for loops longer on a P-IV???

[Matt Gerrans]

Additionally, you might want to try a similar test in C or C++.   Note that
this could be a side-effect of the granularity of the Sleep() function.
For instance, the following little snippet of C++ on my P-III demonstrates
that the Sleep() function usually rounds up to the nearest 10 ms:

   DWORD zzz     = EditSleepAmount->Value,
         started = timeGetTime(),
         slept;
   Sleep( zzz );
   slept = timeGetTime() - started;
   LabelResults->Caption = (AnsiString)"That *actually* took " + slept +
                           " milliseconds.";

Anyhow, it is possible that the P-III is faster than the P-VI -- I think the
first round of P-VIs that came out actually had worse performance than their
predecessors.