I've often wondered about this - especially as (according to the timelines on wikipedia) the Acorn Communicator was released before the IIgs ...It's a curious one, the 65816... iirc the reason for its creation was essentially to be a CPU that could do 24-bit addressing whilst retaining backwards compatibility with the 6502, because that's what they wanted for the Apple IIgs.
I might have thought one way to "expand" it might be to invest in a bytecode of some sort - Sweet16 springs to mind and I've used that lot in the past on some 6502 projects but there is also Acheron - http://acheronvm.github.io/acheronvm/I think if one weren't saddled with the requirement for backwards compatibility, then the 68000 would have probably been the more natural choice. As it is, the 65816 is an 8/16-bit chip with some 24-bit stuff bolted on, and the downside as you folks say is that it's a bit fiddly to program as a result. That's why the direction I've been exploring is to try to optimise the 6502's existing 8/16 architecture. I've mostly kept in mind the "ah, if only it were easier to do x!" moments I've encountered from my own assembly language programming. But that's not stopped me pinching a few of its ideas, as you'll see below!
And I think I have given up on the '816 now - having written far too much code for it now. Mostly a 32-bit bytecode VM which overcomes the banked memory issues to present a nice 32-bit interface - at the expense of speed obviously. I achieve an effective clock speed of about 200-300Khz on a 16Mhz '816 CPU and most of the cycles "wasted" are to do with the '816s way of reading a byte from RAM when in 16-bit mode.
And of-course, having PiTube does make it easy to make your fantasy come true by writing it in C or ARM and running it as a co-pro...
Cheers,
-Gordon
Statistics: Posted by gordonDrogon — Tue Apr 09, 2024 9:19 pm