I'll cover all SNES special chips:
DSP-1/1A/1B, DSP-2, DSP-3, DSP-4: all NEC uPD77C25 cores. We have 1B so far. We really, really need the DSP-3 because the game that uses it is completely unplayable with our HLE simulations. We could definitely use DSP-4 as there are lots of differences. And DSP-2 would allow all games to be playable. It would be nice for completeness' sake to get the 1 and 1A, but technically all DSP-1 games run with the revised DSP-1B.
I am raising money to get the DSP-2,3,4 dumped. If you would like to donate, please see this thread:
http://board.byuu.org/viewtopic.php?f=16&t=1249
Cx4: this one would be really nice to have as well. Our high-level emulation of it is terrible. The code is half floating point, half fixed point; and very inaccurate with no timing. We continue to have new bugs pop up every couple of months, Jonas Quinn just fixed a starfield effect bug two months ago.
ST-0010, ST-0011: these seem to be identical sans the program ROM. The ST-0010 is emulated well, the ST-0011 nowhere near enough to be playable.
ST-0018: this one is apparently an entire RISC CPU, with a hardcoded program ROM inside of it. It most likely has the same Shogi AI program that the other SETA Shogi games use. This will likely be by far the most difficult core to emulate. This is a 21MHz 32-bit processor, it's more powerful than any other SNES coprocessor, leagues beyond even the SuperFX2.
SuperFX, SA-1: these are actual processors, the games upload their own programs -- there is nothing to dump. They aren't 100% emulated perfectly, but then again nothing is ever perfect in emulation.
SPC7110, S-DD1: these chips have some complex decompression algorithms, and some simple extras (MMC, etc), but nothing timing-related at all. It all happens immediately upon request in real-time, and our algorithms are bit-perfect. Emulating their internal ICs would net us absolutely no difference in emulation, other than a slower framerate and greater complexity.
S-RTC, OBC1: these are beyond trivial. Absolutely no timing, absolutely nothing to them. They aren't even DSPs, there won't even be an internal ROM of any sort. It'd be like emulating the original Pong.
SNES S-DSP: this supposedly has an internal program as well, but it's 100% bit-perfect and timing-perfect in simulation. Hard to justify trying to emulate the microcode, it'd only slow things down and make the emulator require a copyrighted program to play *any* game at all.
SNES S-PPU1, S-PPU2: mostly the same as above but even less likely to use microcode. Could be hardcoded logic.
There's lots of other stuff, nothing that has any internal programs that I am aware of or that would really be useful from where I stand: BS-X base, BS-X town cart, Exertainment bike control software, hotel SFCBOX, NSS, etc.
Execute summary:
- DSP-2,3,4: very important
- DSP-1,1A: nice for completeness' sake
- Cx4: important but difficult to emulate
- ST-0010, 0011, 0018: would be nice
- everything else: not really necessary
Also as an aside, I looked into Air Rescue. It's definitely a uPD77P25, a PROM variant, same core. There's almost no actual code inside this image, and the data ROM is completely unused. It apparently does very little, and MAME can apparently play the game already with only twenty lines of boilerplate code in system32s.c. The comments in that file make it sound like it's used as a protection only. Even for MAME, the prospect of adding an 8MIPS DSP emulator just for a protection only used on a single game seems unlikely, but if you guys want to, my uPD77C25 core will be public domain. Feel free to port it over for MAME/MESS.
Also, the game is creepy as all hell, especially when you rescue and release prisoners.
|
Index 
Flat 
Re: Dr. Decapitator: Maybe ready for Round 3?
Reply
