Heya Sorin ...
ah -
the concept of a 'test bed' ..

what's your test bed look like ?
and - moving forward - what criteria for any one else's test bed
would be significant from a measurement standpoint ?

I suggest:
1. OS name/version/SP
2. amount o RAM
3. details on cache / VMM / swap file settings
4. complete disclosure on OS config files
5. complete disclosure on VFP config files
6. LOC
7. Number o Loops / Iterative processes [per module]
8. any OUTSIDE drivers required
9. LOG file of runtime [from sysinternals FILELOG]
10. driver types for data at test time [odbc/ole-db/native]
11. disk type
12. hd interface type
13. cache on HD
14. cache on HD controller

I would suggest any test on W2K with 256M ram to be significantly different than on W2K with 1Gig ram, for example, as VFP is a memory hog [uh - .NET is as well on the server side of CLR]

if you make a RAM DISK for the temp space, that's skewing the results and that 'test bed' shouldn't count.

--Nota Bene - I'm not singling *YOU* out for this - but any real test ? should at least have *THOSE* metrics available, or the results will seem un-objective [at least in me small wee heid] ...

I'm sure there's the same kind o content for blot net tests - so lets see what other 'named' metrics will surface that match what VFP can do [that is measure-able] ..

Note To JayVeePee - not your baliwick, this time - you don't OWN the hardware.


mondo regards [Bill]
ps - that list wasn't inclusive - but is a good place to start .