As of tonight:
- Spectrum, which is very good - for 48K stuff I've got no bugs filed on it. Can't work out why but the 128K works for some stuff and not for others so far though.
- Megadrive, which is fairly good. Plenty of known small issues and it still needs battery-backed cart RAM implementing, but from the games I've tried it seems to have better compatibility than anything except Gens.
- ST is good too now I've put my 68000 core in (which does prefetch and ST-accurate timing). Runs most games and even the special effects in most demos. Sync-scrolling is the one thing badly busted.
- Atari VCS has lots of small issues and one big one (no bankswitched cart support)
- Jaguar is currently the bastard stepchild - I haven't even run it for a while. At the moment it won't work inside the same timing system as the other systems because the timing really isn't suited to running multicore. I think I've worked out what's wrong and I've got a route to a fix but it will mean a complete teardown of the timing system, so I need to know I've got several uninterrupted hours of coding before I can implement it. I think I broke the OP with a display system rewrite I needed to do to support border removal on the ST too, but that should be a quick fix.
- the C64 core doesn't count at the moment because I'm still gluing all the bits together. Should be reasonably easy though, as I've got the CPU, the video chip doesn't look to have the (different) complexities of the MD or the ST - the only tough bit will be the sound and the format support.
The Speccy and ST were 'easy' because I programmed them myself and know them pretty much inside out. The VCS and Megadrive I played a lot of, but never coded on. The Jaguar I only really knew T2K. The systems I didn't know have been really fun to do - the VCS for it's utterly mad 'control the raster gun' architecture, the MD because the custom hardware is so complicated (god knows how long it took me to get the FM working right!).
The C64 is a particularly interesting challenge because not knowing many of the games well it's harder to know what's right and what's wrong

.
I've got several other possibilities stacked up too - SNES is a strong contender (but is a big job because it needs a new CPU core) and I'm looking at the PC Engine and Amiga as well. N64 too, but that's a truly massive job. This project is there to keep me going for my emulation and spare-time coding needs for the next 10 years or so, so I've got lots of time

.