Sometime ago, I wrote to the MorphOS team about an idea for the future of MorphOS,
having received no answer, I am putting it here for all to analyze and discuss, so
please read the initial message and the two files, comment and discuss:

Here is the original message which I sent to the MorphOS team (slightly edited):
> Hello to all the MorphOS team. First let me tell you who I am, my name is
> [removed for privacy], I was an Amiga user in the past. I beg you to read
> this message and the attached files without trashing it and to pass it
> around the team. I am writing about the future of Amiga-like operating
> systems on the market. I have analyzed the market and have concluded that
> it is possible for an amiga-like systems to become common again. This was
> not true 10 or even 5 years ago.
>
> This message was written due to my realization that there was a spot in
> the market for an Amiga-like operating system again combined to the fact
> that the powerpc is a dead-end, as the team has already stated. The team
> has, in the past already stated that it thinks that moving the system to
> the x86 or ARM platform is the only way for the longer term. I have to
> disagree. The x86 architecture is an ugly monstrosity with several
> execution modes, endless extensions, inssuficient registers and is
> generally disgustly unclean. The x86 deployment has plateaud and will
> hopefully start to reduce its market share (I detail why this is likely
> in the attached file in the market analysis section). Before you think I
> am crazy, I am not saying that it will dissapear, simply that it will
> loose market share compared to now. The ARM architecture on the other
> hand, while much more clean, still has several execution modes and it is
> not the solution to everything that some tout it to be. For one, it is
> literally a toy, scaling poorly in power. Its power efficiency becomes
> poor as the clock speed increases. In fact it is the spiritual successor
> of the MOS 6502 and was designed as such, the makers originally made it
> as a RISC successor of the 6502. The design will probably not become
> popular, for example, in the server market contrarly to what is commonly
> said (this is detailed in the attached file). The ARM servers are
> supposed to be out since three years ago yet are still unavailable for a
> good reason. Finally the ARM native byteorder is little endian (it does
> however have limited big endian support) rather than the big endian of
> the Amiga. On the other hand another instruction set is making a comeback
> and could be declined in both server and mobile chips, is perfectly clean
> and optimal and is extremely power efficient, allowing high performance
> with low power and low heat dissipation, uses the big endian byteorder of
> both the Amiga and the Internet. This instruction set is the MIPS
> instruction set. It is making a comeback (Imagination Technologies has
> signed over 20 new licensors these last years), becoming popular again in
> printers, networking gear as well as chinese mobile devices. While I
> would prefer ARM to the x86 architecture, the MIPS platform is the
> obvious best way to go for an Amiga-like platform.
>
> In the attached file called rebirth.txt I detail everything from the market
> analysis to the way to go about the hardware. Of course, I supply a
> rationale for everything I write because a suggestion without a
> justification is worthless.
> In the other file, I state how I can help with such a project, it would
> not be with programming since I am no programmer, just a (very) educated
> layperson. It goes without saying that I would help as much as I can and
> that I would do so for free.
>
> Please think about it, either MorphOS could remain a small alternative
> hobbyist platform which will never be popular, or it can become "the new
> Amiga", being truly used and liked by many people again.
>
> I have lots more ideas and trics to improve the chances of success and
> ideas to improve the performance of the code, however, they will be sent
> later on.
>
> Again, to all the team, please read the documents and discuss them.
>
> Note: the documents are in the plain text format.

The documents are here:
http://www.warmup-asso.org/download/test/rebirth.txt
http://www.warmup-asso.org/download/test/helping.txt

Do not hesitate to dissect the proposal and discuss it. If you think it is not doable, please say why, don't remain silent.

Edit: after reading the first few replies, I wish to add that the proposal, which I recommand that you read and dissect, is in the file rebirth.txt. The content of the message which I initially sent to the MorphOS team is merely a bit of context to the proposal which follows, If you all only address that content, you will merely discuss a small point, which, taken without the remaining, is of medium importance and you will never get to dicussing the main point. Please make sure to read the files and to discuss the content of rebirth.txt as it is the main part.

Note, reading the documents again, I realised that there are quite a few minor errors, even though I checked the document for errors many times in the past, and one major one which prevents major understanding. The major one is the following, i state:
> The MUI framework should be split in two, a generic part !include!d in the proprietary framework, supplied with Amiga E bindings
The second part of the sentence is missing, I wanted to say: a generic part !include!d in the proprietary framewors, supplied with Amiga E bindings and the other part, containing workbench oriented functions, which should be !include!d in the upgrade kit which turns the box in a desktop computer.

Please note: this message is mostly a cross-post from a topic on the Warmup forum.



[color=#f2f2f2][ Edité par amigabeliever 09.03.2016 - 23:05 ][/color]

PPC died because of lack of support. Isn't MIPS in danger of going the same path here?

Quote:

x86 is, x64 isn't.

> ARM servers are [...] still unavailable

http://www.nextplatform.com/wp-content/uploads/2016/01/arm-servers-muller-keynote.jpg

> another instruction set [...] uses the big endian byteorder [...]. This instruction set
> is the MIPS instruction set. [...] the MIPS platform is the obvious best way to go for
> an Amiga-like platform.

Endianness depends on the specific microarchitecture.

https://morph.zone/modules/newbb_plus/viewtopic.php?forum=11&topic_id=11137&start=35

The MIPS64r6-based Warrior P6600 core does seem to support big-endian, though.

> Imagination Technologies has signed over 20 new licensors these last years

Wouldn't these be licensees?

> I can help with such a project, it would not be with programming since I am no
> programmer, just a (very) educated layperson. [...] I have [...] ideas to improve
> the performance of the code

Is this supposed to sound convincing? ;-)

MIPS instead of PPC or X64?
Why not Power8 (or just stick with PPCs for a little longer)?

> Why not Power8 (or just stick with PPCs for a little longer)?

https://morph.zone/modules/newbb_plus/viewtopic.php?forum=3&topic_id=9463&start=68 ;-)

> x86 is, x64 isn't.
x86-64 is one more execution mode on top of the horrible x86. If they made x64 circuits without x86 backward compatibility, maybe you would have a point.

>> ARM servers are [...] still unavailable
http://www.nextplatform.com/wp-content/uploads/2016/01/arm-servers-muller-keynote.jpg
Well, I guess you proved that there are some ARM servers arriving, I wouldn't say that they are currently the slightest competition to x86 though. It is still however true that the power efficiency of ARM CPUs becomes somewhat dissapointing once you raise the clockspeed. In other words, they are not what datacenter operators hoped them to be, in other words they are not the machines which would finally allow them to significantly lower their energy bills (those devices would merely allow to lower energy consumption a little).

> PPC died because of lack of support. Isn't MIPS in danger of going the same path here?
The MIPS market is growing. While it was going badly in the past, the change of ownership to Imagination Technologies is bringing some new growth. Unlike the previous owner, Imagination knows how to sell the thing.

@thread
If you only address the bit of content quoted in my initial message (which is merely a bit of context for the proposal which is in the files), you are going to miss the point.
The topic, isn't about simply discussing which CPU for the future of MorphOS, in which case several architectures may be a viable solution. It is about a specific strategic commercial approach to make an Amiga-like operating system mainstream again (and no I am not saying that simply changing CPUs would make the platform popular). PLEASE PLEASE PLEASE read the two linked files (especially rebirth.txt) and discuss the content of rebirth.txt, otherwise, as I said, you will miss the point.

But i have a question who are you to beleave that morphos team must choose your way ? A rich partner ? Cjoosing MIPS is not the best choice since there is no common hardware using it. No laptop ....

The only way to be mainstream again is to have money.... For now i think it's impossible. To be mainstream we must have an hardware partner wich produce stuning machines at low cost. We must have a bigger team to develop the operating system and Morphos should be more secured than now for viruses and hack. We must grant also an anti-piracy system full of fucking DRM. Because big firm will not invest... So i think in this way with killer apps and games we can be mainstream. But the actual morphos team will not find any pleasure with à such à way. They are working for there plesure and they want certinly to improve morphos but they won't kill themselve in building a mainstream operating system full of security and DRM . the world have changed. Big firm like adobe,Microsoft, electronic arts and others are building must have software and if you want them on your operating system you must have ......much money

[ Edited by acepeg 10.03.2016 - 07:29 ]

Quote:

From nothing to slightly more than nothing? This was also the case of PPC in the mid 90's, but then PPC had a world leading company supporting it and one of the most famous companies using it (Apple). With Apple the future looked brighter and brighter for PPC until the early 2000. Everyone knew about Apple and PPC, only geeks know about MIPS.

In the end Dave Haynie was right: with only one influential company using it against hundreds using Intels it was only a matter of time before their processors surpassed the PPC. Intel was selling hundreds of millions of processors every year against the PPC's millions, which translated to a lot more money for R&D for Intel. I don't see how MIPS can stay ahead or relevant for the very same reason.

If MorphOS is going to use a non standard, mostly unused processor with hardly any consumer hardware (if any) the Team might just as well stick with PPC forever.

Quote:

With all due respect, this is poppycock. By that logic we shouldn't use PC video cards either, since they can still operate in a VGA compatible mode.

I think so... Why continue only on big endian if WE rewrite the whole operating system breaking the backward compatibility ? We must introduce many features like memory protection and smp and...

[ Edited by acepeg 10.03.2016 - 12:59 ]

> discuss the content of rebirth.txt

Okay.

> Here are some suggestion for hardware components: [...]
> the main SoC
> SoC containing the following licenced IP cores:
> MIPS p6600 2Ghz single core processor 64kb+64kb L1+L1 instruction+data cache
> PowerVR GT7900: comes from the same company as the MIPS processor
> [...] there is only one cpu core (better for MorphOS)
> [...]
> For the cost of IP cores, it may be possible to negociate a fee per unit produced
> instead of an upfront cost. [...] the most likely path to success would be through
> an arrangement with Imagination Technologies, makers of the MIPS processor. [...]
> In exchange for being supplied with one of the best operating systems in the world,
> they cound supply some advantage in return such as financing the development or,
> at worst, a discount on the IP licencing cost for the hardware project.

Who is going to develop and manufacture this 2 GHz single-core P6600-based SoC with PowerVR GT7900 GPU for the MorphOS team? And as soon as this SoC exists, who is going to develop and manufacture a mainboard for it?

@thread
I you read the file rebirth.txt carefully, you may wish to skip this message (it is unfortunately a bit long), since acepeg states points which are addressed in the file, I am answering his points by quoting a few parts of the file. I will address points which are not treated in the file in a second message for ease of reading.

@acepeg
> The only way to be mainstream again is to have money.... For now i think it's impossible.
> To be mainstream we must have an hardware partner wich produce stuning machines at low cost.
> We must have a bigger team to develop the operating system and Morphos should be more secured than
> now for viruses and hack.
I agree that to go upfront against the Wintel duo would require huge amounts of money and for MorphOS to try to do this would be foolish. There is however potential for disruptive technology. This is why I suggest to exploit an untapped market (the tv world convergence box) and using it as a back door (through a desktop upgrade kit) to reconquer a fraction of the desktop market. Have you even read the file rebirth.txt? In the file I state the following (in section 2a):
> There is one market which is still underdevelopped, the boxes which are plugged on the TV.
> There is particularly an absence of that elusive convergence box. [...]
> A typical cord-cutters media setup includes the following:
> DVD or Blu-Ray player: to play back movies
> OTA-television DVR: to watch and/or record TV programming [...]
> Game console: for playing of big studio games
> Roku, AppleTV or equivalent: for OTT streaming, casual games and streaming from one's NAS

> This list is obviously a problem, not only will many people find that there are too many boxes,
> but it doesn't even cover all one may want to do with the living room video display (TV).
> There are many more potential functions:

> Play indie games (consoles mostly have big studio games)
> Play high quality amateur (free) games
> Allow children to play educational games
> Acess textual data services (news, weather, road traffic, etc.)
> Watch/record a video filmed with one's cameraphone / tablet / digital camcorder
> View interactive multimedia content
> View HBBTV
> etc.

>These functions in the second list, combined with those of the first list, could be combined in a convergence box.
> Here are some facts about the above liste:
to see the facts, please refer to the file rebirth.txt (section 2a) as this message is becoming unreasonably long

I also state (in section 2c):
> In the same way as an upgrade kit was sold which transformed a playstation 2 into a rudimentary computer,
> an upgrade kit should be sold which allows to use the convergence box as a basic computer. [...]
> For MorphOS users, this would have the advantage of garanteeing a sufficient number of applications
> and games in the "TV mode" which would cover multimedia needs and gaming; while at the same time
> it would allow to run software started from the workbench environment [...]

> Offering such an upgrade kit, [...] would also be useful to other people, mostly those who only wish
> to add web browsing and document viewing to using a media/gaming center. Some people do not need
> a full computer and will be satisfied by that, while other people have many computers with one or more
> of them only used for basic capability (ex: web browsing), capabilities for which the upgrade kit would
> be sufficient to constitute a replacement.

> Finally, over the time, it may be possible to educate the customers to make them realize that the
> TV functions convergence box with the upgrade kit is usable as more than a said box with added
> web-browsing and document viewing, and as such, it may be possible to reconquer [a small] part
> of the desktop market.

> To be mainstream [...]
> We must grant also an anti-piracy system full of fucking DRM.
Well, it depends, if we go the way of the TV-world convergence box with an optional desktop upgrade kit, we may not need so. Do you remember the controversies on the Android platform where developers are furious at google for its lack of a robust copy-protection to protect their apps (the developers were considering the Android platform inferior to iOS becauso of this)? Has this resulted in a lack of applications to Android (even though everybody is pirating en masse)? The Android platform has a DRM framework to protect videos but not much in term of protecting applications. Moreover, for video on the convergence box, since there should be an application store, why not let the developers of the video service be responsible for implementing any potential DRM which they consider necessary inside their own app without supplying such services at the operating system level (except for a single function to turn-on the HDCP on the HDMI port, which everybody knows can be removed with one of those made-in-China devices bought online)?
Finally, if you want DRM-free videos on the platform, there are many small such services and one big one EZtakes, it would be easy to make sure there is an EZtakes app on the platform. And there are DRM-free classic games on GOG, with some of the games bordering on legendary status. This is to say a (partly or maybe even wholely) DRM-free platform is viable

>> If they made x64 circuits without x86 backward compatibility, maybe you would have a point.
> With all due respect, this is poppycock. By that logic we shouldn't use PC video cards either, since they can still operate in a VGA compatible mode.
Well, this might be a worthwile point if the x86 64bit long mode was a purely separate mode (even though the processor including a backward compatibility mode is equal to carrying a bag of one's trash with oneself at all time), but the fact is the longmode includes the legacy instruction set with its horrors.

>> The MIPS market is growing.
> From nothing to slightly more than nothing? This was also the case of PPC in the mid 90's,
The same could have been said of the ARM processor 10 years ago. They targeted a different market than the desktop one and were successfull. The same can happen for MIPS. Moreover, the change of ownership has just happened. Give Imagination a bit of time to fix the flaws of its predecessors. Finally, the networking gear manufacturers are switching from PPC to MIPS, the printer manufacturers are switching from PPC to MIPS (or, less frequently, ARM) and the chinese are embracing MIPS as their favorite platform (they are not sheep that need to do like everyone else). The real reason why PPC is loosing ground is not because Apple left it, even though this fact doesn't help; it is because the two main companies behind it are not trying to push it. Until recently, with the creation of the OpenPower consortium, IBM only developped the Power Architecture for its own computers. On the side of Freescale they only care about their three main lucrative markets: automotive, industrial and aerospace and do not care about the rest of the world. They almost respond to the loss of the printer and router market by shrugging their shoulders.

@thread
I am baffled by the fact that most people seem to prefer the x86 processor to the ARM processor which, at least, while suboptimal, is much cleaner and is nonetheless widely available.

Quote:

OK Andreas, OK.
Power8 IS PPC.
Sheesh!

[ Edited by Jim 10.03.2016 - 19:58 ]

Quote:

Compatibility does NOT have to be sacrificed if we run new apps in a separate box (outside Abox).

Quote:

They would not be integrated, which mean that the same thing (separated new and old binaries and their data) could be achieved by running the *old* environment in a separate box instead, like UAE. This correctly brings the primary focus to the new OS with the evolved features, and the legacy compatibility becomes even higher since also the Amiga HW is "present" through emulation.

> Who is going to develop and manufacture this 2 GHz single-core P6600-based
> SoC with PowerVR GT7900 GPU for the MorphOS team? And as soon as this
> SoC exists, who is going to develop and manufacture a mainboard for it?
There are several ways to go about this. The MorphOS team could take on board a chip designer, as there are people able to do this in the Amiga world. As an alternative, Imagination Technologies could be asked to do it (they supply services around their own cores). We may get them to do so in exchange for an operating system. They are many hardware designers in the Amiga world. Also, Acube can make a mainboard.

In the file, I state the following (in section 4b):
> To design the hardware, the MorphOS team could partner with an other Amiga company,
> such as Acube, while the team works on the software. To manufacture the hardware
> (PCB ICs) it could pick a subcontract manufacturer based in Europe or North America,
> which has a good quality control. To distribute and market the unit, a partnering with
> a distribution and marketing company can also be a solution. For the cost of IP cores,
> it may be possible to negociate a fee per unit produced instead of an upfront cost.
> There are also some rich people who are fans of the Amiga and may be willing to invest
> some money to cover some start-up costs. I have seen some names of such people.
> Ultimately though, the most likely path to success would be through an arrangement
> with Imagination Technologies, makers of the MIPS processor.

I also state (in section 4c):
> The MIPS architecture is very popular in the embedded world. The higher end
>MIPS processors [...] are only used in a moderate number of chinese
> smartphones and tablets. [...] the mobile world, with smartphones and tablets,
> is built around the google android operating system and the android system
> is built for the ARM architecture. Google treats the MIPS architecture as a second
> class architecture [...]. Imagination Technologies needs an operating system
> which treats the MIPS architecture as first class to become more popular. [...]
> the smartphones and tablet [...] market is now saturated, [...]. The next big
> thing will be the boxes-pluged-on-the-TV market,
> this market [could] soon be dominated by ARM based devices.
> Imagination Technologies could instead push its own operating system
> and MIPS processor so as to become to the smarttv world what android
> and ARM have become to the mobile world. And the MorphOS team is able to
> supply them with the said operating system. This way Imagination Technologies
> would have an operating system, superior to android, to supply alongside the
> MIPS processor. [...] In exchange for being supplied with one of the best operating
> systems in the world, they could supply some advantage in return such as financing the
> development or, at worst, a discount on the IP licencing cost for the hardware project.

note: a more interesting arrangement could be: the MorphOS supplies an operating system for free to Imagination and in exchange, they design the SoC for free for the MorphOS team. Then, you can go to Acube or something, and partner with them for the design of the hardware, suggest to them that they design the board at their expense, as they already did for several Amiga-compatible boards, and receive a percentage of the final sales in exchange.

Finally, in the file, I also said:
> The MorphOS team could supply the said operating system without giving away MorphOS and here is how: see section 4d.
Please have alook at the said section.

@thread
The change of CPU to the MIPS which I suggest is mostly motivated by its potential within this general scheme to become mainstream again. If MorphOS is to remain a purley desktop based and purely hobbyist platform, it is probably best to stick with PowerPC for compatibility. While sticking to PowerPC would be my first choice in such a case, the MIPS would, however, be my second choice and the ARM my third. The x86 is a disgusting monstrosity which shouldn't be touched.

Finally, I want to put emphasis on the following point. When you want to enter a bulding, against peoples will, which is guarded at the front door, you don't force your way in. You enter through the service door at the back mixing in with the employees. It is the same thing for the computer market. You don't go upfront to comfront Wintel. You enter an underdevelopped market (Android became popular by entering an underdevelopped market), here I suggest the TV-world convergence box; you make sure to put a back door, here it is in the form of a desktop upgrade kit; and you use the backdoor to get into the place. Think of it, deception always wins over brute force. You can compensate a lack of funds by being smarter.

> the change of ownership has just happened.

...3 years ago.

> the networking gear manufacturers are switching from PPC to MIPS,
> the printer manufacturers are switching from PPC to MIPS

They are switching from PPC, yes. Have you got any sources backing up the claim they are switching to MIPS?

> the chinese are embracing MIPS as their favorite platform

They are embracing anything from ARM (too many to mention) over MIPS (Loongson, XBurst) and SPARC (FeiTeng) to Power (OpenPOWER CP1) and even Alpha (ShenWei). There's even a custom ISA called MVP.
Edit: As has been established meanwhile, ShenWei uses another Chinese custom ISA which does not resemble Alpha ISA.

> The real reason why PPC is loosing ground is [...] because the two main companies
> behind it are not trying to push it. Until [...] the creation of the OpenPower
> consortium, IBM only developped the Power Architecture for its own computers.

As you're indicating, IBM actually *is* trying to push (its implementation of) Power Architecture through OpenPOWER since 2½ years ago. Better late than never.

> Freescale [...] only care

Freescale does not care about anything anymore ;-)

https://morph.zone/modules/newbb_plus/viewtopic.php?topic_id=11100&forum=3

> about their three main lucrative markets: automotive, industrial and aerospace
> and do not care about the rest of the world. They almost respond to the loss of
> the printer and router market by shrugging their shoulders.

NXP (Ex-Freescale) cares about many more markets, including printers/imaging and routers/networking/telecom, just not (exclusively) with Power Architecture anymore but (also) with ARM.

[ Edited by Andreas_Wolf 09.05.2021 - 12:24 ]

> OK Andreas, OK. Power9 IS PPC.

...and so are POWER3, 4, 5, 6, 7 and 8 :-)

Quote:


All your convergence box things is pretty 2000ish. Markets have long moved on.
You know where the future is? Look for example to the new Samsung Galaxy S7. That's how the market evolves., Who cares about some strange set top boxes. They are yesteryears technology.

My guess for the future:
Real potent mobile phones, plus a few tablets will cater maintream consumer computer needs. Special purpose/geeks will stay laptop/big box.

The smartphone market is no option to target (unless you have quite some billions to burn).
Normal/classic computers will stay. The so called Wintel duo is lonhg gone. Intel is no more that tightly bound to MS. Intel is just a chip provider.
Why Intel above ARM? All those übercheap ARM solutions are usually pretty closed stuff (mostly allwinner SoCs). Intel is comparatively open and powerful. Is there any sanely priced cpu out there which can compete against an i7 in performance (total and per core)?
I don't see that. And why this "ugly" Intel? Well, because it just works. And as a coder I don't need to mess around with any x86 legacy stuff. The best compilers avaialble are for x64. It's the de facto standard of the IT industry. Why looking for problems when there are none?

Sorry, all in all I don't think your strategy is the right one.
And honestly I don't feel the urgent need to make MorphSO mainstream at all. I would very, very, very welcome a significant increase of users, but MorphOS doesn't need _millions_ of users to be nice. It's even nice already with just this handfull of users (albeit it definitely is too few!). If MorphOS NG gets say 20.000 - 50.000 users it can easily sustain in a healthy tiny niche which operates way under the radar (which has some pretty advantages).

[ Editiert durch Zylesea 10.03.2016 - 23:24 ]

>> Who is going to develop and manufacture this 2 GHz single-core P6600-based
>> SoC with PowerVR GT7900 GPU for the MorphOS team?

> The MorphOS team could take on board a chip designer, as there are people able
> to do this in the Amiga world. As an alternative, Imagination Technologies could
> be asked to do it (they supply services around their own cores). We may get them
> to do so in exchange for an operating system. [...] a more interesting arrangement
> could be: the MorphOS supplies an operating system for free to Imagination and in
> exchange, they design the SoC for free for the MorphOS team.

I don't think you know what you are talking about here. To me, you sound clueless regarding the resources required to design and manufacture a modern IC as complex as a multi-GHz SoC.

> The higher end MIPS processors [...] are only used in a moderate number of chinese
> smartphones and tablets.

Loongson has been intended for routers, netbooks/laptops, desktop computers, servers and supercomputers.

http://en.wikipedia.org/wiki/Loongson#Loongson-based_systems
http://blog.imgtec.com/mips-processors/loongson-mips64-processors-performance-barrier

@Andreas_Wolf

What did Trevor say the development cost for the x1000 was? And how much did every beta board cost to make? I can't remember but it was a lot, and he had to add that price to the board which made it even more expensive than it would have been with it's small batch number alone. People loudly complain about it's price but it seems to be fairly accurate considering. People just don't expect design, development and testing to cost that much. But it does.

Quote:

Again, incorrect.
This was discussed awhile ago.
Integration with one copy of Ambient for both boxes, possibly upgraded to hypervisor status, with pipelines and threads enabled for both boxes.
Not only could you maintain 31 bit compatibility, but you could farm some processes out to the 64 bit part of the system.