Do you think that the simple variance in how the time might be obtained between the two systems could have impact here?...Also possibly variances in 'intervals' at which time is obtainable?
I suppose it is also possible that some OS task could have preempted one of the runs and not others, or even have occurred at different places vis-a-vis the test runs. Was the PC hooked up to a network?...Was there some other task alive but inactive?...Was an Internet connect alive during the test?...that kind of thing.
I presume that you recompiled between VFP7/VFP8 runs.
For a loop of 20 MILLION I would say in any case that the impact of the differences observed is negligible.
>Hi
>
>
>T1=SECONDS()
>PRIVATE i
>FOR I=1 TO 20000000
>NEXT
>? SECONDS()-T1
>
>
>I run this program on my Compaq EVO P4 2,2Ghz and on a PC Athlon XP 2000+.
>
>Result:
>
> | VFP6SP5 | VFP7SP1 | VFP8
>---------------------------------------------------
>P4 2.2GHZ | 1.232 s | 1.191 s | 1.503 s
>---------------------------------------------------
>AMD XP 2000+ | | 1.232 s | 1.171 s
>---------------------------------------------------
>
>
>VFP8 vs VFP7SP1 is 26% SLOWER on P4
>
>VFP8 vs VFP7SP1 is 5% FASTER on ATHLON XP
>
>Why ?
>
>Fabio