Originally Posted by 8bitbubsy
However, the pattern offset calculation is rather slow... Hencing the "Seek" box when you jump to another pattern. What's up with that? Should be instant...
it shouldn't take longer than around 500ms to seek.. if the status bar shows the string "discarding frames" for a noticable period of time, possibly your cpu is overworked?
seeking to a random point in the song involves a silenced run of the emulated replayer code from the first vblank to that point. i have done this for accuracy.
a song may play a string instrument in one channel that never stops, or changes note (the dma isn't reset).. just a constant waveform loop. let's say we seek 3m 30s into the song.. the seek resumes play with an accurate representation of which part of the string waveform is being played at that point.
possibly there is a better way to do this?
but i can't imagine what it might be..
i had tried other methods like, flagging points in the song which the dma is either off/being turned on ("safe" points), and seeking from there instead of the first frame.. but it saved a mere 10ms, it wasn't worth the added maintenance.
additionally, and most importantly, the repayer code could be processing all kinds of things in those skipped 3m 30s. i want that processing to be represented with complete integrity.
Originally Posted by TheCyberDruid
Rather odd format though
i love dm's sound. perhaps rather loud in it's average output.. but interesting replayer and synth methods