Some minor (potential) size opts

[emoon]

Hi,

I noticed some minor size opts that can be done as part of the sin-table generation

https://github.com/askeksa/Cinter/blob/master/player/Cinter4.S#L70

At this point it would be possible to do

neg.w d0

instead as d0 is -16384 after this loop.

Also here these two pointers (+ the offsets) points to the same address so actually one write is needed.

https://github.com/askeksa/Cinter/blob/master/player/Cinter4.S#L74-L75

Now given that the code above here

https://github.com/askeksa/Cinter/blob/master/player/Cinter4.S#L63-L64 looks identical maybe the compressor will actually do a better job if it stays the same so I'm not as sure about that one.

[askeksa]

Thanks for the optimization tip.

The first one is definitely an improvement. I experimented with a few different variations of eliminating the redundant moves, and what seems to work best (after compression) is to swap the two offset writes around (inside the loop) and keeping the middle 3 instructions of the block outside the loop, so it becomes:

neg.w   d0
move.w  d0,-(a1)
neg.w   d0
move.w  d0,CINTER_DEGREES/2*2(a1)

[emoon]

Nice! np :)