I plan to close the survey end of this weekend, and analyse the results.

Thanks to all who have filled in the survey already. If you haven't contributed yet, then here is your last chance. Click the following link:

https://keasigmadelta.com/store/surveys/amigaos-4-x-hardware-emulation/

Hans

Last edited by Hans on 16-May-2023 at 07:13 AM.

_________________
Join the Kea Campus - upgrade your skills; support my work; enjoy the Amiga corner.
https://keasigmadelta.com/ - see more of my work

@Hans

Done. Thanks for waiting. I had forgot until I saw the reminder.

@all
And the results are in...
https://keasigmadelta.com/blog/amigaos-4-x-hardware-and-emulation-survey-the-results

Hans

_________________
Join the Kea Campus - upgrade your skills; support my work; enjoy the Amiga corner.
https://keasigmadelta.com/ - see more of my work

@Hans

Interesting survey and results, I get the general feeling that more people would give OS4 a chance if it was on cheaper hardware.

I think we got to see the inconvenient truth that PPC is at a dead end / too expensive. I blame no one for this, when OS4 project got going PPC seem to make sense as it where since 1997, Amiga software and community had a toe in PPC market. Plus in 2001 when Amiga Inc & Hyperion decided to make AmigaOS for PPC, it seem like it had a good future, I think G4 PPC where doing good on Apple, all the console manufacturer where pouring development money into it for there next gen systems, now no one left pouring money into PPC development.

We are where we are, I think it good / maybe easiest way to maybe help the OS4 market is to improving emulation support for OS4.

I do think targeting Pi support for PPC emulation might be a good idea, I know Pi might seem underpowered to do PPC, but I did read it can do roughly G3 800Mhz speeds ( video of some one running it MacOS 9 on Pi3 https://www.youtube.com/watch?v=hwX-QMPqrRg ) Plus hopefully Pi5 not too far away.

With targeting Pi hardware we could maybe even make a Pistorm PPC edition for Classic Amiga.

_________________
AmigaNG, YouTube, LeaveReality Studio

@amigang

Incovient truth is Amiga is interesting because it is not a pc.
Amiga is hobby. Hobby don't have to be cheap but hobby have to be interesting.
Our community has cheap and fast pc solution (uae jit
for almost twent three years (since autumn 2000)
and nothing come from it.
Nobody sane will resign from use win/lnx/osx on pc or from ios/android on arm
to use os on win 98 level.
If You want to switch to pc work on szulc, karlos, szonwejs and other that want to switch from ppc to provide something decent.
Something that will be good enough to switch to win no often than one a week.
Somethign like Mac OS X which means Amiga gui and graphics on top of unix.
for now aros clowns are when they were in 1996 most important thing amiga gui clone open source is not done.

Last edited by ppcamiga1 on 20-May-2023 at 06:53 AM.

@All

What I've noticed is a lot of comments about OS4 crashing in emulation. Both in the survey and elsewhere. I don't know any more details about this. Sometimes an error is interpreted as a crash but may not be. On 68K a crash would end up in an alert or worse as a reset so don't know where this confusion comes from.

I know AmiUpdate can give an obscure error because of a broken package and save it in some hidden log. Updates lack sanity checks as people can upload broken packages that aren't rejected and sit there endlessly.

A problem AmigaOS4 has, due to it's inheritance from OS3, is that it is too sloppy. Like the example above, there are too many things broken, that remain unchecked and make it look bad. I think this also comes from a sloppy attitude to releasing software. OS4 is based on an old OS software model where you must manually check your software and it lacks automation. As simple as it is, compared to a modern OS, too much work is involved. On top of this is that OS4, like OS3, is just too technical for an end user and there hasn't been enough development to make is user friendly. I think this happens when all the devs are tech heads and don't seem to have any concept of user friendlessness or user empathy because that's hard to program and too much effort when the hard user way still works.

@Hypex

Quote:

What I've noticed is a lot of comments about OS4 crashing in emulation. Both in the survey and elsewhere. I don't know any more details about this. Sometimes an error is interpreted as a crash but may not be. On 68K a crash would end up in an alert or worse as a reset so don't know where this confusion comes from.

When I last tried AmigaOS 4 on UAE, it was crash prone. It would crash with software that is stable on real hardware.

Quote:

A problem AmigaOS4 has, due to it's inheritance from OS3, is that it is too sloppy. Like the example above, there are too many things broken, that remain unchecked and make it look bad. I think this also comes from a sloppy attitude to releasing software. OS4 is based on an old OS software model where you must manually check your software and it lacks automation. As simple as it is, compared to a modern OS, too much work is involved.

True, automated tests would do a lot to make releases more stable.

Quote:

On top of this is that OS4, like OS3, is just too technical for an end user and there hasn't been enough development to make is user friendly. I think this happens when all the devs are tech heads and don't seem to have any concept of user friendlessness or user empathy because that's hard to program and too much effort when the hard user way still works.

I don't think so. AmigaOS was designed to be user friendly, but with plenty of space for power users to tinker with it. At least, the OS itself was. Things got less user friendly when we started relying more and more on *nix ports (e.g., samba, smbfs, etc.). The additional work needed to make it easy to use often never got done.

IMHO, OS4 itself is easy to set up and use. A lot of the software you can install on it, not so much.

Hans

_________________
Join the Kea Campus - upgrade your skills; support my work; enjoy the Amiga corner.
https://keasigmadelta.com/ - see more of my work

@Hans

Quote:

When I last tried AmigaOS 4 on UAE, it was crash prone. It would crash with software that is stable on real hardware.

Long ago? From my experience, software that crashes in emulation crashes also on Pegasos 2.

Quote:

IMHO, OS4 itself is easy to set up and use.

For us, yes. Others need point by point guides to even install our OS (both for WinUAE and Pegasos 2/Qemu). Some pecularities (SLB/boot partition, what filesystem to use) are really confusing for new users.

I agree. Amiga Os 4 is what it is. It is just better Amiga Os 3.
It is something like Mac Os 9 from Apple.
Amiga Os 4 has all drawbacks of Amiga Os 3 because it has binary compatibility
and allows integration with old 68k Amiga software.
Thats ok as long as we use powerpc.
Powerpc work in 32 bit big endian mode like 68k.
ppc and 68k apps exchange data in the same format
and everything work like thirty years and more ago only better because many times faster.


Switch to little endian cpu breaks binary compatibility and integration.
First thing that should be done after breaking binary compatibility should be add memory protection and unix compatibility.
It was not done by aros clowns.
After so many years they have still nothing worth of use on x86 and arm.
Still no reason to switch to x86 or arm.

I don't want to sugarcoat anything, but as a tester of Qemu/Peg2 I think I know what I'm writing about.

First of all, a switch to ARM might be a solution to the current unavailable hardware problem, but there would not be enough resources to do it again and it would take a lot of effort.

Again briefly to the configuration of Qemu/Peg2....

We had some problems in the beginning e.g. the default CPU settings of Qemu was wrong, which caused many problems. SmartFileSystem was not usable, Altivec was not usable and the CPU was not recognized by AmigaOs4.1. This led to many incompatibilities.

With the right configuration these problems were already solved and improved in future versions of Qemu8.x. It was detected by the usage and feedback of the Amiga community.

Qemu/Peg2 is already on some machines and despite the many limitations faster than on real hardware and how fast the development of new hardware progresses everybody knows.

In my opinion you should build on this and improve this emulation, Qemu is not only an emulation, because some parts are translated natively from the host system and this makes it of course very pleasant to use.

I don't think that newcomers invest in expensive hardware (PPC) for an OS they don't know and which also has many limitations.

An adaptation of AmigaOs4.1 to Mac mini G4 could also be a solution, I would buy such a hardware with AmigaOs4.1 immediately.

There was at that time the "mona loader" what failed, why actually? That would really interest me.

@Maijestro

So you’re saying OS4 isn’t portable code? I thought that was one of the major points of the transition from 68k to PowerPC.

Last edited by kolla on 21-May-2023 at 07:05 PM.

_________________
B5D6A1D019D5D45BCC56F4782AC220D8B3E2A6CC

@kolla

Oh well C code is always portable, precompiled binaries not so much, retaining backward compatibility certainly would be a challenge, I can't say for sure that we will need power PC binaries to run on the ARM architecture, well 68,000 codes certainly would be an advantage to support. I know many who want to drop 68000 supports, because they want SMP and better memory protection. It certainly would make it easier to ignore compatibility, but without software who would use it? Certainly, making another AROS makes no sense.

_________________
http://lifeofliveforit.blogspot.no/
Facebook::LiveForIt Software for AmigaOS

@Maijestro

I agree, half step, similar to how macros 7 had primitive support for power pc, but had direct access to the hardware (GFX/Sound/floppy/cdrom) certainly, allow for low price hardware, and availability that we do not have today.

_________________
http://lifeofliveforit.blogspot.no/
Facebook::LiveForIt Software for AmigaOS

@Hans

Quote:

When I last tried AmigaOS 4 on UAE, it was crash prone. It would crash with software that is stable on real hardware.

It could be in the emulator then and not OS4 itself in that case.

Quote:

True, automated tests would do a lot to make releases more stable.

Given all that's involved I wonder how enormous a task it would be to automate it. On systems built with this this is mind it's less of a deal. But on OS4 it would need some automater to be built to test it all and that would take time to write.

Quote:

I don't think so. AmigaOS was designed to be user friendly, but with plenty of space for power users to tinker with it. At least, the OS itself was. Things got less user friendly when we started relying more and more on *nix ports (e.g., samba, smbfs, etc.). The additional work needed to make it easy to use often never got done.

It's user friendly from a 90's perspective. For example, plug in a printer, then load up printer settings and chose a driver. But that's backwards to today where things work the opposite and automated. Plug in a printer (that's turned on), the printer is detected and it locates the driver to set it up for you. But Amiga also lost it when it depended too much on CLI. So many things had a GUI but some things lack it. One basic example is drives on the system, as they can be hard to list without a system standard program, and then you end up typing Info into a CLI.

Quote:

IMHO, OS4 itself is easy to set up and use. A lot of the software you can install on it, not so much.

It's only easy if you know the quirks and are used to customising it all. An example is setting up a boot partition for OS4 such as for a classic system. That can be done under emulation. It doesn't setup any nominal defaults. When you chose a boot volume it can default to FFS3 or something. So the user just selects this then later the system complains or breaks in some way. If it needs FFS7 it shouldn't have picked a lower FFS itself. In fact anything but FFS7 should be banned in the list. You are expected to change a lot of things that should be preset. Even seasoned Amiga users can get caught out. But it's based on the old 80's HDToolbox where you needed to be an "Advanced User" to install a HDD. (And still do.)

Contrasting to Linux, though it can be quirky, you select filesystem and then a purpose for partition. So it has some defaults and can be customised later. Even if text based it's also easier to set a partition size than in MTB trying drag a slider around and being unable to scale it to the next GB or even enter it directly. AROS an MorphOS also have their own ideas on setting up boot volume.

@pavlor

Quote:

Long ago? From my experience, software that crashes in emulation crashes also on Pegasos 2.

It was years ago. IIRC, even GCC was crash prone.

Quote:

For us, yes. Others need point by point guides to even install our OS (both for WinUAE and Pegasos 2/Qemu). Some pecularities (SLB/boot partition, what filesystem to use) are really confusing for new users.

Ugh! I forgot about the partitioning bit. Shows you how long it's been since I last installed from scratch. I remember advocating for giving people an easier partitioning tool for OS installation...

Hans

_________________
Join the Kea Campus - upgrade your skills; support my work; enjoy the Amiga corner.
https://keasigmadelta.com/ - see more of my work

@Hypex

Quote:

Given all that's involved I wonder how enormous a task it would be to automate it. On systems built with this this is mind it's less of a deal. But on OS4 it would need some automater to be built to test it all and that would take time to write.

Start with automating just one module...

Hans

_________________
Join the Kea Campus - upgrade your skills; support my work; enjoy the Amiga corner.
https://keasigmadelta.com/ - see more of my work

@kolla

Quote:

So you’re saying OS4 isn’t portable code? I thought that was one of the major points of the transition from 68k to PowerPC.

It's not fully, still only partially. There's still code in there that is 68K such as ARexx, that needs a new license for a native port, if it could still be licensed from IBM.. Some other code like XAD unpacking libs or what one they have is in 68K. So they need 68k emulation for some functionality. 68k emulation is cheap these days but they still need a native emulation module for any host CPU as well as a JIT engine for same CPU, which tends to be more involved and written in assembler. Then, there are internal ties to PPC with ABI and low level functions for interrupts. Plus they still use the old 68K structures as base structures for the OS which aren't suitable for PPC even. No issue for x86 where it's built off a CISC design with dynamic alignment. Oh yeah, it's still stuck in big endian, so without conversion it's only portable to another big endian CPU currently if at all.

Also, each kernel needs to compiled specifically for each PPC board. This is common as the Mac had the same issue, where each OS had to exactly match the machine. This is a PowerPCism and it's kinda annoying as each PPC is incompatible with another. Against x86 where you could boot a generic Windows disk on pretty much most x86 machines which were more divergent.

I know the kind of thing Maijestro is talking about. There are quirks in the hardware of OS4 machines. The Sam series for example. I compiled and run ub2lb code targeting Sam hardware on my XE. It had a bad habit of crashing. But at times it would be fine and other times it crashed. It tended to depend on cold boot or reboot. I trace it down to a routine that read some device blocks but passed a NULL handle. According to the API docs it must pass a pointer for device handle. This would explain the crash but not why it works sometimes. Somehow, on the Sam, it always works, even though the code itself is bad. I recall they ran into this when configuring Sam emulation for QEMU because they were also getting a crash. Somehow, a hidden hardware quirk of the Sam allows poorly written code that passes NULL pointers to run transparently in firmware, that on other nominal hardware like me XE will crash as expected.

Last edited by Hypex on 24-May-2023 at 01:08 PM.
Last edited by Hypex on 24-May-2023 at 01:04 PM.

@Hypex

IBM didn’t create ARexx, have you never read the copyright notice when you start RexxMast? :)

Quote:

ARexx Version 1.15 Copyright © 1987 by William S. Hawes All Rights Reserved

_________________
B5D6A1D019D5D45BCC56F4782AC220D8B3E2A6CC

@NutsAboutAmiga

I think it would still need PPC compatibility. There's apps out there in PPC that won't all be recompiled. Just like old 68K apps. And they would still need 68K compatibility but that shouldn't be a problem now days. However, 68K compatibility wouldn't exactly be holding he OS back. After all, it's all emulated so should be a moot point. But if any software relies on 68K OS objects, since the OS design does support this, they will just need to put a middle man 68K library interface in place. The main problem is OS4 is built off and uses the same 68K designs, so apps need to peek and poke on system objects, and they didn't exactly replace the access with a clean room implementation for PPC so got stuck. Especially the big one being Forbid/Permit since a global lock is put in place to protect local data.

@Hans

Hardware for AmigaOS is only rasperrypi ...
Have it as native will made AmigaOS 4 (if smp) probably one best buy from the pi shop.
Good alternative compared os3 for classic machine that have PiStorm as accelerators.
This is what i think in my point of view.

_________________
I love Amiga and new hope by AmigaNG
A 500 + ; CDTV; CD32;
PowerMac G5 Quad 8GB,SSD,SSHD,7800gtx,Radeon R5 230 2GB;
MacBook Pro Retina I7 2.3ghz;
#nomorea-eoninmyhome