Fleecy Moss on AmigaOS 4 Release Date and More

[Floid]

Lando wrote:
I'm disappointed that AmigaOS won't be available in fully 64-bit within the next year but at least he says there'll be a transitionary version in the mean-time that will run in 32-bit mode with enhancements for 64-bit systems.

This is pretty much Standard Operating Procedure with every OS out there.  Viz the interim MacOS Jaguar for 970, and what the first Windows for AMD64 will look like (though unlike Apple, they're more obfuscating about what it will/won't do).  

Linux and *BSD (FreeBSD) are making a bit more of a jump, because they can - less 'penalty' for trying native builds, if you don't like it run the 32-bit tree(s), and so on.  (That's with an eye to AMD64, though; I can't remember if the PPC970 needs a bit more 64-bit-mode love somewhere in the kernels before you can get it back to running straight 32-bit code.)

So basically, not really a surprise... and we should remember that one big impact of 64-bit is in the VM system, which is fairly new for us, still something of a work in progress, and so forth.  (In other words, tradeoff in terms of development hours; do we want it running with *all the features we want* on 32-bit first, or do we want to delay those for the testing and so-forth to ensure it works/scales portably across 32- and 64-bit systems?)

The Amiga Generation 2 Visual Services Technology sound very impressive also.  It seems these are slated for OS4.2 - he doesn't say whether this is the "transitionary" 32/64-bit AmigaOS that will be released in preparation for moving to 64-bit though?

It's not really that hard to read; he's said 4.1 for "AG2" - at least its introduction - and nothing particular about when we can expect 64-bit.  Methinks some 64-bit hardware has to hit the market from a non-Apple vendor before anyone can allow themselves to care...

...and if they're going to pull AG2 off, they can try to keep that code 64-bit clean from the start.

No idea if they can pull this off, but it makes sense.  Some platforms will beat them to 64-bit, but... You could probably port something like MS-DOS or CP/M to 64-bit rather easily, but Linux 2.4 on 32-bit is still going to be more 'useful,' y'know?  Bad metaphor, it'd actually be neat to see AmigaOS as a sort of small, light 'MS-DOS' that can blow video streams or whatnot out of 8GB RAM... but since that chunk of memory is still rather expensive for the home user (have they even sorted the standards for cheap-unbuffered-mass-market gigabyte DIMMs yet?), there's at least a year's time to worry about it.

Anwyay... 4.0 early next year, followed quickly by 4.1 and 4.2, the transitionary 32/64-bit OS (4.5?) and a fully 64-bit OS (OS5?) by August!  Very impressive, looks like Amiga Inc have turned the corner!

Man, maybe I haven't been reading these lately, but that sounds a little 'optimistic,' no offense.  (We saw what it took to get 4.0 rolling, once it's out, some inertia will be overcome, but still...)

From the pace of things so far, I'd keep my hopes up for a 32-bit 4.1 or 4.2 with an evolved VM (again, not to knock the existing work, just that we know it's supposed to keep improving) and some of these 'AG2' features following "within" ("about") a year from 4.0's launch.

But that'd still put us on par with Win'9x and NT/XP in a number of ways, so it's not something we'd have to complain about.  If the next-generation APIs are really well thought-through, then that'll help things 'rocket' toward the capabilities we want.  ("64-bitness," the 'mythical' portability of OS5, and so on.)

Edit: Removed one of the redundant uses of "y'know," so I sound like slightly less of an idiot.

[Floid]

Just to clarify...

*IA64 (Itanium) has been the special case out of the current crop of 64-bit chips, in terms of "instant OS support."  But that's because it's been practically incompatible with IA32 (something they're *now* planning to address in hardware, I gather; it's been what, six, eight years?) ... and despite all the effort 'poor' Microsoft and friends put in, Intel's only been selling like 3,000 units a year.  ... Because, guess what?  Few third-parties have recompiled *their* applications to the platform.

*From there, practically every other 64-bit platform to date has been UNIX of one flavor or another (VMS being the exception).  Like it or not, that supports a slightly different development model... and certain *other* aspects of UNIX have long kept third-parties targeting it ready to recompile or port at the drop of a hat.  As it turns out, this enforces some professionalism in that market (software vendors have to know enough to 'stay on the treadmill,' or they'd have long since fallen off), but given some of the moans targeted at Linux on the AmigaOne, I doubt many of us would really be happy with 'UNIX' (as stood ~1990-2000) on the desktop.  No matter how cool SGIs seemed at the time.

OS X is addressing some of the issues; from my perspective, DragonFly BSD is set to give them all a knockout punch; desktop Linux has gotten around many of them by promoting vendor-maintained package repositories.  So I do encourage the haters to try things again in two years, while at the same time saying "Yeah, on 'the desktop,' you want software you bought six years back to 'just run.'"

*64-bitness offers big fat wads of memory addressing.  That's basically it.  (Some registers get wider, but in tradeoff, all your addresses and pointers or whatnot tend to get wider too.  It does help out with a few specific calculations, but yes, 32-bit chips are basically "just as fast," and could remain so as long as we can keep scaling them by the GHz.)  Big fat wads of memory addressing do let you do some cool things, though... like storing entire CDs in RAM.  Or decent chunks of video.  Or artwork for a building-covering banner at 300DPI.  Or game data that includes an entire topographic map of New York state with one-inch resolution or something.

Like I said, it'll be a while until we can afford this much RAM, so as users, we don't need to fret much right now.  But when 64GB is as cheap as 64MB is now, you'll hope we got around to having the option open, right? ;-)

[Floid]

Dietmar wrote:
>Why not? Infact while we're at it let's use a really good and polished 16 bit OS?!

You have to understand the concept of bits. 16 bit allow values from 0 to 65535 (64 KB). That's obviously not ideal for a polished OS.

Exactly.  Just remember that we have to question what what 'polished' is from one year to the next.  There was a time when window gradients seemed ridiculously 'expensive' - especially before and during the period when it still took $500 of add-ons to get the average machine running a browser stably.  

Now the machine itself costs $500, and we want a bit of eye-candy to satisfy our sense of aesthetics and act as a natural 'demo' of our hardware's capability.  (I used to pick on my Mac-using friends about this... "Sure, you've got shading on the window borders, but how long does it take to boot?")

Who's to say we won't finally see basic voice controls, or as we're already seeing (finally!), scalable graphics become part of the 'necessary' user experience?

For the business end of things, the trick is to introduce these features *when* there's a hope they'll integrate quickly and unintrusively on hardware your users can afford.  MS learned how to play this game by the time they introduced XP - it's huge, it's bloated, yet the features (and misfeatures, like MSN Messenger integration) -- and the extra 256MB of RAM to run them -- are far, far more "affordable" to the average user than the $300/16MB it cost  to get OS/2 running suitably in '94.

In retrospect, I'm still happy I spent those $300, but I did have to turn voice navigation (let alone dictation) off to get some reasonable speed from the system.

So let's agree to hang back at 32 bits for now -- and kick some ass despite our 'limitations,' since our 64-bit brethren won't be able to afford much better outside the server farms -- but agree to revisit the idea in a year or two, lest we become a bunch of bus arch trolls convinced the status quo can never be improved upon.   :quickdraw: