I've had similar problems measuring bandwidth. The greatest difficulty was the
limited resoution of the available timers. It's a darn shame that a high
resolution counter isn't available given all these fancy hardware. Milliseconds!
Pah! You can do an awful lot in a millisecond. Ever since the Pentium was
release there have been high resolution counters available - anyone written some
code to access them ? - back then it was all restricted access only! Does Intel
give a damn now ?
Jerry
Nathan Meyers wrote:
> Willi Richert wrote:
> >
> > I have written an ftp-client (connected to the ftp-server residing at
> > the same computer) which I tried to measure how fast it can handle I/O.
> > I used getCurrentTimeMillis() (or similar) which gave me for the same
> > config (retrieve 1meg, same file) sometimes values that differed by a
> > factor of 2.
> > Now, how can I get "valuable values". Maybe I missed a similar
> > discussion, so I would be thankful for every pointer.
>
> To explore a couple of obvious questions:
>
> 1) Are you sure the different values are wrong? Have you stood by with a
> stopwatch to compare what the program is reporting with what is really
> happening?
>
> 2) Is the time of what you're measuring long enough to make minor
> variations insignificant?
>
> Perhaps the most important caveat is to realize that a call that reports
> milliseconds or microseconds or whatever isn't necessarily accurate to
> the millisecond or microsecond... it's only accurate to the resolution
> of the underlying clock. If, to get a little extreme, your system ticker
> is counting off in tenths of seconds, then you're always going to get a
> time that's accurate to + or - 1/20th of a second.
>
> Nathan
>
> ----------------------------------------------------------------------
> To UNSUBSCRIBE, email to [EMAIL PROTECTED]
> with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
----------------------------------------------------------------------
To UNSUBSCRIBE, email to [EMAIL PROTECTED]
with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
