Sound overhaul for TKG

[Karlos]

Originally Posted by patrik

Perhaps something more with the MMU setup is dysfunctional with RemApollo+MMULib. Can make it do speedy chipmem for that setup too by running MuFastChip separately and if that is slower than full MMULib, that would show if that is the case.

This is the case 5 version of the code

https://github.com/0xABADCAFE/tkg-mi...040/predelta.s

There shouldn't be any chip ram accesses until we get to line 221, where we write a single word to the volume buffer. We then do 4 long word writes to the sample buffer whether we use the shift-normalisation or multiply normalisation path. After those writes, we move on to mix the next frame of 16 samples.

[Karlos]

I can't understand how the single last write there would impact the next frame of data being mixed the way it appears to. There's some dimension to it I'm just not seeing.

[patrik]

Originally Posted by Karlos

I can't understand how the single last write there would impact the next frame of data being mixed the way it appears to. There's some dimension to it I'm just not seeing.

The dimension was that something else becomes dysfunctional when RemApollo+MMULib is used. With RemApollo+MMULib+MuFastChip, which rectifies the difference in chipmem read speed, I get the same slow result as my first test:

Test case 5:
	Mix : Delta Lookup (Pre-encoded source)
	Norm: Multiplication/Shift
	Info: Move16 fetch, target 68040

	Mixing  1 channel(s):   102186 ticks 189 packets
	Mixing  2 channel(s):   142704 ticks 189 packets
	Mixing  3 channel(s):   178237 ticks 189 packets
	Mixing  4 channel(s):   213528 ticks 189 packets
	Mixing  5 channel(s):   254121 ticks 189 packets
	Mixing  6 channel(s):   299491 ticks 189 packets
	Mixing  7 channel(s):   335369 ticks 189 packets
	Mixing  8 channel(s):   374262 ticks 189 packets
	Mixing  9 channel(s):   409171 ticks 189 packets
	Mixing 10 channel(s):   444816 ticks 189 packets
	Mixing 11 channel(s):   479108 ticks 189 packets
	Mixing 12 channel(s):   515068 ticks 189 packets
	Mixing 13 channel(s):   555132 ticks 189 packets
	Mixing 14 channel(s):   596589 ticks 189 packets
	Mixing 15 channel(s):   628154 ticks 189 packets
	Mixing 16 channel(s):   651306 ticks 189 packets

So please disregard this tangent, it was helpful for me to understand that my system was performing badly, but not really useful for the thread.