Re: [stella] 2600 Sound, that time of year..

Subject: Re: [stella] 2600 Sound, that time of year..
From: "Andy Mucho" <andy@xxxxxxxxxxxxxxxxxxxx>
Date: Fri, 1 Aug 2003 20:37:22 +0100
Paul Slocum wrote:
> Yeah, I suspect it will have to be limited to title screens, but it'll be
> an awesome title screen!

I hope so, whatever uses it :)

> How much RAM and ROM does it use?

RAM - It eats the entire zeropage except for 5 bytes at the moment, but
that's because I've put the code in there to save about 15-20 cycles.. It's
one giant self modifying animal, but can be made to run from ROM, with a 3
channel piece eating about 45 bytes of RAM.. There's a lot to remember ;)
But only while its playing.. And of course this drops linearly with the
number of voices.. So a 2 voice tune could eat as little as 30 bytes ram

It's worse on the RAM front than I'd like, but because of having to break
all the sequencer processing into tiny little, less than 40 cycle states, I
end up with this constant trade off between speed and ram.. To little ram
and I eat too many event states to do something simple.. Too many event
states and the whole sequencer processing takes *WAY* too long.. It's
already too long..
I aim to get a 3 channel complex tune, running in about 30 scanlines of
processing when interleaved with the music synthesis.. But this timing is
worse than it seems, because I always have to allow for the worst case with
regards the number of event states processed per frame, even though some are
going to be idle most of the time..

ROM - The current player eats about 900 bytes of ROM, but that's *very* lazy
there's a lot of wasted space, but then there's obviously the amount of
memory for your tunes, which I've tried to make quite compact... Could
probably easily fit into 512 bytes or so..

The actual main sequencer actually fits into 256 bytes ( it has too ;) ) but
there's a lot of other fluff that comes with it, like actually setting up a
song, and various other ancillary functions..

When I've got the EGs working (I've just sacraficed 3 bytes RAM to make them
faster), I'll try to put something more musical into it and send another
.bin to the list.. The last one was a terrible programmer music test, and
see what the consensus is.. I've got a nasty feeling the PWM method just
won't cut it flexbility wise..

We shall see, or should I say, hear :)


Archives (includes files) at
Unsub & more at

Current Thread