Formula One emulation bugs

[Red2010]

• Incorrect Musical Instruments

When playing in NTSC, the music may load inappropriate instruments and sound different than in other emulators.

• On-screen Bugs:

The image is definitely self-explanatory.

[Clownacy]

This is a strange one... the audio issues are not caused by inaccurate audio emulation, but rather the wrong data is somehow being uploaded to the YM2612. If the instrument data is extracted and played in other emulators, it sounds the same as in this emulator. For instance, the title screen music is meant to make use of FM5 and FM6, but, in this emulator, those channels never receive any data.

[RealMalachi]

Maybe the game uses mirrors of the YM registers? I've confirmed in one of my own drivers that changing yma1 and ymd1 to 4012h and 4013h respectively works on hardware and various emulators, but not clownmdemu (v1.5). IIRC the mirrors end at 5FFFh, similarly the psg mirrors end at $C00017, but they might have more mirrors alongside the entire VDP

[Clownacy]

I can't recreate the issue you're having: Sonic 2 modified to use 4012h and 4013h for A1 and D1 works fine in v1.5.

[RealMalachi]

I did my first test on a 68000 driver, I just couldn't be assed to type it in 68000 address space because I assumed both CPUs would have identical address spaces. Did some tests on the the Z80 and yeah it handles the mirrors just fine
Although I'm not actually sure if Formula One uses a 68000 driver, since the v-int is kinda hard to follow

[Clownacy]

Ohhhhh, right, the 68k memory map was wrong, preventing the YM2612 from being mirrored. Thanks for finding that.

Unfortunately, this does not fix Formula One (and the debug log would have pointed out if it was accessing invalid addresses like that).

[Clownacy]

This game used to sound fine in older versions of ClownMDEmu. It was broken between v0.7.1 and v0.8 by commit ca3637c (particularly these lines).

Worryingly, all these lines do is make 68000 instructions use accurate durations. I can only imagine that the game has some kind of awful race condition, and that these changes to the emulator's timing were enough to cause it to create a bug. Other than that theory, I am stumped as to how this commit could result in broken audio.