Pros and Cons of Amithlon type systems

[B00tDisk]

I can do it, I write more complicated things for a living, I'm just seeing if it's really worth the time investment.

The main missing bits are the glue between the emulation and AmigaOS that give you native x86 code and direct hardware access.  Luckily those are pretty straightforward after seeing the headers you compile against and the patches to UAE and the Kernel.

The things I'm still weighing out are how much to emulate vs. pass through to the Amiga side (all PCI? SCSI? USB?) and which OS to use as a base.

Bernie was already rethinking how devices were handled for v2 and seemed to be going more emulated to increase compatibility. (No special Amiga side drivers to write)

Linux is a proven host but isn't interesting IMHO.  The big benefit there is that Bernie already patched the kernel so that you could get a lot of the memory mapped to the correct locations for a speed increase, but is that even needed with multi-core computers with memory faster than the CPU's Amithlon was designed to run on?

There are just a lot of variables to take into account before I decide to commit to it so I thought I'd gauge interest.

Bernie has mentioned that you can compile and run x86 apps on the Amiga emulated by Amithlon - I'd always hoped that meant (back when Amithlon was active, mind you) that someone would tackle some of the "must have" applications and we'd get the best of all worlds: an Amiga OS and killer, modern apps.  

If somehow someone managed to revive Amithlon and make that again possible...wow, that would be neat.

[B00tDisk]

Of course, thinking on it, part of the problem is that Amithlon was started in 2001 - when the core of Amiga OS was 7 years stale already (I don't consider two shareware bundles "service packs").  The difficulties involving bringing the Amiga back into a modern realm was merely "almost completely" insurmountable.  In three more years, Amiga OS - the 3.1 branch - will be twenty years dead.  The problems now are not "almost completely" insurmountable, they're entirely insurmountable.

It goes beyond mere hardware support, too.  You can write drivers until your fingers fall off to support every video card, every motherboard chipset, and so on, in the book and it doesn't change the fact that your target is possibly 2000 users.  Rare and weird strains of Linux have larger dev groups than that, forget user base sizes.  Then there's the issue of applications - once you have Amithlon updated to run on new hardware, once you get there, what do you do?  Run Aweb?  Ibrowse?  Or start scratch developing apps or porting apps from scratch from the x86 side?  Taking how long with the abovementioned tiny developer's side?

Don't get me wrong: I think Amithlon is neat.  Hell, I think the Amiga is neat.  I think its awesome that AOS 3.x has USB2 support, that someone can tweak out an A600 to perform like a mid 90's m68k mac (and in fact run MacOS 8.1).  When someone shows that their Bodega Bay chassis is still kicking over and they've got a video card, USB controller, CD-ROM etc. etc. that's awesome.  Likewise running AIBB on Amithlon and watching that red indicator blast up past the fastest pure 68k (and PPC!) systems just brings a smile to my face.

In a perfect world I'd wake up tomorrow and fire up this PC running Amithlon and type in replies on a ported and running emulated or native and running through an x86 interpretive layer version of Firefox, while the underlying host bits of Amithlon did cool stuff like handled multi-user, MP and VM, sandboxing the parts of AmigaOS 3.1 that would have a fit over such behavior.  But this is a far from perfect world.  And anyway, what I just described sorta-kinda already exists in having WinUAE - except there's the chipset support if I needed it.