Conversion of Spectrum 128 music to Mockingboard

["TomCh" <tomcharlesworth@blueyonder.co.uk>]

re. BLuRry's posting in "8 bit 3-d game"
"http://www.brendandy.com/apple/demo disk.zip"
Fire-up AppleWin & BRUN CYBERNOID

My motivation was to really push the MAME AY8910 emu code, as the Apple
Mockingboard music (Skyfox, Ultima) wasn't really exercising it that
hard.

So as a little R&D proj, I thought I'd convert Dave Rogers' Spectrum
128 Cybernoid routine (written in 1988).

Here's a summary of how I did it:
. I got cybernoid.ay from
http://www.worldofspectrum.org/projectay/gdmusic.htm
. Split it into bin's with AYSplitR
. Disassembled with Inkland's dz80w (www.inkland.org)
. I wrote 6502 macros to replace the z80 opcodes
. For cybernoid, I hand converted the z80 code to 6502 (using the
macros)
. I added a few extension to AppleWin's debugger to help debug the 6502
code (ACME symbol loading & ZP pointer support).

I use Skyfox's MB detection routine.

The Z80 regs are emulated with zero-page memory locations $F0..$F8
The playback routine is very inefficient, as it:
. saves the ZP memory
. restores the Z80 regs
. runs the IRQ handler
. saves the Z80 regs
. restores the ZP memory

This allows playback to work simultaneously with Applesoft & ProDOS. If
DOS3.3 doesn't disable IRQs around disk I/O, then it won't work on a
real Apple (under DOS3.3), but it'll still work on an emulator :-)

I profiled the IRQ handler and IIRC, it takes about 20% of the frame on
average. This is poor, but the code can easily be optimised. Remember
this was really just a proof of concept.

After this, I wrote a python script to do the Z80->6502, and quickly
converted Cybernoid-II. Problem is this has a section of music that
doesn't sound quite right. It's one of 3 things:
1) AY8910 emulation
2) Subtle difference between Mockingboard & Spectrum's AY8910 interface
(eg. timing)
3) Bug in my converted code :)

I can't eliminate (2), as my Apple is currently knacked. Any one with
real Apple & Mockingboard h/w prepared to test this for me?

Tom