>I've tested like
>

>FOR ln1 = 1 TO 10
> ?SYS(2015),seconds()
>ENDFOR &&ln1 

>
>
>I see two hits per milisecond, with increasing sys(2015)
>Since there are two per milisecond there must be two within the same millisecond, even considering rounding. (I'm not sure if there is something that counts smaller then a millisecond, so the result shown could be rounded)

There isQueryPerformanceCounter, and this one counts. But you have to keep QueryPerformanceFrequency in mind to get at the ms...

>Now the only problem remaining is that this valid ony on a single thread :) Multiple threads still can create the same return value.
>
>Simple program
>create table
>

>CREATE TABLE murks (c1 c(10))
>use
>

>and a prog snippet
>

>USE murks SHARED again
>FOR ln1 = 1 TO 100000
> INSERT INTO murks VALUES SYS(2015)
>ENDFOR &&ln1 

>Simultaneously run this snippet on two instances of VFP on the same comp.
>This gives 200000 records
>
>now test
>

>SELECT COUNT(*) as ncnt, c1 FROM murks GROUP BY 2 HAVING ncnt>1
>

>
>I come up with 32787 hits ... 16%

Guessing that on a single 1 core CPU the single comp would show no duplicates. To lazy to test by setting which core to run on....