info to themselves.

I don't say that MorphOS isn't faster, only that the Pegasos II doesn't give a fair indication (the Sam 460 will), and that a benchmark that compares speed of a mature OS with a beta pre-release is completely meaningless.

[spirantho]

@itix

As long as the morphos devs have full access to the hardware details, then once MorphOs has reached a mature state on the 460, then I suspect we'll find there's very little difference. Of course before that state is true, then yes, MorphOS may well appear slower and that wouldn't be a fair comparison either. Look at G5 benchmarks on MorphOS now compared to their first beta releases, there's quite a difference.

@everblue


There was a link above to an OSnews article that compared non-optimal beta version of OS 4.1 on a Pegasos II with a mature MorphOS version and said "oh look! Morphos is faster, therefore it is better".  
To make a true comparison (as if that would help anybody), you need to compare a fully open hardware platform running mature versions of both of the OSes.

[spirantho]

Yes, MorphOS and AmigaOS both have their own strengths and weaknesses - there will always be some variance, but overall I don't think there'll be much difference.

As for the review being fair - yes, it was a fair indication of the state of AmigaOS 4.1 on the Pegasos II at that time, but in no way should it be used - several years later - to say that because of that AmigaOS as an OS is slower than MorphOS as an OS. All the review proves is that a pre-release beta of AmigaOS is slower than the full release of MorphOS at that time on that hardware.

[spirantho]

Funny that you claim Hyperion was selling pre-release beta to their customers. They never advertised it as such.

The OSNews article was done on the beta of AmigaOS 4.1 for the Pegasos II:

It is important to note, though, that AmigaOs 4.1 for the Pegasos II is still in its bets stage, so things can certainly improve once it moves out of beta

But yes, it nly reviews state of affairs in year 2009. Today if AmigaOS 4.1 FE (not released yet) was compared to MorphOS 3.7 (released) it could be even greater victory for MorphOS... ;-) I mean, it is also possibility, isnt it?

Absolutely, it could be. But I still say we need to wait until we have a fully open platform and a mature version of each OS on that platform before we can possibly make any decisions.

To me personally, I don't care which is faster - but I just get annoyed with people saying x is better than y and giving totally misleading 'evidence' such as that OSNews article, which - by its own admission - was not representative of the full story.

[spirantho]

@itix
Fair enough (though it doesn't say much for the quality of the article if they don't even know what version of the OS it's running).

I still say that the only fair comparison will be when MorphOS and AmigaOS have both reached a stable, mature build on a completely open board like the Sam460 though.

[spirantho]

The problem is that the Classic would need so many upgrades to come close to the performance of even a Sam440, that there'd be no Classic left! You'd need to upgrade everything, and the actual machine would be little more than a keyboard and disk drive dongle.

[spirantho]

@OlafS3

You mention something which would be very useful, actually.
I can see that for many parts, there are similarities rather than differences, but for some bits such as the PCI and USB that you mention, the APIs are quite different. If they were standardised across platforms it would make driver development much easier across platforms. Even graphics APIs are separating more and more now.

We should celebrate the differences in the underlying OS between MorphOS, AmigaOS and AROS, while striving to keep software flowing on all the OSes... but as the APIs move away from each other this will get harder and harder.

Ideally I would like there to be an independent "Amiga-like" council which would define APIs - each OS would contribute submissions via RFCs to the multi-platform council which would then ratify or deny that submission before it became standard.

This could never happen, though, because each of the OS owners would say "Why should we? We're a different OS - we're not responsible for the other OSes. We'll do what we like, thanks".... plus in my experience there would be precious little agreement about how the APIs should work - but it would be nice.

[spirantho]

Spirantho mentioned that he perceives an "End of the Road" for MorphOS development on PPC hardware, when they have completed support for the last few Mac PPC hardware choices that are not currently supported.

That is funny statement.

Just to clarify - what I meant is that once all the Mac PPC models are supported, they're going to be stuck with increasingly ageing hardware which is going to get more and more obsolete. They're going to need to change to another ISA such as x86 or ARM if they're going to stay at all current - I don't think there can be any argument to that.
If they don't, then in the future AmigaOS will be using multi-GHz multi-core low-power brand new hardware, AROS will be using brand new low-cost x86 hardware, and MorphOS will be stuck to ancient Mac Behemoths without any support for the current technologies - that would be a bad thing for MorphOS.

[spirantho]

@OlafS3

The other problem with abstraction layers like that (or even bytecodes like .NET) is that of speed. We will always suffer from a lack of speed compared to Windows machines, and as programs get heavier and heavier in their resources it's going to become more and more impractical. We need to grab as much power out of our "Amiga" machines (whatever they are) - the primary advantage we have is that we have much more lightweight operating systems. If we go the way of bytecode and abstractions we lose that strength entirely.

Abstraction is great for porting software, but useless for ekeing the most power out of the hardware, which is especially necessary for Amiga-like systems (particularly so when porting MAME, trust me )

[spirantho]

It depends on what you want to do.
If you want to encode (or decode) video, play MAME, that sort of thing, then using bytecode would be impossible.
I know what you mean, though - for less demanding applications, high level programming languages like C# or Java can speed up the writing massively - but as usual different applications have different requirements, so we can't force people to use high-level code as it could be disastrous (especially for things like device drivers).

[spirantho]

Bytecode is not solution to everything but you could write a MP3 encoder in C# and use encoder libraries written in C.

Yes, that's a good description of when to use bytecode (final app) and when not to (the guts of it) - and illustrates how we need both quite well.

[spirantho]

I think the difficulty that's being faced is that AmigaOS has to change the API, whereas MorphOS doesn't. AmigaOS's long term goal is SMP and memory protection. That's a heck of an undertaking, and where a load of work has gone already (that's not been noticed because it's invisible to the end user). As MorphOS has been concentrating on single-core hardware, I haven't seen it making the groundwork towards SMP - which would require changing the API (like AmigaOS has done). SMP will never work on AmigaOS unless the API-changing groundwork is done first.

If and when MorphOS starts work on SMP (have they done so already? I really don't know) they too will have to change the API somewhat, but I just wish it would be with changes that are compatible with AmigaOS's API changes. Not going to happen, though, I suspect....

[spirantho]

The AmigaOS devs aren't stupid. They don't just change and break APIs for no reason - it's far too much work to do unless they need to.
As a maintainer of an OS, the first thing you need to do is to take the code you're given (in this case a patchwork of M68K and PPC 10 year old OS3 code). Then you need to clean it up and lay the groundwork for the future of the OS (in this case SMP and memory protection). Only then can you actually start on the new technologies themselves. That's exactly what Hyperion have been doing.

Already we're seeing some of the fruits of their labour (e.g. the Extended Memory system for >2GB RAM which is one of the main benefits of 64-bit architectures). This sort of thing would have been much harder without the API changes.

I think many of the misconceptions surrounding Hyperion developers is that they're idiots. They're not. There are reasons why they do what they do (but they're not always obvious to the end user). They don't create work for themselves just for the heck of it. It's not in their interests to do nothing except break APIs - they need to keep pushing the OS forward otherwise they'll never sell another copy, and they're well aware of this.

[spirantho]

I wasn't referring to you, particularly, more to the whole feeling I get from other people on forums like this. Usually the same people who accuse AmigaKit of making lots of money on the X1000, as though they're all driving Ferraris or something (but that's a different argument for another day ).
SMP has to be transparent, otherwise it wouldn't be SMP. The difficulty will be getting SMP working without breaking binary compatibility, as you say. I suspect the solution would be to have "SMP-enabled" binaries and "legacy binaries" which can only run on the one core. I don't really know, though, I'm just guessing there.