Thanks. I try

actually the IIGS uses a 65C816 at 1 or 2.8MHZ and comes with 256k to 1 meg of ram, expandable to 8 (a modified 65C816 is used for the super Nintendo console, which is a 16 bit processor, there is also a SCSI card available for an apple II, however unlike pc compatibles the apple IIGS expansion slots are not general purpose, each one can only accept a certain type of card, and some slots can only be used at the expense of certain on-board devices (trust me a friend of mine has a IIGS), not to mention the proprietary nature of the IIGS' firmware would make converting a full Linux os difficult at best, however perhaps a specialized port with a stripped down kernel (2.4 or earlier more then likely) with a specialized set of utilities (forget gnome/kde/xfce, think twm, mwm or possibly windowmaker) but doable perhaps.

there is however a project that you might be able to base a linux-like os for the IIGS
http://www.gno.org/gno/
never seen it run but i've heard about it

The 65C816 has essentially the same kinds of limitations as the 6502, except that the flat memory addressing limit goes from 256 bytes to 65536 bytes. By today's standards that is still quite limiting (equivalent to an 8086 CPU). However, the 65C816 is a better choice for a Unix/Linux like OS. The graphical environment is what really requires the power of a modern CPU. The actual Linux kernel and shell are efficient and lean enough to fit in a small memory footprint. The speed of the CPU might still prove to be inadequate for anything but small file-systems and programs with mostly integer calculations.

I've developed a lot of Motorola 68030 and 68000 software in the past and wouldn't mind doing something more with those. It is possible to get floating point emulation packages for the 68000 family. Those can intercept the unimplemented instruction trap and allow software to use the standard floating point op-codes without modification to the software. It is a bit slower than compiling emulation code in-line but not significantly slower.

It's really unfortunate that the IBM PC did not use the Motorola 68000 architecture because that is a lot better for flat memory models than the 80X86. The 80X86 also requires very complex optimization to use the special-purpose registers and instructions for the most efficient execution. The Motorola 68000 treats all registers equally, except for the distinction between address and data registers. Of course, one can argue that the 68000 is not suitable for the kinds of hardware optimizations done in Intel CPUs. I have to wonder if there might have been better, more general hardware improvements with a 68000 based architecture that benefited all CPU designs rather than just those using Intel architectures.

Because of the left-over baggage from earlier X86 CPUs, a lot of an Intel CPU's hardware is unnecessary for modern software. I think we may eventually see PCs start using software emulation for the 16-bit real mode and eliminate that completely in the CPU hardware. The BIOS would either be rewritten or executed by emulation software. I don't think 32-bit execution can be removed from the CPU hardware yet, but that is considerably simpler than the 16-bit segmented model. Microsoft has essentially abandoned 16-bit execution and 16-bit Windows emulation in their 64-bit operating systems.

I would like to see a CPU design some day that has instruction emulation facilities for each process or thread using microcode. That would allow any process to efficiently emulate one instruction set for virtual machine applications. The process or the OS would load the microcode for the emulated instruction set and then each thread would have a designated micro-code set for instruction execution. One could limit the number of emulated instruction sets based on microcode storage or some arbitrary number. This has been done before, of course, but never on a consumer personal computer (unless I count the IBM 5100 desktop computer).

I tried gno, it is working fine, but it is not stand-alone OS, it is installed and started as Prodos application. But as I already say, there is already Linux OS, which is working fine on 6502 CPU (6510 and 8502 actually, it is for Commodore 64/128):

http://lng.sourceforge.net/

lunix/lng btw is not linux, it is writen in 6502 assembly as opposed to C
it is simply a unix-like os

I agree. It is Unix clone. And everything written in assembly is much faster than anything else. Of course, if it is written by good programmer
About 18 year ago I saw Windows 3.0 clone rewritten in assembler and it worked on 5 MHz 8088 CPU much faster than the original Windows 3.0 on 386 and 486 CPU. Too bad I can't find my copy...
Anyway, I am looking for any Unix like OS, that is working fine on Apple II. As we already know, there is Minix for 8086 CPU. At this time 8086 and especially 8088 aren't so advanced CPUs, so there is no reason for any Unix like OS to work on 8080/8085, Z80 or even 6502 and 6800 CPUs. This is opposing to what Linus Torwalds said - that Linux can't be used on anything lower than 386 class CPU. I know about ELKS project - is is nothing more, but Linux, working on 8086, 80186, 80286 class CPUs: http://elks.sourceforge.net