A Moogle  🏳️‍🌈 💻 🎮 🎶 🇸🇪

I love 68k!

I'm not sure I really *love* any CPU architecture other than perhaps SPARC. But they all have their warts, even SPARC. Still, I appreciate how easy 68k is to reverse-engineer. I'd be up a creek without a paddle with regard to the CD-i's CDIC otherwise.

I could absolutely talk your ear off about SPARC, though. It's the closest thing to a perfect set of opcodes that I've ever seen. It's light and streamlined, with an abundance of synthetic opcodes that can be constructed from a relatively small handful of real instructions.

I haven’t done SPARC assembly, just MIPS. I think the last time I did hardcore 68K reverse engineering was trying to get the 1.44M mode working on MAME’s SWIM chip in the Mac II driver.

At the moment I'm trying to reverse-engineer every function in the Philips CD-i 220/F2 player's BIOS that pertains to CDIC access. It's going rather well so far.

The CDIC comprises a gate array, an ADPCM decoder IC, a DSP IC, and 8192x16bits of SRAM, and is wholly undocumented. It's now emulated well enough that most non-MPEG CD-i games work fine, but there are stragglers, and I don't want to push until they all have working audio.

Ooh, yikes. That’s a lot of undocumented equipment. I remember having to search through patents to figure out how the RSP depth buffer worked.

Just being pedantic, but RDP. RSP was the vector coprocessor, the RDP was the rasterizer itself. And yeah, it's a lot of undocumented stuff, but the current emulation clocks in at under 1400 lines. The alternative option for fully-working CD-i in MAME is objectively worse.