> The SID has long been successfully emulated

Emulated, but from what I understand, not perfectly. Mainly because the SID was a hybrid digital and analog IC - and it's that analog part that made it unique.

Also - I don't think (I could be wrong) that any successful reverse engineering of the chip was ever done, nor do I think that there were any or much documentation about the chip (that is, official docs)? It was very much one of those custom chips made for one specific purpose that wasn't used in any other product (this was a common thing in many home computers of the 1980s).

It was originally conceived as a multi-purpose synth chip. Bob Yannes, designer: "When I designed the SID chip, I was attempting to create a single-chip synthesizer voice, which hopefully would find it's way into polyphonic/polytimbral synthesizers. I never realized this however."

http://sid.kubarth.com/articles/interview_bob_yannes.html

Thanks both. I was relying mostly on the existence of the SIDStation some years back.

Actually the opposite. The demoscene has done phenomenal work squeezing all the tricks they can out of that chip. Some of its quirks have been used to great advantage in demos over the years.

AFAIK, you can't buy new 6502s any more (nor can you buy the 6510 actually used in the C64). You can buy 65c02 chips, but those are slightly incompatible with the 6502 and 6510 on the opcode level.

Not really significantly though. The Apple //c (and some ][es) had 65c02s rather than the 6502 in the ][ and ][+, but they could still run all the old software.

There are undocumented opcodes that still perform useful operations on the NMOS 6502, for example loading the A and X registers simultaneously, decrementing and comparing memory, along with a variety of more exotic instructions that can replace quite long sequences of equivalent "valid" opcodes. These are NOPs on the 65c02.

I know little about the Apple 2 series, but in quite some C64 software, these operations are used wherever they can save a few cycles. Mostly demos.