PPC accelerator poll

[SamuraiCrow]

@Varthall

The fact that the 68k is dead doesn't change the fact that the Amiga has been dead for years anyway but we still keep using them.  Besides, softcores are more software-upgradable therefore a larger FPGA is going to be more cost-effective than fixed logic.  Especially when you consider the costs of trying to buy a system that's fast enough to emulate a 50 MHz 68060 vs. having a 100+ MHz 680x0 softcore.

PPC doesn't have a suitable JIT in its versions of EUAE, after all, and still has no backward compatibility for the chipset in it's OS JIT.  Factoring in for the 30% code-bloat associated with PPC's RISC instruction set (which translates into poor cache performance), this leaves the choices at 68k or Pentium.  I'd choose 68k if it can be made competitive per clock (and the N68050 softcore is well optimized in that aspect).

[SamuraiCrow]

I'd rather it if someone brought out an Amiga-style keyboard wedge case that fits a Mini-ITX motherboard and stick in a Sam440 to run OS4, or an iMica to run Aros.

Since Aros is making its way to 68k Amigas we're going to have a major OS update soon, meaning there's even less reason to upgrade to PPC just to use a newer OS on our classics.

Even if we got a new PPC accelerator, the Sam still has USB, RTG and 16bit audio, fast ethernet and a whole bunch of other features that would be slower on a PPC-powered classic Amiga.

I agree with this.  There is no point in putting a MiniMig and PPC on the same accelerator since you'd just be using a few CIA chips in the host motherboard.

Implementing CIA chips in the FPGA softcore would be better because then you could leave the old hardware behind.  You'd only need a Kickstart 3.1 ROM image to run the new system under AmigaOS 3.x and if you really NEED PPC compatibility, give it a Sonnet Crescendo G4 accelerator in a PCI slot.

The only point I'd disagree upon is that AROS will be a major upgrade for 68k.  AfaOS already backported most of the features of AROS to AmigaOS 3.9 a long time ago.  A few features of 3.9 need to be implemented in AROS before I consider it useful on 68k:  bit-planar support for AGA or ECS would be chief among them.  Otherwise you'll still be forced to use a graphics card on all AROS apps the same way as AfaOS does now.  I think that will come eventually though.