Subj : Re: which for loop is better?
To   : comp.programming.threads
From : Sean Burke
Date : Sun Mar 27 2005 07:32 pm


Elie De Brauwer <elieNO@SPAMde-brauwer.be> writes:

> Sean Burke wrote:
> > ssam_iit@yahoo.com (sam) writes:
> > 
> > 
> >>David Hopwood <david.nospam.hopwood@blueyonder.co.uk> wrote in message news:<dfL0e.1131$IH3.327@fe3.news.blueyonder.co.uk>...
> >>
> >>>sam wrote:
> >>>
> >>>>Hi, 
> >>>>   which one of the following codes in C would you prefer to print
> >>>>those "hello"s and why?
> >>>>int i,j;
> >>>>1) for(i=0;i<=99;i++)
> >>>>    for(j=0;j<=999;j++)
> >>>>      printf("hello");
> >>>>
> >>>>2) for(i=0;i<=999;i++)
> >>>>     for(j=0;j<=99;j++)
> >>>>       printf("hello");
> >>>
> >>>As if it matters. What a stupid question, teaching people to micro-optimize
> >>>without thinking about whether it can possibly lead to any difference in
> >>>perceptible performance. Whoever set this homework question should be
> >>>ashamed of themselves.
> > 
> > 
> >>I should've elaborated the problem. Here it goes : for the same number
> >>of total iterations(100*1000 in this example) in nested-for-loops, is
> >>it more effective to have the outer for loop iterate more than the
> >>inner one (case 2), or the other way around (case 1)?
> > 
> > 
> > 
> > The days when runtime was determined by the number of instructions
> > executed have long since passed us by. On modern machines, performance
> > is governed by cache misses.
> > 
> > And please don't top post.
> > 
> 
> But in this case (using a compiled, non interpreted language) won't the 
> number of instructions be the same ?

You obviously weren't around in the days when

   for (i = n - 1; i >= 0; i--)

was faster than

   for (i = 0; i < n; i++)

-SEan

.