>>Unfortunately VFP only reports to thousandths of a second and I suppose that rounding may have some effect (if rounding is done).
>
>Since each is for 100,000 executions, my concern is tempered. But it still seems to be a step in the wrong direction.
>
>Is there a sensible explanation for this?
>

Jim, if you do tests that don't take advantage of the things VFP 6 in a Win32 environment will do, such as take direct advantage of >16MB of RAM, then FPDOS is very likely to be far leaner - the control structures are simpler, fewer complex calls to things like the Win32 API, and specifically to the GDI, are made, which would require significant attention.

If you start doing things that take fuller advantage of things going on inside the Win32 environment, then you'll likely see significant improvements in the VFP side tests, because it interacts more directly with the Win32 services.

You can strongly bias your tests to favor FPDOS by doing a few things:

Place all files on the local drive
Set your system's performance to favor allocation of memory to cache
Work with relatively small files
use a fully defragmentsed swap file of fixed size on another physical drive, preferably on a separate channel
Run full screen
Run in a separate VDM (NT/Win2K)
Disable your screen saver while a DOS VDM is in foreground
If using 2.6, modify your PIF to only allocate ~15,500KB of DPMI memory for the DOS session, with 2.5, disable DPMI/XMS and make ~15,500KB of EMS memory available for the DOS session.