While and For Loop Performance

[StuartBruff]

Can't remember if this has been discussed before ...

I was messing around with prime number generation and hit a problem with a for loop, in that the iteration range definition (first, next..last) could sometimes mean next was greater than last. The immediate solution was to simply rewrite the for loop as a while loop. However, I noted that the performance was considerably worse.

As a quick check on the relative performance of the two forms, I put together the attached worksheet. It shows that a for loop executes much quicker than a while loop in M11, M13 & M14. (M14 is quicker than M11, BTW).

Stuart

[AlvaroDíaz]

I think that this what you want see:



This into winxp sp3 with mathcad 11 eating 23460 Kb, as report task manager.

I have the following rule: "if the iterating number is knowed, then must to be a for loop ... ". Next is the exception: .. except that you can vectorize."

There is a thing that I not understand: can't you check which is first,next...last before the for loop?

You testing file coul be more interesting if you put it in the "real" math world. Here an interesting example:

Try to write the usual procedure for the bisection method for the zero of a monodic function, i.e. using while abs(b-a)>TOL and then try to write one other using a for loop. Notice that the iteration number is given for this algorithm.

Then call the zero's routines for a lot of functions (probably a parametric family to not write dozens of code pages) or call over the same function a lot of times to find the same zero to amplify the bussy time.

That's as suggestion.

Regards. Alvaro.