PegasosIII should become the next high-end Amiga!

[Ranchu]

I really don't understand why so many Amigans pine for a PPC based platform.  Is it just to be different?

Developing a modernised Amiga compatible OS seems to be enough of a challenge without the added headache of developing a custom hardware platform for it also.  Why not take advantage of the x86's enormous marketshare, product range, unmatched desktop performance and commodity pricing?

The old arguments against x86 from back in the 80's and 90's don't apply anymore.  For example: -

- 32-bit protected mode, which virtually every modern OS/app now operates under, is quite sane and comparable to other architectures.

- The AMD K5 pioneered an instruction decoding front-end where complex x86 instructions are decoded into simple micro-instructions that are handled by the "RISC like" execution units.  This same technique was refined with the Pentium Pro (P6) and subsequent generations.  Although instruction decoding logic had a high overhead initially, this now accounts for minimal overall die space in recent generations.  Interestingly, other "RISC like" architectures have adopted similar approaches in recent years.

- Rename registers and caches are another technique designers have used to get around x86 architectural limitations.  In fact this can provide more effective than having a large number of general purpose registers (another characteristic of "RISC like" designs).

- SIMD extensions have effectively replaced the archaic stack-based x87 floating point unit in almost all cases.  SSE2/3 is a very good set of ISA extensions.

- Finally with x86-64 we have an excellent 64-bit implementation that addresses some of the other remaining architectural issues.  The number of general-purpose registers is doubled from 8 to 16 (all 64-bit wide naturally).  Incidentally, 16 registers seems to be an ideal compromise between having enough to perform most tasks quickly while still allowing reasonable context switching speeds.  Support for some remaining "legacy" features have been removed such as virtual mode and segmented addressing.

AROS on x86 seems like the "ideal" solution for Amigans.  An open source Amiga-like OS for the most powerful desktop hardware on the market.  If only more developers would get behind it!

[Ranchu]

@dammy

I selfishly wish that the AROS team wouldn't get side tracked with ports to EKIFA and other architectures (including x86_64 to some degree).  I would love to see them focus on x86 native and Linux hosted and develop in a way that it can be ported to x86_64 down the track when it is really required.

The time spent porting could be better spent writing and improving device drivers.  This would improve AROS's reach much more than creating ports to obscure platforms.

EFIKA may be a neat toy right now, but the fact is that PPC is going to be relevant only for embedded applications.  The same thing happened to other once great RISC architectures like MIPS and ARM.

Actually I don't really see the interest in EFIKA.  Aside than embedded applications (what it seems to have been designed for) the small form factor makes it neat as a media PC.  But with only 400MHz/760 MIPS it doesn't have enough power to handle standard-def MPEG4 content let alone hi-def formats.