Neal,
OK, I'll zip next time..
I'm sure you've noticed that if you only need to write strings, there is
room for optimalisation. Because data can have \00 in it, there is need for
interlocking (the "writing" tag, that "Wait For Write" is waiting for). If
you only need to write one string, windows will probably do this
interlocking for you, saving several ms per run.
Anyway, another thread (previously posted) is asking for memory mapped
files, so the code is probably getting used. (Nothing better that home made
code being put to use).
Regards,
Wiebe.
"Neal" wrote in message news:404ddb60@newsgroups....
> I had already taken out the 1ms wait when I ran the tests. I attached the
> VIs I used for testing both the UDP and memory mapped methods. I ran it
> again today and although my UDP tests were a lot slower than the last time
I
> ran it, they were still 2x faster than the memory mapped tests. The only
> reason for the UDP slow down that I can think of is the fact that the last
> time I tested it I was in an airport and was not connected to a Network.
>
> P.S. If you post another VI please zip it first. VI files get blocked on
my
> computer for security but zipped files still make it through.
>
> "Wiebe@CARYA" wrote in message
> news:404d951a@newsgroups....
> > Neal,
> >
> > There's a "wait 1 ms" in "Wait For Write.vi". It is there to prevent
100%
> > proccessor load, but it should not be executed the first iteration.
> >
> > After adjusting the VI, I get a typical 10X speed up (16X for read, 9.1X
> for
> > write)!
> >
> > Regards,
> >
> > Wiebe.
> >
> > "Wiebe@CARYA" wrote in message
> > news:404d91b3@newsgroups....
> > > Hi,
> > >
> > > That's a supprise to me. The memory mapping can only be used with
lot's
> of
> > > overhead, but I guessed UDP had even more overhead... It seems not to
be
> > the
> > > case...
> > >
> > > Thanks for letting me know.
> > >
> > > Regards,
> > >
> > > Wiebe.
> > >
> > >
> > > "Neal" wrote in message
news:404c9b95@newsgroups....
> > > > Thanks for your example code. I tested it against the UDP code that
I
> > > wrote
> > > > and the UDP was about ten times faster than the memory mapped file
> > method.
> > > > For speed reasons I think I'll stick with UDP. If I ever need a
> simple
> > > > Windows method on a computer that has no networking support I will
use
> > the
> > > > memory mapped method. Thanks again for your help.
> > > >
> > > > "Wiebe@CARYA" wrote in message
> > > > news:4044b740@newsgroups....
> > > > > Hi,
> > > > >
> > > > > Here is an example. Test Mapping.vi and Test Mapping.exe
communicate
> > > with
> > > > > each other. Test Mapping.exe can run multiple instances.
> > > > >
> > > > > Please let me know how it works for you.
> > > > >
> > > > > Regards,
> > > > >
> > > > > Wiebe.
> > > > >
> > > > >
> > > > > "Neal" wrote in message
> > news:40449905@newsgroups....
> > > > > > I'm now using UDP and it is working well. However, I am
> interested
> > in
> > > > the
> > > > > > memory mapped file method you mentioned because I may have
> problems
> > > with
> > > > > UDP
> > > > > > if there is no network card installed. Could you please send an
> > > example
> > > > > of
> > > > > > the method you mentioned?
> > > > > >
> > > > > > "Wiebe@CARYA" wrote in message
> > > > > > news:40431acb@newsgroups....
> > > > > > > Hi,
> > > > > > >
> > > > > > > After doing some test with this method, I think it might be a
> good
> > > > > > solution.
> > > > > > > There are some disadvantages, but also some advantages.
> > > > > > >
> > > > > > > Disadvantages:
> > > > > > > Windows only.
> > > > > > > One PC only.
> > > > > > > Uses API's.
> > > > > > > Low level (unless wrapped).
> > > > > > > Memory errors when changing control type (during development).
> > > > > > >
> > > > > > > Advantages:
> > > > > > > Fast, little overhead.
> > > > > > > Full control over synchronisation etc.
> > > > > > > Can also be used from other languages.
> > > > > > > All applications can read or write to the "global" at will.
> > > > > > > Use as many applications as you like.
> > > > > > >
> > > > > > > If you're interrested, let me know.
> > > > > > >
> > > > > > > Regards,
> > > > > > >
> > > > > > > Wiebe.
> > > > > > >
> > > > > > > "Wiebe@CARYA" wrote in message
> > > > > > > news:403f0d05@newsgroups....
> > > > > > > > HI,
> > > > > > > >
> > > > > > > > You could use the windows (MS only) way; by using memory
> mapped
> > > > > files...
> > > > > > > >
> > > > > > > > Regards,
> > > > > > > >
> > > > > > > > Wiebe.
> > > > > > > >
> > > > > > > > "Neal" wrote in message
> > > > > news:403b6c93@newsgroups....
> > > > > > > > > I have several built LabVIEW applications that I need to
> > > exchange
> > > > > data
> > > > > > > > > between when they are running simultaneously on the same
PC.
> > > Data
> > > > > > > Socket
> > > > > > > > is
> > > > > > > > > probably one of the best ways to accomplish this task but
it
> > > > > requires
> > > > > > > some
> > > > > > > > > extra installation and file handling that could complicate
> > > things
> > > > > for
> > > > > > > the
> > > > > > > > > end user.
> > > > > > > > >
> > > > > > > > > Does anyone have a better suggestion for communicating
> between
> > > > built
> > > > > > > > LabVIEW
> > > > > > > > > App.s?
> > > > > > > > >
> > > > > > > > >
> > > > > > > >
> > > > > > > >
> > > > > > >
> > > > > > >
> > > > > >
> > > > > >
> > > > >
> > > > >
> > > > >
> > > >
> > > >
> > >
> > >
> >
> >
> >
>
>
>
