Notes

So you need to have 4 functions per task (qsort/msort/fib) - divide, combine, base, threshold
Divide should take an input and return two inputs
Combine should take two outputs and return one output
Base should take an input and return an output
And threshold should take an input and return a boolean

Divide is where you split tasks into multiple tasks, so fib(10) becomes fib(9) and fib(8), or ... divide(10) = [9, 8]
Combine is where you join them, so if the results from fib are 1 and 2, combine(1, 2) = 3
Base runs an input through the actual function to get the output, so base(10) = fib(10)
Threshold takes an input and determines if you need to split it into multiple tasks or if you should go straight to base

Note on threshold, it controls how parallel your operation is - say if you have fib(50) and your threshold is fib(30), that means any task under fib(30) is going to be run sequentially
If you have fib(50) with threshold fib(1), that means every task is going to be split into parallel before fib(1) can be resolved sequentially

So for fib, the only time you call it is in base
for msort, my combine calls merge and my base calls mergesort
for qsort, my divide calls partition and my base calls quicksort

No you have the set of DC methods for each task
You'll probably want a struct with function pointers to the divide/combine/base/threshold functions for each DC
then just give that to your scheduler thing


Like you'll need some sort of control logic to coordinate the worker starting/ending/returning the result