This utility is a class for timing PARFOR loops. It can be used to observe the various overheads that may exist in parallel for loops and the utilization of each worker. Various examples of using this utility can be found in Example_Script.m.

This class should be used in the following way:

p = Par(n); (1)
parfor id = 1:n
Par.tic; (2)

<usual computations>

p(id) = Par.toc; (3)
end
stop(p); (4)

plot(p); (5)

1. Construct a Par object, with the number of iterations as the input. This constructs the object.
2. Call Par.tic just inside the PARFOR loop. This records the start time of the iteration.
3. Call Par.toc just before the end of the PARFOR loop. Store the
output to the appropriate index of the Par object. This is necessary for PARFOR to recognize that the variable is sliced.
4. Stop the measurement. This records the final end time.
5. Visualize.

There may be some overhead in adding the Par construct. Because of this, the numbers you get may not accurately portray the true timing, especially for short computations.