Hans van Pelt wrote:
-------------------------------
How do your tasks allocate RAM space for parameter storage?

- - - - - - - - - - - - - - - -
At assembly time. My objective is speed and power... not universal applications. At assembly time, I know all the tasks and equipment I must interface.

That might not be included in your definition of RTOS, but I call it a background tasker and this keeps the tasker loop very small (only a few bytes) and thus very quick. I don't us CALLs and RETs; this avoids deepening the stack and wasting time pushing and popping. It is an interrupt period after all.

Mutally exclusive tasks can be identified at assembly time and their resources overlapped for greater efficiency.

The background tasker gives all I/O the apearance of running concurrently. The foreground through its logical interface with background, decides to what it will pay attention and when it will do so. Foreground can initiate tasks and checkup on their completion status whenever.

IrDA or such protocols look easier with this sort of tasker because it handles the protocol state machine without bothering foreground processing. Foreground doesn't need to know what phase the background task is in, only its termination status.

aka j