I got this knowledge really too late, but recently I've learned how the music is (was?) made on old computers like Atari 65XE or NES (the same processor 6502). The amount of work specified in the article above, is increased by the calculation of vsync of the monitor used, and correlating it with the sound frequency. This leads for example to the same game playing in different tonation on PAL and NTSC. Today it's already obsolete, but the emulator still has to be emulating the one or the other standard, to comply with the code. Today we have great privilege to abstract the sound from the monitor sync (by OS) but this is not the case in some embedded devices.