Poster | Thread |
SHADES
| |
Re: x86/ARM - viable way Posted on 6-Oct-2021 22:37:48
| | [ #21 ] |
|
|
|
Cult Member |
Joined: 13-Nov-2003 Posts: 865
From: Melbourne | | |
|
| @NutsAboutAmiga
Well put.
Time for the OS to migrate to a platform that's cheap to purchase and remains, at the least, a semi-standard. Last edited by SHADES on 06-Oct-2021 at 10:43 PM.
_________________ It's not the question that's the problem, it's the problem that's the question. |
|
Status: Offline |
|
|
terminills
| |
Re: x86/ARM - viable way Posted on 7-Oct-2021 2:13:05
| | [ #22 ] |
|
|
|
AROS Core Developer |
Joined: 8-Mar-2003 Posts: 1472
From: Unknown | | |
|
| @SHADES
Quote:
I am mainly talking about modern feature sets like SMP and memory management tasks for the OS etc |
These are the features I was excited about for AROS64-SMP. However that seems to be on the back burner and I once again am losing interest. _________________ Support AROS sponsor a developer.
"AROS is prolly illegal ~ Evert Carton" intentionally quoted out of context for dramatic effect |
|
Status: Offline |
|
|
ppcamiga1
| |
Re: x86/ARM - viable way Posted on 7-Oct-2021 6:49:29
| | [ #23 ] |
|
|
|
Cult Member |
Joined: 23-Aug-2015 Posts: 762
From: Unknown | | |
|
| @agami
So Your idea is to cut cost on binary compatibility everything else be the same. It is stupid. It will not work. Swittch to unix is only way forward.
|
|
Status: Offline |
|
|
g_kraszewski
| |
Re: x86/ARM - viable way Posted on 7-Oct-2021 7:50:14
| | [ #24 ] |
|
|
|
Regular Member |
Joined: 3-Sep-2010 Posts: 343
From: Unknown | | |
|
| @ppcamiga1
Quote:
It is obvious that AROS should be made usable by people that want to switch from ppc. For example You. |
In 2015 I have decided to abandon the whole dead end of Amiga NG. I have turned back to classic M68k, considered solely as a retro platform, without ambitions for development.Last edited by g_kraszewski on 07-Oct-2021 at 07:51 AM.
_________________ RastPort |
|
Status: Offline |
|
|
Rose
| |
Re: x86/ARM - viable way Posted on 7-Oct-2021 8:30:28
| | [ #25 ] |
|
|
|
Cult Member |
Joined: 5-Nov-2009 Posts: 982
From: Unknown | | |
|
| @ppcamiga1
Quote:
Yes, because you lack basic understanding from what you are talking about. |
|
Status: Offline |
|
|
amigang
| |
Re: x86/ARM - viable way Posted on 7-Oct-2021 9:56:23
| | [ #26 ] |
|
|
|
Elite Member |
Joined: 12-Jan-2005 Posts: 2018
From: Cheshire, England | | |
|
| @SHADES
Quote:
I am mainly talking about modern feature sets like SMP and memory management tasks for the OS etc. |
Don't get me wrong i would like to see that in a modern Amiga, but even if OS4 or Aros or MorpthOS team pull it off in supporting such system, I unfortunately don't see the developers/ software coming out taking advantage of it.
I kinda started to feel we should go back more to our roots and show just how much can done on low level hardware. I think most people are surprised to learn that Amiga made so much CGI in the early days for such low level hardware. I remember showing a younger friend my Amiga setup and he asked what specs they where, when I told him its 16mb Ram, 40Mhz CPU, he thought I was lying, didnt think anything could run at them low specs.
Quote:
I would bet you that the interest in emulation would not be growing the Amiga user base by much. Certainly not in development of the OS itself. I'm not saying even a single user gained is bad, at this stage, AMIGA needs even a single user but OS development, for emulation? I don't see it. |
I kind of agree but then again, I think the emulation market of the Amiga has been massively underutilized by companies and the community. Cloanto (who lets not forget has basically made all its money and been able to raise funds to buy Amiga by just selling a emulation package!) should be making a complete Amiga Forever setup for mobile phone market with maybe a nice easy to use front end like whats being used to power the A500 mini but why not add a built in App store so devs could sell maybe Amiga games with in it, that could open up a whole new market for Amiga devs, the A500 mini could of included the same thing, they could also add it to there Windows version. Money raised could go into improving emulation and the OS.
Just look at the sales of AmigaOS 4.1 Classic when PPC support was added to WinUAE, look how many copies of Amiga Forever / AmigaOS 3.2 are selling just for emulation purposes. Amikit XE for Pi has raised funds to improve emulation support on the Pi. Pistorm is creating a whole new market for cheap low cost upgrades for Amiga users, in theory this could be expanded to adding virtual PPC support on the cheap!
they could also do the same as SEGA does, they sell there old mega drive games and collections on multiple platforms, where is the Amiga Collection for PS5 or Xbox. Even Good Old games sell Cinemaware like this, https://www.gog.com/game/wings_classic
there are many interesting markets and ideas to explore, I just wish we had better leaders to explore these low cost options. Last edited by amigang on 07-Oct-2021 at 10:08 AM. Last edited by amigang on 07-Oct-2021 at 10:01 AM.
_________________ AmigaNG, YouTube, LeaveReality Studio |
|
Status: Offline |
|
|
OlafS25
| |
Re: x86/ARM - viable way Posted on 7-Oct-2021 10:21:53
| | [ #27 ] |
|
|
|
Elite Member |
Joined: 12-May-2010 Posts: 6321
From: Unknown | | |
|
| @g_kraszewski
I agree to that... 68k is offering most fun today
@thread
Regarding asking modern features you very propably break most software to get on par with the big platforms. But still you lack the developer resources the big platforms have, not only for OS development but also to have at least a basic set of applications most users need. You need f.e. developers who contantly improve the web browser to stay on same level as other platforms. You must develop lots of drivers for different hardware including USB support. I see two problems (even if I got bashed for it by some aros supporters), from my experience one source for all platforms is advantage and curse at the same time. I frequently experienced it that features or software on 68k no longer worked that worked 2-3 months before. Obviously commits broke it. So you would need a huge infrastructure with lots of test cases and teams working on different sources with managers coordinating it to speed development. Or only 2-3 platforms but with testers and perhaps different sources. But there were no big numbers of skilled OS developers working on it, in good times perhaps 2-3 and some devs doing smaller commits. I do not think that any platform has a chance to compete anymore. Why using something like that at all when you have plenty of options already? Many people even not use a desktop anymore so to create a real market you would need to rethink amiga at all and create something new that is task orientated so even interesting to people not knowing amiga. But that would not be very interesting to most of us I guess
The discussion with using hosted or native I would agree that hosted versions that use the OS they run on as much as possible and offer chance to share data would be best option in current situation. Or developers who create lots of drivers... Both options would certainly need a lot of work and there we are at the big problem of AROS... lack of developers
I use it only on 68k and because it runs on UAE (and V4) and you can also easily mix binaries the situation is not as problematic as it is on X86/X64 or ARM. PPC always was dead because noone interested in it. On PPC competition to MorphOS/AmigaOS would be too strong anyway because AROS is only source compatible and is not directly integrating 68k like with Petunia only using UAE for 68k.
To attract new users outside the whole concept (expecially the user interface) must be process orientated and intuitive to use, very different from now. And perhaps object orientated and flexible. And you would need to invest lots of resources to integrate it in todays world. People want to use what they have (mostly cloud based). Unfortunately many of these cloud based service are closed. Last edited by OlafS25 on 07-Oct-2021 at 10:53 AM. Last edited by OlafS25 on 07-Oct-2021 at 10:34 AM.
|
|
Status: Offline |
|
|
OlafS25
| |
Re: x86/ARM - viable way Posted on 7-Oct-2021 10:25:30
| | [ #28 ] |
|
|
|
Elite Member |
Joined: 12-May-2010 Posts: 6321
From: Unknown | | |
|
| @amigang
emulation and FPGA based hardware will keep amiga alive. And yes I like emulation too, it is option to use something old on modern hardware without lots of problems with configuration and missing drivers |
|
Status: Offline |
|
|
IridiumFX
| |
Re: x86/ARM - viable way Posted on 7-Oct-2021 14:08:50
| | [ #29 ] |
|
|
|
Member |
Joined: 7-Apr-2017 Posts: 80
From: London, UK | | |
|
| @OlafS25
Quote:
OlafS25 wrote: @amigang
emulation and FPGA based hardware will keep amiga alive. And yes I like emulation too, it is option to use something old on modern hardware without lots of problems with configuration and missing drivers |
Indeed, if it runs Amiga apps, it's an Amiga, afaiac. I totally agree.
I am rediscovering the joy of C/Asm hardware programming on the AGA/68K. It's nice, interesting and leaves a pleasant feeling after every session. And by the time the AmigaMini (and later on the maxi ?) are released, I hope to be back in shape for proper coding. |
|
Status: Offline |
|
|
liquidbit
| |
Re: x86/ARM - viable way Posted on 7-Oct-2021 16:54:22
| | [ #30 ] |
|
|
|
Member |
Joined: 9-Dec-2004 Posts: 38
From: Unknown | | |
|
| The idea of having a next generation 68000 processor instead of a PPC is a lot better option.
Who would not admit that he would prefer have a computer that runs Amiga 68x programs natively rather via a simulation.
We do not want sandboxes any more, we want the REAL thing but FASTER!
A great example is the implementation of the Spectrum Next (https://www.specnext.com/about/).
If it was a product available like this I would buy it yesterday!
If there is no way to win the PC/MAC world at least stay RETRO.
|
|
Status: Offline |
|
|
ppcamiga1
| |
Re: x86/ARM - viable way Posted on 7-Oct-2021 18:25:04
| | [ #31 ] |
|
|
|
Cult Member |
Joined: 23-Aug-2015 Posts: 762
From: Unknown | | |
|
| @Rose
Yes, because I don't want to switch from better Amiga than made by Commodore to cheap shit that is not compatible and at the some time not modern and not fast.
Last edited by ppcamiga1 on 07-Oct-2021 at 06:25 PM.
|
|
Status: Offline |
|
|
ppcamiga1
| |
Re: x86/ARM - viable way Posted on 7-Oct-2021 18:27:42
| | [ #32 ] |
|
|
|
Cult Member |
Joined: 23-Aug-2015 Posts: 762
From: Unknown | | |
|
| @g_kraszewski
As a software developer I don't care about cpu as long as os is binary and source compatible. It means cpu must be big endian. 68k is worse because many times slower. I don't want to waste my precious time for pointless optimisation.
|
|
Status: Offline |
|
|
ppcamiga1
| |
Re: x86/ARM - viable way Posted on 7-Oct-2021 18:29:07
| | [ #33 ] |
|
|
|
Cult Member |
Joined: 23-Aug-2015 Posts: 762
From: Unknown | | |
|
| @liquidbit
Binary and source compatibility with 68k software is important. But 68k may run via simulation. Why not. |
|
Status: Offline |
|
|
ppcamiga1
| |
Re: x86/ARM - viable way Posted on 7-Oct-2021 18:30:54
| | [ #34 ] |
|
|
|
Cult Member |
Joined: 23-Aug-2015 Posts: 762
From: Unknown | | |
|
| @OlafS25
It is simple solutions for that. Amiga gui and graphics on top of unix. It will be fast, modern, will have Amiga feeleng, will be enough compatible.
|
|
Status: Offline |
|
|
TRIPOS
| |
Re: x86/ARM - viable way Posted on 7-Oct-2021 22:22:08
| | [ #35 ] |
|
|
|
Super Member |
Joined: 4-Apr-2014 Posts: 1204
From: Unknown | | |
|
| |
Status: Offline |
|
|
SHADES
| |
Re: x86/ARM - viable way Posted on 7-Oct-2021 23:26:30
| | [ #36 ] |
|
|
|
Cult Member |
Joined: 13-Nov-2003 Posts: 865
From: Melbourne | | |
|
| @OlafS25
Quote:
@thread Regarding asking modern features you very propably break most software to get on par with the big platforms. But still you lack the developer resources the big platforms have, not only for OS development but also to have at least a basic set of applications most users need |
Pi is not a "big-boy" started from f-all. Costs F-all. Tons of developers on that platform because, it was so darn cheap! Even Windows runs on it because it's becoming so popular. It's cheap, it's design stays compatible, not a ton of hardware to re-program for every time.
You won't get "developers" on an expensive niche board running a niche paid for OS. IT's not going to happen unless you pay them.
As for it won't ever happen on a cheap, widely supported platform, I think that's very unrealistic. Pi would have failed if that's the case. Why run Linux on Pi when you can just run it on x86. That's rubbish. Sure, it will grow slowly, but it will grow because it's cheap. AROS moves forward regardless of zero company backing, same with HAIKU, ReactOS I mean, there are so many examples of this I can't see your point.
You use only 68k because that's your only choice. If you got the same experience on ARM, you'd use ARM. Hell, Pi-Storm plugs into the CPU BUS and only emulates the CPU but offers access to other Linux sub systems via emulation. Kills the competition in real hardware. Have a look at the JIT 68k emulator being tried on it on a Pi 3a getting 800 mips. But again, it's old 20+year old hardware that it plugs into! Move on already. It CAN be done, it should be done, it's possible and it's affordable to move. A native OS can develop to gain access to all the modern features over time, especially if it's cheap and feel great like an AMIGA.
Pi-Storm feels exactly like an AMIGA, right now. Just hasn't got a Motorola 68k running it.
What holds it back is the 20+ year old hardware that it plugs into and that it has to emulate 68k instead of run native. No one would bother trying to bring forward the 68k OS for emulation only. It's a dead end. It remains in FPGA/emulation only because it's expensive to use a long dead subsystem. 68k doesn't have access to modern hardware because it's long dead. Imagine if the AMIGA OS just ran native. Started being developed towards moving itself under ARM / RISC-V. Cheap platforms to run on, modern hardware to write access for.
Yes, it's not easy, yes it will take time but OS development will grow if it's affordable to purchase and work on and the hardware remains semi-standard and well known.
It already is happening! https://linuxjedi.co.uk/2021/10/03/this-week-in-pistorm-2021-10-03/
They are being used and writing emulation for the 68k CPU which, is also, not easy! and being developed for free!Last edited by SHADES on 07-Oct-2021 at 11:40 PM. Last edited by SHADES on 07-Oct-2021 at 11:34 PM. Last edited by SHADES on 07-Oct-2021 at 11:31 PM.
_________________ It's not the question that's the problem, it's the problem that's the question. |
|
Status: Offline |
|
|
kolla
| |
Re: x86/ARM - viable way Posted on 8-Oct-2021 3:45:23
| | [ #37 ] |
|
|
|
Elite Member |
Joined: 20-Aug-2003 Posts: 2859
From: Trondheim, Norway | | |
|
| @ppcamiga1
Quote:
ppcamiga1 wrote: @OlafS25
It is simple solutions for that. Amiga gui and graphics on top of unix. It will be fast, modern, will have Amiga feeleng, will be enough compatible.
|
Yes, and that’s what OS4 was supposed to be, originally - follow the path of NeXT. Instead the route went via QNX and TAO before the whole thing nosedived into just porting OS3 with all its flaws and limitations to PowerPC. And here we are, quarter of a century later, and same old. Meanwhile, Apple did exactly what was originally proposed for Amiga, first with A/UX and later with Rhapsody/OSX. So if you want to see what AmigaOS could (and should) have been, had the original post-CBM plans been carried out, then just look at what Apple did. CBM had no solid long term plans for OS development, the idea there was to just use WinNT, though I guess the CBM developers would have preferred to further build on AmigaOS and AMIX.
Regardless though, binary compatibility with existing 68k software would largely have been thrown out the window, and with architecte agnosticism and portability being keywords for 90s OS development, it wouldn’t have been locked to PowerPC or any other specific architecture.Last edited by kolla on 08-Oct-2021 at 03:46 AM.
_________________ B5D6A1D019D5D45BCC56F4782AC220D8B3E2A6CC |
|
Status: Offline |
|
|
Turrican3
| |
Re: x86/ARM - viable way Posted on 8-Oct-2021 9:22:19
| | [ #38 ] |
|
|
|
Regular Member |
Joined: 20-Jun-2003 Posts: 386
From: Italy | | |
|
| @SHADES
Quote:
Pi is not a "big-boy" started from f-all. Costs F-all. Tons of developers on that platform because, it was so darn cheap! Even Windows runs on it because it's becoming so popular. It's cheap, it's design stays compatible, not a ton of hardware to re-program for every time.
You won't get "developers" on an expensive niche board running a niche paid for OS. IT's not going to happen unless you pay them.
As for it won't ever happen on a cheap, widely supported platform, I think that's very unrealistic. Pi would have failed if that's the case. Why run Linux on Pi when you can just run it on x86. That's rubbish. Sure, it will grow slowly, but it will grow because it's cheap. AROS moves forward regardless of zero company backing, same with HAIKU, ReactOS I mean, there are so many examples of this I can't see your point.
You use only 68k because that's your only choice. If you got the same experience on ARM, you'd use ARM. Hell, Pi-Storm plugs into the CPU BUS and only emulates the CPU but offers access to other Linux sub systems via emulation. Kills the competition in real hardware. Have a look at the JIT 68k emulator being tried on it on a Pi 3a getting 800 mips. But again, it's old 20+year old hardware that it plugs into! Move on already. It CAN be done, it should be done, it's possible and it's affordable to move. A native OS can develop to gain access to all the modern features over time, especially if it's cheap and feel great like an AMIGA.
Pi-Storm feels exactly like an AMIGA, right now. Just hasn't got a Motorola 68k running it.
What holds it back is the 20+ year old hardware that it plugs into and that it has to emulate 68k instead of run native. No one would bother trying to bring forward the 68k OS for emulation only. It's a dead end. It remains in FPGA/emulation only because it's expensive to use a long dead subsystem. 68k doesn't have access to modern hardware because it's long dead. Imagine if the AMIGA OS just ran native. Started being developed towards moving itself under ARM / RISC-V. Cheap platforms to run on, modern hardware to write access for.
Yes, it's not easy, yes it will take time but OS development will grow if it's affordable to purchase and work on and the hardware remains semi-standard and well known.
It already is happening! https://linuxjedi.co.uk/2021/10/03/this-week-in-pistorm-2021-10-03/
They are being used and writing emulation for the 68k CPU which, is also, not easy! and being developed for free! |
I agree with every single word you said. It just makes sense and it should have been done ages ago!
We have UAE for legacy 68K, there's literally no reason to avoid switching to modern, cheaper, faster hardware if (and that's a big if, sure) we want the Amiga to survive. |
|
Status: Offline |
|
|
OlafS25
| |
Re: x86/ARM - viable way Posted on 8-Oct-2021 10:46:24
| | [ #39 ] |
|
|
|
Elite Member |
Joined: 12-May-2010 Posts: 6321
From: Unknown | | |
|
| @SHADES
PiStorm with Emu68k is indeed interesting. Michal doing Emu68k is one of the best Aros developers. I only wonder if it not makes more sense to have something 68k based running native on ARM. It sounds a little like Amithlon, only for ARM instead of X86. it would be a good concept at least for applications and games only using OS. I understand that PiStorm is trying to offering a cheap accellerator but still there is the disadvantage that it needs the old hardware and that the hardware cannot be improved like on FPGA. So you have a fast processor running in a old hardware. Vampire f.e. more or less replaces all components and only uses keyboard and some interfaces. PiStorm is much more dependent on the old hardware and its limitations. From my point of view both concepts can live together, different goals behind.
If I remember right the people behind PiStorm claim that they do what Vampire promised but not delivered. But that is obviously a misunderstanding of the apollo project. I discussed with gunnar long time ago and he even then said vampire cards were only to get something for testing as a kind of shortcut because the standalone devices were always the goal. Vampire was never only a relative cheap accellerator card but designed as a new amiga generation. PiStorm are a good concept for affordable new accellerators. As I wrote both concepts can and will live together. Last edited by OlafS25 on 08-Oct-2021 at 11:16 AM. Last edited by OlafS25 on 08-Oct-2021 at 11:07 AM.
|
|
Status: Offline |
|
|
OlafS25
| |
Re: x86/ARM - viable way Posted on 8-Oct-2021 11:25:14
| | [ #40 ] |
|
|
|
Elite Member |
Joined: 12-May-2010 Posts: 6321
From: Unknown | | |
|
| @ppcamiga1
it is certainly possible. But why would people install and use it?
From my point of view you need something new to attract people. And you need professional developer support.
But then it would be not much amiga related anymore. And you need lots of money for it. Last edited by OlafS25 on 08-Oct-2021 at 11:27 AM.
|
|
Status: Offline |
|
|