🚀 go-pugleaf

> Good remark, but I think the call to the function itself
> introduces a delay longer than the pause you wish. That's
> why I think also GetSystemTimeAsFileTime() can't give a
> _precision_ of 100ns although the _resolution_ is 100ns. And
> I'm quite sure a call to GetSystemTimeAsFileTime() requires
> a longer time then a call to GetTickCount().
>

That didn't sound entirely right to me, so I tested it. On my Athlon 900mhz,

GetSystemTimeAsFileTime takes    0.00000000958222 seconds to execute

GetTickCount takes               0.0000000135771 seconds to execute

I averaged 10,000 calls in each case. However, the basic premise that
GetSystemTimeAsFileTime isn't any better than GetTickCount is true. I also
recorded one second of GetSystemTimeAsFileTime timestamps and one second of
GetTickCount timestamps. In both cases, the resolution was exactly 10
milliseconds.

...
9:38.477
9:38.487
9:38.497
9:38.507
9:38.517
9:38.527
9:38.537
9:38.547
9:38.557
9:38.567
9:38.577
9:38.587
9:38.597
9:38.608
9:38.618
...

So both are using the same internal clock, which gets updated every 10
milliseconds on a single processor Win2K machine.