@Everyone

Hi

I'm not so experienced with porting, but I will try to look at it... actually, I already started :)

Now I have problem with porting scons, there is no GNU makefile for Godot so I will probably need to build it manually or whatever...

I will keep You informed

Quote:

KimmoK wrote: "Haiku has some posix support" That explains a lot.

Some of the POSIX standard was incorporated into C standards. It was not all incorporated because it is not all good. GCC incorporated more of POSIX outside of the C standard but this has trade offs like reduced C portability and much longer and messier include files. Providing the missing POSIX functions is not difficult as Frank Wille's PosixLib shows but it is difficult to glue all the various support together without errors (different versions of BSD/Linux use different variations and versions of POSIX).

Quote:

"a lot more modern than Amiga" Memory protection + resource tracking are minimum to many. Any single-core using OS is very "unsexy" for development.

I believe the biggest reason the Amiga is no longer a "commercially viable platform" is lack of a large enough standard and supported user base.

AmigaOS 3 is likely the largest Amiga user base with tens of thousands of users between classic hardware, UAE and FPGA hardware but it has been ignored by Amiga IP owners and support has just recently been upgraded to life support with AmigaOS 3.1.4. Emulation and simulation of hardware are seen as end of life solutions for dying platforms.

AmigaOS 4 likely only has a few thousand users which is not enough to attract attention. The PPC architecture being dead and niche market prices make it less attractive now.

MorphOS 4 likely only has a few thousand users which is not enough to attract attention. The PPC architecture being dead and PPC Mac hardware getting old make it less attractive now.

AROS likely only has a few thousand users which is not enough to attract attention. At least it supports architectures which are likely to grow, can leverage affordable commodity hardware and avoids the bad management collapse of business entities. The lack of standardization on one architecture and fewer hardware options means the support for AROS will likely remain divided and incomplete.

While multi-core support, memory protection, resource tracking and security have become more important, they are not as important as a large user base. Cheap enough hardware tends to lower expectations for a product.

Quote:

And 8Ghz + 8GB is minimum for monsters like firefox. "https://en.wikipedia.org/wiki/List_of_BeOS_programs"

FireFox runs fine with half those resources. I expect an efficient and small footprint machine could get away with 1/4 of those resources or less. Haiku is probably more attractive for bloated software as most users are on a high performance x86_64 computer. Most Amiga users do *not* have a high performance computer and I doubt the majority of Amiga users would move to x86_64 as the standard Amiga architecture, especially if compatibility was lost.

Quote:

That's very pathetic. Only a few big names, like mainstream browser, otherwise... ~nothing. We have 50000 apps on Amiga compatibles? (but yes, 99% of it is obsolete/retro SW)

Losing Amiga compatibility certainly hasn't helped AROS x86_64 gain market share. Then again, they haven't attracted the "big names". Maybe it is their lack of "Memory protection + resource tracking".

Quote:

I vote for fresh start for AmigaLikeOS5. (I would not mind BeOS/BSD under the hood if it looks&feels&behaves in good&simple&reliable&FAST Amiga-familiar way.)

A new girl who looks like my girlfriend would be no replacement for my girlfriend (it is what is inside that counts). Organic growth all the way with custom hardware to help. Let the innovators innovate.

@matthey

You are spot on with the user base!!!!!!!!!!

@kolla

Quote:

building and porting software to Haiku is pretty straight forward

i wonder what makes this possible. Is it because an OS is Poxix compliant? Is it just because many development tools are available?

_________________
Philippe 'Elwood' Ferrucci
Sam460 1.10 Ghz
AmigaOS 4 betatester
Amiga Translator Organisation

@elwood

The first thing I can see is that Haiku has Scons available, which makes compilation easier...

Quote:

elwood wrote: i wonder what makes this possible. Is it because an OS is Posix compliant?

AmigaOS can be made POSIX compliant with a compatibility layer.

ixemul.library (BSD kernel Amiga shared library)
libnix (link library)
PosixLib (Frank Wille's link library)
posixc.library (AROS shared library)

These libraries have functions with the best Amiga equivalents of POSIX functions. The Amiga way of handling threads, signals, etc. is different sometimes resulting in incompatibility but overall compatibility can be good using these POSIX compatibility layers. Other than missing functions in these libraries, the biggest problems are include file errors usually as a result of the source failing to follow the POSIX standard or using an unsupported version or variation. Linux/BSD have compatibility problems between flavors as POSIX is a compromise standard.

Quote:

Is it just because many development tools are available?

I believe it is because of the scale of development. There are nearly 100 developers listed for Godot which is likely more than all the active Amiga developers combined.

Healthy Platform
developers work together in groups and sometimes even sub-groups
developers work with the main fork of software and the port is "official"
major updates occur and minor updates and bug fixes are common
developers use official development tools which are maintained

Dying Platform
developers work alone or not at all
developers work with old forks of software and the port is "unofficial"
minor updates and bug fixes are rare
developers use "unofficial" development tools which are unmaintained

The "official" development tools still worked on by multiple people that I can think of are vbcc (including vasm and vlink) and Cubic IDE (an argument could be made that they are in maintenance mode). The end of Amiga draws near without a re-launch of a more prolific Amiga. It amazes me that people show up at these Amiga forums and want to know why we can't port more software to the Amiga. Isn't it obvious?

@NoCache

Thank You - my X5000 is hungry

Which OS will You try to Support - OS4, AROS, MorphOS?

Best Regards

_________________
X DIRT-EDGE X Entertainment

A1200/X5000

@matthey

Thanks for the details about includes and posix...

I tend to agree with you with the status of AmigaOS.

Last edited by elwood on 05-Jan-2019 at 08:29 AM.

_________________
Philippe 'Elwood' Ferrucci
Sam460 1.10 Ghz
AmigaOS 4 betatester
Amiga Translator Organisation

@KimmoK

Quote:

Memory protection + resource tracking are minimum to many.

The first is by far the most important. Resource tracking would be nice to have, but without memory protection an OS isn't even in the game.

_________________
"Unix is supposed to fix that." -- Jay Miner

@matthey

The Amiga is dying, can we rescue it somehow?

You talk from time to time about getting some embedded 68k processor produced. Would that help and is there any real work being done in that direction?

Is the 68k key to the Amiga, or would AROS RPi be an alternative, switching over to ARM instead?

@MaximvsPayne
I don't have any plans in this state sorry :)
I have only OS4 available at the moment so I will see what I can do here first...
In case I will be able to compile it, I will probably release it so others can continue on this to port it to other systems....

@hth313

The currents situation makes me really sad.
But considering Aaron's plans it is more that we could imagine, question is if it is not too late but I believe that Aaron's point of view is reasonable.
About the OS porting... I was thinking about it a little already but I think that switching to OS is not an option... There are few attempts to port it to x86 architecture, it works, but it lacks the unique amiga feeling. Continuing this way will make Amiga just another PC like system.
ARM architecture would be a possible option but it depens on Hyperion entertainment and I think it will not happen any time soon.
There is still a chance for the powerpc architecture, it is just my personal opinion, but it may come that ARM architecture will reach its boundaries and then it is time for powerpc to come...
Now it depends most on Aarons plan and on developers, but Aarons presentation on Amiwest made me to buy X5000 and join developers team.
Another question is if the comunity itself want to safe amiga platform? 90% of the community are stuck with classic hardware cos they thing that amigang is not amiga anymore. It is like to say that dualcore pentium is not PC any more... Hardware evolves, so Amiga HW should as well. So, the only option is with new HW only, but community doesn't want to support it...

@NoCache

What exactly is the difference between X5000 and standard PC except processor and the high price?

To sell something you must offer something that is different to competition and in best case better. Something like Vampire is different and it offers features people want so there is a market and is interesting for many current amiga fans.

Aros on Raspberry on the other side offers a cheap and powerful NG hardware and certainly average Raspberry users are more open to a alternative OS than average users today, mostly not even using PC hardware.

Both projects are most promising today in my view.

AmigaOS on expensive custom hardware is it not, it not offers enough amiga feeling to most "classic" users and it is not good enough and lacks needed software while being too expensive to attract users from outside. Short: dead end

Last edited by OlafS25 on 06-Jan-2019 at 10:55 AM.

@OlafS25

Quote:

AmigaOS on expensive custom hardware is it not, it not offers enough amiga feeling to most "classic" users and it is not good enough and lacks needed software while being too expensive to attract users from outside. Short: dead end

We'll see soon enough if the Tabor can change your mind. Emulation is the only long term option if new hardware ceases IMHO.

_________________
"Art challenges technology. Technology inspires the art."
John Lasseter, Co-Founder of Pixar Animation Studios

Quote:

hth313 wrote: The Amiga is dying, can we rescue it somehow?

I believe the Amiga can be rescued but that there are more paths to failure than success. Trevor estimated the Amiga could be relaunched for $10 million at the Amiga Ireland event but did not give specifics on what this included. People talked about it in the following thread.

https://amigaworld.net//modules/newbb/viewtopic.php?topic_id=41613&forum=2

Quote:

You talk from time to time about getting some embedded 68k processor produced. Would that help and is there any real work being done in that direction?

If the 68k hardware path was chosen for the future, having available, affordable and faster 68k CPUs would be key to success. Embedded partners could help to share development and/or production costs. An ASIC needs production of tens of thousands of units but a 68k SoC could be competitive with ARM SoCs if production was high enough (hundreds of thousands of units). The 68k does still have appeal for some embedded players which target very high quantity production. I began talks with potential embedded partners as part of the Apollo Team but Gunnar largely ignored my efforts. I have communicated with embedded players since I left the Apollo Team but I am not in a position to make anything happen so nothing is happening. The other option to lower risk is a product which appeals to users outside of the now small Amiga market (the 68k was popular in consoles and other computers).

Quote:

Is the 68k key to the Amiga, or would AROS RPi be an alternative, switching over to ARM instead?

An ARM AmigaOS path is an option and in many ways a better choice than x86_64. The Raspberry Pi is the kind of standard and affordable hardware which the Amiga should have brought out. It is the closest modern feel to what the Amiga used to be and makes me think a 68k Amiga Pi could be successful in a post Moore's Law computer world. I don't think switching architectures and riding their wave is going to be as successful as more compatible organic grass roots development. This does require hardware investment but I can't see a software only Amiga ever being successful again, especially if existing AmigaOS compatibility is lost.

Quote:

NoCache wrote: About the OS porting... I was thinking about it a little already but I think that switching to OS is not an option... There are few attempts to port it to x86 architecture, it works, but it lacks the unique amiga feeling. Continuing this way will make Amiga just another PC like system. ARM architecture would be a possible option but it depens on Hyperion entertainment and I think it will not happen any time soon.

AROS x86_64 has been available for some time and even has multi-core support but I doubt it has as many users as UAE due to lack of compatibility. AROS ARM for the Raspberry Pi may attract more Amiga users but I doubt it will be a replacement for 68k AmigaOS users. The Vampire is a replacement/enhancement for some 68k users but isn't cheap enough to be a game changer for the Amiga. There is a way to make the 68k cheaper and faster while retaining compatibility though. The majority of AmigaOS users are 68k users and that is what they want. Moving away from the 68k loses the largest group of Amiga users.

Quote:

There is still a chance for the powerpc architecture, it is just my personal opinion, but it may come that ARM architecture will reach its boundaries and then it is time for powerpc to come...

I expect AArch64 to have similar or better performance potential compared to PPC. I expect AArch64 to have better performance than PPC on low spec hardware, with 64 bit software, with SIMD unit usage and with security software. AArch64 has a significant advantage in instruction counts and code density. The only way I see PPC AmigaOS continuing would be to develop PPC CPUs and hardware but I don't think that is worth the effort for another RISC architecture with a history of lackluster performance despite efforts from giants IBM and Motorola. The other member of the original AIM alliance, Apple could have brought PPC back when they bought P.A. Semi but they chose to drop PPC and develop ARM instead. The 68k is more competitive and has more potential than PPC for the embedded, retro, hobbyist and educational markets. The 68k is a unique product which can sustain higher profit margins.

Quote:

Another question is if the comunity itself want to safe amiga platform? 90% of the community are stuck with classic hardware cos they thing that amigang is not amiga anymore. It is like to say that dualcore pentium is not PC any more... Hardware evolves, so Amiga HW should as well. So, the only option is with new HW only, but community doesn't want to support it...

The Amiga community is supporting Vamp hardware which will likely out sell Tabor. Tabor will sell too but neither offer anywhere near the value of a Raspberry Pi (thus unlikely to attract attention outside the Amiga community). It is unlikely either will save the Amiga platform by selling enough units. Hardware is what is exciting but the price, performance and compatibility thresh hold necessary for relaunch is unlikely to be achieved by either of these products.

@matthey

Vampire is about 233mhz 68060, wich is nice for a hobby Computer like old amigas are, but not enough for quite much anything.

I'm prety sure that Tabor will over sell Vampire.

@utri007

I do not believe that

"Classic fans" are much more than NG fans and there will be only few outside buy Tabor. So I think Vampire (Standalone and the accellerator cards) will outsell Tabor by far. But we will propably never know for sure because sale numbers are kept secret by anyone

BTW Tabor is embedded hardware too, hardly hardware you would use for daily tasks

@matthey

RPi is a cheap hardware and powerful enough for many tasks, additional the RPi owners are certainly more open to alternatives than the average PC user or even worse people only use smartphones. I think there is a chance to attract new users who were not using amigas before or left amiga a long time ago.

Last edited by OlafS25 on 06-Jan-2019 at 11:47 PM.
Last edited by OlafS25 on 06-Jan-2019 at 11:43 PM.

@matthey

Seems to me that a new ASIC 68k may be the best way forward if Amiga is to be rescued.

There are plenty of these small development board coming out, RPi is just one of them. It cannot be all that expensive producing them giving the number of offers in the market. Makes me think a 68k Amiga Pi variant could have a place here as well.

Do you know why these embedded partners do not go ahead? I suppose that if a 68k ASIC is made, the majority of sales will be in the embedded field. What I mean is that we probably need the embedded industry more than they need us. They should be able to move ahead without us really.

There must be some kind of blocker, like it being too risky, the 68k IP may still be protected in some way, too much effort, lack of sales and support infrastructure or lack of development tools?

Quote:

utri007 wrote: Vampire is about 233mhz 68060, wich is nice for a hobby Computer like old amigas are, but not enough for quite much anything.

The Apollo Core is *not* twice the performance of a 68060 at the same clock speed. The Apollo Core has much faster memory and higher clock speeds than most 68060 accelerators from the '90s which makes it much faster in some benchmarks but the 68060 can use faster memory (no built in memory controller) and the rev 6 68060 can reach similar clock speeds. The Apollo Core SIMD unit probably does push performance past twice that of a similarly clocked 68060 for a few benchmarks but requires hand optimization. The 68060 was a potent low clocked CPU for the '90s and the Apollo Core is currently limited to a low end FPGA.

Quote:

I'm prety sure that Tabor will over sell Vampire.

Tabor loses half its potential performance without multi-core support (and little hope to get it from Hyperion). It is already cost reduced without a standard classic PPC FPU, MMU or SIMD unit which can reduce performance, increase bugs and/or reduce compatibility (AmigaOS 4 already has reduced Amiga classic compatibility.). It has become clear that PPC is no longer a viable architecture which is likely to further reduce demand. Tabor relies on the GPU doing most 3D processing as modern GPUs have increased processing capabilities where the Tabor CPU has reduced them. Why not port AmigaOS to OpenCL or similar and run it on the more powerful GPU while avoiding the slow PCIe bus (AmigaNG as a bus board with no/minimal CPU)?

Quote:

OlafS25 wrote: RPi is a cheap hardware and powerful enough for many tasks, additional the RPi owners are certainly more open to alternatives than the average PC user or even worse people only use smartphones. I think there is a chance to attract new users who were not using amigas before or left amiga a long time ago.

I think there is a chance to have a successful Amiga niche on the Pi. I see efficient multi-core support, improved memory protection, seamless Amiga classic emulation, an improved AROS look and building an active Amiga community for the Pi as keys to success. I expect it will be difficult to convert non-Amiga Pi users to the AmigaOS/AROS but there is a huge user base to market to. An Amiga/68k Pi would have much more retro appeal though.

Quote:

hth313 wrote: Seems to me that a new ASIC 68k may be the best way forward if Amiga is to be rescued. There are plenty of these small development board coming out, RPi is just one of them. It cannot be all that expensive producing them giving the number of offers in the market. Makes me think a 68k Amiga Pi variant could have a place here as well.

Yes, the Single Board Computers (SBCs) have become cheap and popular. Most are used for embedded markets sometimes with an attempt to market to hobbyists. The problem with marketing to hobbyists is usually a lack of software and failure to differentiate from the crowd. A 68k board can have retro software compatibility and is unique in comparison to the many ARM and x86_64 offerings.

Quote:

Do you know why these embedded partners do not go ahead? I suppose that if a 68k ASIC is made, the majority of sales will be in the embedded field. What I mean is that we probably need the embedded industry more than they need us. They should be able to move ahead without us really. There must be some kind of blocker, like it being too risky, the 68k IP may still be protected in some way, too much effort, lack of sales and support infrastructure or lack of development tools?

Good questions. Simple RISC CPUs are easier, cheaper and safer to develop. Modern higher performance CPUs are generally professionally developed and good designs which are difficult to compete with. The 68k hasn't been developed in many years so there are unknowns where businesses often want to minimize risk. The Apollo Core was seen as an amateur development despite the promising performance which is true considering the mistakes. I would bring in a professional veteran CPU engineer with CISC and variable length instruction experience (specialized and not cheap but worth it). I've communicated with a well known ex-Motorola and ex-AMD engineer who might be available and would give a project instant credibility and experience.

The 68k patents should all be expired but it would be good to have a lawyer specialized in patents take a look at possible problems. New CPU patent violations are a more likely concern. While 68k development tools are not on par with modern architectures, they are in better shape than the tools I've seen some embedded developers try to use. The biggest obstacle to better 68k tool support is a lack of modern 68k hardware.

Last edited by matthey on 08-Jan-2019 at 04:07 AM.

Quote:

Daedalus wrote: Free and open-source doesn't automatically mean an OS gets lots of support. .

But could make developer hardware cheap ...