Originally Posted by DarrenHD
I think the real test will be - will the music sputter or click or pop if you are also doing other heavy, cpu intensive things, like web browsing, lots of disk activity, or lots of graphic activity?
Hopefully it will be rock solid.
It won't have zero CPU load because the data still has to come from the hard disk drive and travel through the system to the zorro bus (or clock port) and out to the prisma, so I estimate minimum 5-10% cpu usage on a 68060 just to get the data "out the door" so to speak. It could be more if coming from a PIO hard drive.
You will never get the audio rock solid if you're doing lots of disk activity. If the process can't get the data (because another process is chewing up the data source, especially the slow IDE bus if you use that) then obviously the card won't be able to carry on playing. What's important in this case is that the card cleanly stops until data is available again.
There are ways to lessen the problem, though - by changing the disk buffer size you can minimise disk access, and make it load in larger lumps of data at a time. Of course this requires more RAM (Can't have your cake and it, after all).
It should be less than 5% of a 68060 as the IRQ is pretty optimal now (it has to run on a 68000, after all), but when the disk activity kicks in that'll all change - not a lot I can do about that if you're using IDE.
I'm currently hooking the IRQ into a CIA timer, but I can also use the VBL which may make things more reliable (but it's a bit less "clean").