WDC Re-releases Classic 1986 6502/65816 manual “Programming the 65816”
amazon.com
amazon.com
The reason why this is an interesting chip is because if offers a window on what an 'alternate universe' might have looked like if Motorola and Intel had had a third viable contender when the CPU wars were still in their infancy. Unfortunately the chip never found itself in a machine that was spread out widely enough and with powerful enough hardware that its potential could be seen. And that's a pity because it had some pretty nifty features, the processor that it is easiest to compare to was the 80286.
Likewise I believe the 'direct page' (same as the 6502's zero page, but relocatable, like in the 6809) is limited to the first page, the lowest 64k of memory.
Unfortunately it's not an architecture for which a C compiler is a natural fit.
I'm sure WDC _could_ have developed such a competitor. It's interesting that even now they haven't done so. They still apparently sell millions of dollars of 65xx cores ever year, I don't know why they've never made a truly 16 or 32-bit 6502.
Of course, given how badly Apple screwed up the Apple III, it's unlikely even the 65816 could have saved it.
A 65c816 with Jay Miner's Amiga/Lorraine chipset would have been a natural successor to the Atari 8-bit series.
My impression is the Apple IIGS was deliberately handicapped by pressure of the Jobs/Mac partisans at Apple. It could have gone out with a higher clock and been a contender.
I do wish something had evolved so we could have something to power the low-end PC market in the original 6052 price range.
As neat as it was (I remember it fondly -- I'd love to buy one for myself now), the IIgs was a distraction and a dead end. The Apple II division had been starved for resources forever, and the IIgs was too little, too late.
Just viewing it from a raw computing power perspective: by 1986, it was clear that you needed more than 16-bit addressing capability by the end of the decade.
The 65816 had even less general purpose registers than the 8086, the Direct Page stuff seems like it would be a pain in the ass to use manually and a nightmare for compilers and the instruction set was already saddled with a rather awkwardly implemented backwards compatibility with the 6502 (set this two processor flags and now you are running on a 8bit processor instead of a 16bit one!)
I guess 65c816 kept that speed and its simplicity and offered an upgrade path. But only the IIGS and the SNES used it.
Direct page is pretty cool when you consider that the cost of a memory access was 1 cycle. So it's effectively 256 extra registers (tho still only one ALU).