the fact is that with a carefully crafted workloop you can do a lot of "multitasking things" without the call tree getting its knickers in a knot. This eliminates the use of "stack saves" and all the othere OS thingies that makes the chip/C 'uncomfortable'

solely speaking of using the '51
My absolutely preferred method is to use a processor that is fast enough that every 'task' is executed in a timely fashion without any prioritizing.
My second choice is the one I mentioned in http://www.8052.com/forum/read.phtml?id=157305
If both the above fails (as often happened in the days of the old 'steam driven' 8 MHz 12 clockers, when, in addition to using slowpokes, we used the '51 for 'strange things' for reasons of economy) I resort to things like what was discussed in the last few posts.

Erik