Brien,
>By George, I think you've got it! (Sorry for the terrible pun)
You've come to the right person when it comes to terrible puns.< g >
>I reproduced these results. I am surprised of the difference the TALK setting made. Thanks for pointing this out!
>Another thread mentions that SEEK
ORDER is faster than SEEK . This is true and its speed is also not affected by the TALK setting.
>
>This info is going to change the way I develop. Thanks UT!
I think that it's worth pointing out, that anytime such disparate results occur, something else is going on with the OS. It was strictly by accident that I "stumbled" across this. My machine here at work has TALK set to OFF by default. The home machine has it set to ON. While running the test at home the first time, I noted during its running the messages appearing on the status bar. When I saw the results, that's when I checked TALK, noted that it was ON, turned it off and re-tested.
>>Brien (and Peter),
>>
>>I think I've found the answer. In my original test TALK was set to OFF. When set to ON, LOCATE and SEEK yielded approximately the same. However, when set to OFF, SEEK (using the same test) was about 325% faster on a P200/MMX w/64 Mb of RAM. This was with VFP 6.0 SP3.
>>
>>TALK should be set to OFF to produce valud results. Otherwise, screen I/O is factored in. Further, if one truly wishes to measure the relative speeds, SEEK should not have to continually open and close the index. In doing so, you measure file I/O as well and not the true performance of the function.
George
Ubi caritas et amor, deus ibi est