How much would it cost to port AmigaOS to x64?

[dreamcast270mhz]

I know someone whose a dev for DFBSD, and he entertained some q's for me about the OS in late '08. Can't remember his name for the life of me, but he lives in C'ville VA

[Iggy]

Hmm? I got my prescott core P4 Dell in '04, so thats eight years. And thats pretty sweet there, but since Netburst wasnt too efficient with its pipelines it doesnt surprise me.

Nope, Intel kinda blew it there.
Has to be the only time AMD actually slipped ahead of them (that ain't happening again).

[dreamcast270mhz]

AMD is a pretty good company however, and I like their stuff over much of the stuff Intel is pumping out

[Iggy]

AMD is a pretty good company however, and I like their stuff over much of the stuff Intel is pumping out

Yeah, well you can see where my sympathies are.
Take a look at what's in my x86 box (and for that matter my AROS box uses a 2,7GHz single core A64).

[NovaCoder]

If you set a reference hardware platform specification, it shouldn't cost more than a few million.

That's right, if they choose a reference hardware platform it would make it less expensive for sure.

The problem of course, is that reference platforms get out of date pretty quickly so it would take more cash to keep getting it ported to new reference platforms.  And if the port to a reference platform that was up-to-date when the work started but then took a couple of years to get out of BETA it would also be a bit silly.

There's also the question of what is considered a successful port, just the OS with no app support, the OS + PPC support, WARP support, Open GL support, x68k support etc etc.

[Jupp3]

That may be, but I wasn't intending to say that AmigaOS beats out Linux - just that Linux UI is an unholy mess of warring standards applied with wild inconsistency to the extent that my two-week attempt to switch over last November left me wanting to kill people.

Yes, that might be even worse on linux (or unix-like systems, in general) than it's on Amiga.

And worse still, it's not only about different toolkits either, f.ex. Qt is now split to "classic" widgets and QML.

But in any case, sometimes it's really bad on Amiga too. f.ex. ReAction had some totally non-standard behavior for shift + cursor combinations, at least in text editor (select text instead of moving to the end)

[Digiman]

I don't think OS4 has any viability on most X64 hardware, but I do think if we put it on something like Genesi's ARM boards, then it would be very viable as it would then be on low cost hardware which is powerful enough for most work.

By the time they port it 2.8ghz i7 CPUs+mobo will cost the same as ARM so waste of time.

[Digiman]

That's right, if they choose a reference hardware platform it would make it less expensive for sure.

The problem of course, is that reference platforms get out of date pretty quickly so it would take more cash to keep getting it ported to new reference platforms.  And if the port to a reference platform that was up-to-date when the work started but then took a couple of years to get out of BETA it would also be a bit silly.

There's also the question of what is considered a successful port, just the OS with no app support, the OS + PPC support, WARP support, Open GL support, x68k support etc etc.

Would need to be as complete a conversion as Apple's PPC-->x86 product if they want €100 for it ideally.

Supporting an i7 and a future i7 is no problem, forget bus/RAM speed or north/south bridge uniqueness. PCI-E is here to stay, as is basic i7 architecture. GPU is only possible exception.

The issue is (total potential sales x RRP)/cost of development up front.

If 10,000 copies sold at €100 that's €1,000,000. If they even sold 10000.

The real issue is nobody would invest in that business. If IBM bought it to turn it into a true virus resistant OS after massive improvements that is the only real world mass market scenario I could see for X86 OS4 IMO.

[dreamcast270mhz]

Theres more to porting it to x64 than ARM. Using the i7 as an example, you have several hurdles to jump:

OS4 doesnt support SMP, so only one of an i7's cores is used, limiting its usefulness.
Endianess will have to be addressed, when porting the OS you will have to rewrite any ASM code that may be present (If any)
OS4 doesnt support 64-bit, and since 64-bit instructions use twice as much RAM, it will be grossly inefficient
Locking the system to certain machines (like OS X is) will increase compatibility but cripple market penetration, as we are in a recession and people dont always have cash for yet another system, especiallly not an i7 system
Leaving it open to any x64 machines will cost much more manpower to ensure compatibility


Plus, AmigaOS4 has little to offer over current alternative OSes for x86. Comparing it to ZETA/BeOS/Haiku, for example, you will see more x86 native apps for it (since BeOS had only a short PPC run) compared to OS4, which will require its porting before any applications can be compiled/tested

And whose gonna pay for an alternative OS with inferior hardware support, that they have to relearn and with much less infrastructure than most other OSes on the market. The Amiga community? We are so divided I doubt you'd even get the number of current OS4 users,  many of which will be angry that their hardware is no longer supported.

You seem like your mind is made up Digiman, but you know what they say, the empty can rattles the most, good luck professing your beliefs, but i feel they fall on deaf ears

[commodorejohn]

OS4 doesnt support SMP, so only one of an i7's cores is used, limiting its usefulness.
Endianess will have to be addressed, when porting the OS you will have to rewrite any ASM code that may be present (If any)
OS4 doesnt support 64-bit, and since 64-bit instructions use twice as much RAM, it will be grossly inefficient
Locking the system to certain machines (like OS X is) will increase compatibility but cripple market penetration, as we are in a recession and people dont always have cash for yet another system, especiallly not an i7 system
Leaving it open to any x64 machines will cost much more manpower to ensure compatibility


Plus, AmigaOS4 has little to offer over current alternative OSes for x86. Comparing it to ZETA/BeOS/Haiku, for example, you will see more x86 native apps for it (since BeOS had only a short PPC run) compared to OS4, which will require its porting before any applications can be compiled/tested

It doesn't matter, technical issues are utterly irrelevant to these people. "x64 is the Only Possible Future of Amiga!" is largely a mantra uttered with religious fervor and zero comprehension as a prayer to the Great God Intel, Dominator of the Market. If the PDP-8 made a sudden and inexplicable return and overthrow of the desktop market, they'd be suggesting OS4 get ported to that.

[persia]

The problem with Power architecture is the price of entry, why should I have to spend 1900 quid on machine that in the X86 world is out performed by a couple hundred quid machine?  AmigaOS's are a hobby, as such they should fit a hobby budget.  At 1900 quid you get the truly devoted, but you don't get the vast majority of people who just want to play with it.  I actually don't even want another machine at this point, just make it run in a Fusion or Virtualbox instance.

[dreamcast270mhz]

I think we should just make an ARM port to something simple like the beagleboard, because its low cost, low footprint and open source. We already have an open source Amiga OS, AROS (which I do not care for, but many others do, thats fine). The thing is that OS4 has too many challenges on x86, on ARM, is a lot less and it would be more economical for us users.

If you happen to get loaded enough to buy OS4 then you may do with it what you wish, but until then we must take what we are give or not at all.

[jorkany]

It doesn't matter, technical issues are utterly irrelevant to these people. "x64 is the Only Possible Future of Amiga!" is largely a mantra uttered with religious fervor and zero comprehension as a prayer to the Great God Intel, Dominator of the Market. If the PDP-8 made a sudden and inexplicable return and overthrow of the desktop market, they'd be suggesting OS4 get ported to that.

Hmm, I don't think it's so much that people are in love with Intel as it is that products which just happen to use Intel are readily available and generally inexpensive. So yes, you're right, if there was another commodity leader then people would want to target that instead. I don't see anything wrong with that. I do see something wrong in leveraging the wallets of a few pensioners "just because".