KT>   I am currently taking an algorithm analysis course were we must time
KT>different implementations.  The thought of the day is to use the 
KT>"clock()" yet with very small time increments, the functions returns
KT>zero.  Not a lot of help since the program had to take sometime!  The 
KT>Unix camp is running a program called "time" prior to the
test program.

   The problem with time is that it will time the loading of the executable
which may vary significantly depending on fragmentation of hard drives.  If
you want accurate timing then run the algorithm hundreds of times in one
timing session to ensure that the time taken is well above the minimum
timing interval.


   cya
___
 X MR/2 2.0 NR X Me?  Buy from MICROSOFT?!  Not Today!  Not Tomorrow!

--- Maximus/2 2.02