One unified OS for the future?

[OlafS3]

@Thomas Richter
My whole arguments rests on two simple facts.

1. Given the choice, the common consumer rather buys another cheap piece of hardware (adding additional value to his household. Like hooking it up to another TV, placing it in the child's room, etc) than screw with an existing computer. This is particularly true for sub-100 USD products

2. x86 as a platform has plenty of user friendly OS' people can rely on. ARM users don't have as much choice, in that regard.

But is that not making a bet again? In the mid 90s PPC was the bet, at that time it seemed a "safe bet". We all know history. Apple did not do the same mistake, they kept doors open, others in our community not, they bet everything on PPC and lost. Now people are again demanding to set everything on the "next safe bet". Why are you so sure that ARM will win? If you loose your bet AmigaOS or MorphOS go from one dead end in the next.

Yes X86 has other OSs but people will compare it anyway. Or you create something like the Raspberry that is unbeatable cheap. Other than that I am of a similar opinion like Thomas. What we need is new software and for that we do not necessarily a completely new platform but hardware with geek factor that is different from competition (like FPGA based hardware would be). Even if we have a 64bit OS with full SMP and MP this would not bring automatically new software because it would still have not the same user base and software as Linux, Mac or Windows. And for new markets like Smartphones or Tablets our desktops are not suited.

So if you are so sure you will certainly bet everything you own on that? Not? That is what most do. They talk as long as they do not risk their own money or do the work.

[OlafS3]

Uhh the entire industry of computing is based on speculation - you can carry out years of market research in advance of a product release, but you seriously can't anticipate the impact of said product, or its reception. In the case of Android, which PRIMARILY runs on ARM devices, with x86 and MIPS making up the minority, it has been well-received. An Android desktop would likely be ARM based, as the hardware is cheap enough that it will still be powerful, but competitive. Intel/AMD x86 at the same price range as ARM either is too power-hungry or too anemic to even boot up. That being said, I'd trust the engineers of MS, Dell and AMD, all of which I have worked with in my past job as a data center tech, over a handful of users and developers in the middle of a forum known for sociopathic trolls, hell I had lunch with a manager for the largest AMD data centre in the DC area simply because he was called out to the MS data centre I worked at and invited me and the rest of the crew to lunch at a sushi bar and discussed what he wanted to see done to improve AMD-based Dell server reliability with us, over sushi and beer at that. He also talked to me about the Opteron-A series, and he said that it will, in his own words "Be the smartest move that HQ has made since launching the Opteron line"

You know as well as I do there are logistical and also other concerns with having to support two different architectures, diametrically opposed at that! Without a ports type system like FreeBSD uses, one or the other will simply have little software. Best to focus on one architecture for logistical reasons.



If you're going to make a brash statement, prepare something better than just an opinion: http://www.nvidia.com/object/tegra-k1-processor.html

This, my friend is a SoC which has the power of an Nvidia GPU and a quad-core ARM CPU. I can't find any benchmarks vs an i5 or i7, but most people I know are on the budgetary end of computers, either older gen i-series, older-gen AMDs, or the Pentium and Celeron series of CPUs. Your personal desktop is certainly *NOT* representative of what everyone else has. ARM has scaled up at a logarithmic rate that is even better than x86, I can't really explain it to you other than it simply doesn't waste any space on the die for any legacy-cruft that an x86 CPU does. Your desktop, my workstation and most other x86 devices start up in a 16-bit mode, and have to be initialised from there to protected and then long mode just to even boot a modern, 64-bit OS. ARM? It originally deployed with a 32-bit design and a 26-bit address space, but it can still run a 32-bit binary inside the 26-bit address space. x86 doesn't have this luxury, its address mode is locked to its execution mode.

Furthermore, I will use the same argument that Howard Roark made in Fountainhead, original text below:



Similarly, what is real mode, AKA 16-bit mode, in a modern x86_64 CPU for? To start the BIOS, mostly - the original BIOS comes from the original IBM-PC designs, based on the 8088 and 8086. This was reimplemented by competitors to become PC-compatible. Then the 386 and 486 added a 32-bit protected mode, using an undocumented opcode in the original 8086 design to initialise it. They retained the 16-bit mode to keep DOS running, they simply used extenders like DOS4GW. With the extinction of Windows 9x with the atrocious Windows ME in late 1999, the 16-bit real mode was effectively rendered obsolete. But this was kept and copied into the 64-bit world, where now you had to escalate to long mode from protected mode from real mode. And this is all because the industry decided to use an architecture which is an extension of a 16-bit reimplementation of an 8-bit copy of a 4-bit processor. Why the hell keep all this cruft? DOS won't even run properly on a modern GPT sliced disk, let alone a system with no drivers!

ARM is significantly more towards my ideal of legacy-free than x86, and with the ARM64 releases, they're using a binary translation layer to execute the older 32-bit binaries in microcode on the 64-bit CPUs. In addition, Amiga would do better on a dedicated piece of hardware that is both cheap, and cost-effective, x86 isn't that answer. AmigaOS has a new place in the media-centric world - its low resource usage, efficient memory management and user-centric design would make it a perfect small computer OS, being used in either all-in-one computers or small set-top box computers, and ARM excels in those applications. None of the NG Amigas utilise anywhere near the full potential of workstation hardware, and as I hate to admit it, the days of a large howling workstation are numbered. As we speak I've my Nocona workstation for sale, simply because it is too loud and noisy to keep on, my Challenge S is quiet enough for low-end server applications, the Origin does well for high end, my Octane2 and Beaglebone have been doing very well as my main machines for most applications, and where I need a mobile solution, my trusty Nexus 7 does the job. I simply don't really need x86 except for a few things, which I am considering getting a small low-power computer to do the job of instead.

please keep certain phrases out of discussion

What I have a problem sometimes is that it seems all is either black or white. X86 is evil, your preferred hardware is good, Linux is evil, your preferred OS is good and so on. Always extreme. If you are in business you must be flexible, you do not unnecessary set everything on one bet, you always try to have plan B. As a big processor producer I might have to act this way (we see at Motorola what happens if the bet is lost), expecially as someone concentrating on software (including OS) you have the chance to leave another door open. Aros is showing that this possible with supporting different platforms at the same time. Of course you still need adapted components like UAE for the specific platform and you need everything to be recompiled when you change ISA. Apple was more wise than others in that sense, they are still there but a lot of other companies are not. Over the years I have so many trends seen coming and going that I could not count them anymore. The IT industries is always producing lots of new "bets" all the time, the intelligence is to survive despite of this. I do not know with whom you are having lunch and I could not care less but only because someone has a big salary and a nice title on his card it does not mean that he is right.

And as a application developer (what I do in normal life) I do not really care about hardware or lowlevel OS development. More important for me is how many potential customers are there, can I be sure that there will be development in coming years, how simple and productive development is and so on.

And BTW phoenixconsole is already working on supporting ARM as a new target for Aros (has f.e. just published a new version of his distribution for Raspberry). So if you are really interested you should support this.

[OlafS3]

when you have something portable you are not betting, only if you set everything on one platform without any chance to change direction (like Hyperion and MorphOS team did)

[OlafS3]

Actually the hardware does matter. I think at least MorphOS and OS4 have a similar status quo, but a different long time strategy. AFAIU OS4 wants to stay on ppc and keep compatible to its current incarnation while bringing SMP and stuff like that to the OS. And while I doubt feasability of that approach it actually is their approch. MorphOS evolves and matures a little more in its curent incaranation on ppc but then, eventually, will do a sharp cut to a "MorphOS NG" with SMP, 64 Bit, full resource tracking, and maybe - maybe not - full MP. But this will come at the cost of direct compability. And that sharp cut will come with an ISA switch, too. Either to x64 (_my_ preference) or to ARM. AFAIK the ISA is not yet set into stone. If resources were huge, support for several ISAs for MorphOS NG would be also possible (but resorces aren't huge, hence it's only theory).
AROS (x86) kind of switched ISA (well AROS is available for many ISAs) already for the cost of binary compability while keeping API compability - with the disadvantage that SMP or MP are not present in AROS as they aren't in  OS4 or MorphOS today. Introducing SMP or MP to AROS would mess things up there as much as it messes things up in MorphOS or OS4.

Would it be possible at all to develop a "modern" amiga-based OS without sacrificing not only binary but also source compatibility? If everything would have to be broken what sense would that make at all? And what software could then run on it (except 68k in UAE)?

[OlafS3]

4GB clearly isn't enough. Especially in modern times.
My laptop only has 4GB RAM and it really can become an issue. Especially when reclaiming memory takes too long, there are memory holes in a piece of software like Flash Player or without cache optimization.

It's actually less of a problem for most 32 bit OS', since they usually assume that there's less RAM available. However, that isn't too say 32 bit systems can't be starved.

4 GB RAM is a issue? :confused:

Seriously all my newer systems (with Windows) and I never had a problem with RAM

[OlafS3]

That is true.  Maybe it will never be cost effective to "bake" new ASIC chips for an updated Amiga clone.  I don't know how much performance increase is gained from going from a fast FPGA to a new baked ASIC, or how much performance increase is "Enough".  It has been stated in the past that 10,000 to 15,000 chips would need to be produced to make any new ASIC chip cost effective, which is probably too many for our remaining community, but who knows, maybe with the improvements shown using FPGA's and the Apollo Soft-Core 68k CPU, plus a SAGA core, we will see enough former Amiga users become interested again to reach those numbers.

I have guessed that we only have about 1,000 to 3,500 active users counting all flavors of Amiga & Amiga Inspired platforms today, but others claim there are many more than my estimates.  It will be interesting to see how many users line up to purchase the Phoenix accelerators, so please do NOT keep your sales numbers secret, like so many other sellers of Amiga gear do.  We want to know how many units you produce and sell and how quickly they are sold, which will help all programmers know the size of the community they might consider writing software for really is.

I can understand secrecy about some things from Amiga companies, but have never understood why they refuse to release sales numbers.  The only reason I can think of to hide sales numbers is that the number of sales is so embarrassingly low, they refuse to let anyone know how bad things really are, for fear of having the community shrink even smaller.  I hope the Apollo Team will not follow the pattern of hiding sales numbers for their Phoenix accelerators, or the Viper boards from Majasta.

Edit:  Maybe a big Kickstarter fund would help get hundreds of Phoenix accelerators built quickly and eliminate the risk of producing more than the number you have buyers for them.  This could also help fund the design and production of additional Phoenix accelerator models for any Amiga models you have not already designed an accelerator board for.

+1

You answered why they propably keep sales number secret

[OlafS3]

Great. You summarize in a few lines of text what I actually hate about retrocomputing-minded people like you, here and on other IT fora. Your absolute knowledge about other people's needs about hardware specs and computer uses. Your "for most uses" clearly don't fit the needs of the person you're talking with, and, for your info, they don't fit mine either. My laptop, which incidentally is the machine I use to develop and build Icaros Desktop, is a 8-GB 64-bit Windows 7 PC hosting the Ubuntu Linux virtual machine I use to develop and all needed target AROS guests. Ubuntu VM takes 2 GB of RAM and every AROS VM at least 512-1024 MB each one. For my main job, however, I need Windows. Current 8 GB are fine, but I had to add 4 GB to the ones I got with the laptop at the beginning, since 4 GB only were plain not enough to perform similar tasks.

We're ending yar 2014 DC and we're now paying less than 50¢ per gigabyte on mainstream SSD devices, while 8 GB RAM modules generally cost less than 100 euros. There is no practical, no economical, no moral need to save clock cycles and memory cells anymore. There's no need to be afraid of paging files and memory protection: our SSD, but even our fastest hard drives, can perfectly live with them, fastly and reliably. Being so conservative in resources can be good for embedded applications, but neither with mainstream operating systems, nor with Android, nor even with post-Amiga OSes, we're even remotely targeting to embedded uses. Scalability can be good, but we're definitely using our computers to perform more and more resource demanding tasks. If you still think 4 GB are "just enough" for today tasks, it simply means you haven't ever worked with huge images, with HD movies, with virtualization, with most of CURRENT "computer tasks" that 15-20 years ago we could just dream about.

I use PC for programming and have never had any problems with 4 GB. I do not use it for professional video editing and when I did that I would definitely not use anything amiga-related. Even new games do rarely need more RAM so, 4 GB is enough for "most" users not power-users using it professionally. Many users even use tablets instead of desktops, certainly not for video editing :9. But who here does that at all and where do you get the "power-software" needing that to use on amiga? People use their computers (or tablets) for web-browsing, email, facebook and so on and gaming.

[OlafS3]

It does matter, indeed, if you were performing other tasks while using your text editor. For instance, if you were compiling your sources on cpu #2, downloading files and running other tasks on processors #3 and #4, having your stupid text editor crashing on cpu #1 and brin ging all other tasks to death with the rest of the OS, actually MAKES the difference between amigoid and unixoid OSes. Are you really still performing a single operation every time? Curious to hear, from the "we have had the first and most powerful mainstream multitasking machine in computer history" kind of people.




And they would work far better if they only could be parallelized on different CPUs (or CPU cores, that's quite the same), as any other operating system already proved, without any chance of saying the contrary.




It does not depend on software bloatness, but on how tasks you need to open in order to get your results as quickly as possible. This is the exact reson why we're talking about etherogeneous computing for 5-10 good years now, using GPUs to perform parallel computation instead of CPUs. And yes, 2048 cores are far better than 1024 for chemistry simulations and scientific computation, as like as 4096 would be better than 2048 and so on. And, believe in me, software running on Tesla-based servers are far from being bloated, since the algorhithm sent to every stream core must be as neat as possible.

There's also, indeed, a break even point for "normal" CPU core parallelization on home computing tasks. But this heavily depends on user needs as well. The more tasks you open, the more CPU cores you'll need to keep responsiveness, altohough the rest of your hardware should also cope with that (configuration balancing). Having 16 cores would be pointless without a huge amount of RAM and a good disk subsystem, since some tasks would end up filling available resources and place others on the to-do list, waiting for resources to be available again.




As I said, I love people pretending their poor computing needs should be just enough for everyone.

"As I said, I love people pretending their poor computing needs should be just enough for everyone."

LOL

Why starting to be polemic?

Counterquestion

Why do you think your "bloated" needs are representing everyone? I am just a application developer, perhaps I have no clue...

[OlafS3]

I use my virtual machine for the same purpose (programming, if we can call this way what I do) and I need just 2 GB of RAM. This does not make me any better and does not absolutely means 2 GB should be enough for everyone doing the same things.

But I specially love your bold statement, and the following reprise: "who here does that at all and where do you get the "power-software" needing that to use on amiga?". I can only answer: you won't EVER find the needed software on amiga to do that, if you still keep resources and goals on the lowest possible bar. Lack of ambitions is part of the issue. I would really love to have at least QEMU ported on Icaros Desktop, and SMP added to AROS because my ambitions are higher. I guess all this discussion started because AmigaOS/clones lovers do have the same ambitions, and finally start feeling sad these ambitions are still... well... delusions.

I am realistic, ambitions are all nice but we see at Arix what happens when amibtion is too high. Perhaps we should be more realistic, when full SMP is not possible (or too complicated) why not doing something less ambitious. I would have wished Arix project would have (at first step) only added missing drivers, that alone would have helped a lot. And if "real SMP" is too much why not doing it less complicated (even if it is not "true" and needs adapted software). More than 4 GB are only needed for professional graphic software (including video editing). In reality there are not many people doing that (even on their work systems). But even if you say 4 GB is not enough for "everyone" you can also say it is only relevant for a minority (that needs it professional).

[OlafS3]

My "bloated" needs are just higher-resources hungry ones than others. I perfectly know I don't represent the normality, but I hereby constitute a case where yours/Matthey's/Biggugn's target resources for the OS are just too low. Obviously I wouldn't use a classic Amiga or a FPGA reimplementation to virtualize Linux, but I wouldn't keep the OS underpowered and "cut off" to the bare minimum needs of 68K apps just because they may be "enough" for most people. I would love the OS being right for ALL people, no matter how low or high hey needs are.

I start being polemic because I don't like, really, this "it's right for me, it should be right for everyone" attitude. I consider it utterly unpleasant.

"yours/Matthey's/Biggugn" have a different goal. It is a fun system, a new platform based on 68k certainly not competing with Windows or Linux or Mac. People like you are dreaming of a high-end platform in the same league as those big players. The air is thin in that league, who do you think will port the professional applications to it using that resources? I had contact to former amiga developers, mostly even Linux ports are financial failures. Applications are either on windows or for mobile platforms (expecially games). The other source of new software might be to motivate freeware/shareware developers but for those it is more important to have better dev tools. More than 4 GB is completely irrelevant there. And even more than one core is certainly only used by professional software.

[OlafS3]

No, you're not realistic at all. 4 GB and 32 bits are not enough anymore. single processing is not enough anymore. lack of memory protection is not enough anymore. inability to parallelize tasks is not acceptable anymore. It may be still somehow true today, it won't absolutely be tomorrow, and since - for instance - ARIX is not available today, but it will (hopefully) tomorrow, your target shoud not be todays, but tomorrow's ones. Today games use more CPUs, use huge amounts of memory, expecially for hi-def textures on consoles, and keep resources under an "acceptable" level only on Android devices, but just for the fact they don't share the same specs of computers and consoles.

If you have looked at the consoles, that are mostly multi-million projects now (mostly exclusive tied to one platform). Do you think that such a super-Aros or super-AmigaOS or whatever would be even taken seriously by these developers? As I said, more and better development software will bring much more software than support of  more than 4 GB.

[OlafS3]

Bloated is relative. What is definition where efficiency ends and bloat begins?

C64 with 64 kB RAM used to be plentiful. Then Amiga 500 with 512 kB RAM, then with 1 MB. Then Amiga 1200 with 2 MB and now we are pondering is 8 GB bloat or not.

Bloated is indeed relative. It depends what you want to do with a system. If you want to play in first league competing with Windows or Mac (in applications) and see it running the newest games (in competition to the newest consoles) then 4 GB are not enough in future. But I do not see big chances to get there and compete in that field. So better to build up a niche and to get more software other shortcomings are more relevant.

[OlafS3]

Sure, if you can trust every single piece of software that you run then you don't need memory protection.
 
 I don't think I'd trust it for online banking or entering my credit card details anywhere though.

You trust any amigaoid for that? You are a optimistic person :-)

Seriously all our OSs (including NG) are completely open. And even if there would be some protection and the banking software would run on it I would like to see how you explain what you uses to the banking people. The bank will always try to say it is your fault and they would certianly use it against you. I do not online banking in general but if I would do that I would use a protected Windows-PC or Linux or Mac for that.

[OlafS3]

Teet editor was used just as an example. And please remember because there is no MP some other program could crash it. Perhaps MP3 you are listening to has broken frame your MP3 player can't decode and trashes data in your text editor. Not likely to happen but it is possible.

I am ok with it but it can't be denied how useful MP is.

Yes of course MP makes sense but the problem is that we have what we have. I am not a lowlevel guy but if I understand it right adding what people call "modern" would break everything. 68k software would run on UAE anyway but would it be possible to compile the newer software to it? And right now it is partly easy to port 68k software (as long as it is not hitting hardware or is including assembler parts) to NG, would it still be possible to do that with such a modernized platform? If not what sense it would make for the people (except feeling more amiga than a themed linux)?