RE: Plotting with 3.2.0 on Windows is SLOOOOOOWWWW

[Tatsuro MATSUOKA]

Sorry again 

        if (isempty (str))
#          sleep(0.05);

Then 
N = 100;
n = 2;
m = 2;
ts=cputime()
for i = 1:n*m
    x = rand (2,N);
    subplot (n,m,i);
    plot (x(1,:),x(2,:)); 
end
cputime()-ts

gives
about 2 seconds


        if (isempty (str))
          sleep(0.05);
The same test gives 
0.8 seconds

Anyway further test should be done one core computer and I will do it at home.

Regards

Tatsuro
--- Tatsuro MATSUOKA wrote:

> Sorry
> 
> I have manually change the code
> and 
> > > !         if (isempty (str))
> Is forgotten.
> 
> octave.exe:11>for i = 1:n*m
> >     x = rand (2,N);
> >     subplot (n,m,i);
> >     plot (x(1,:),x(2,:));
> > end
> octave.exe:12>cputime()-ts
> ans =  0.76563
> octave.exe:13>
> octave.exe:13>N = 100;
> octave.exe:14>n = 2;
> octave.exe:15>m = 2;
> octave.exe:16>tic
> octave.exe:17>for i = 1:n*m
> >     x = rand (2,N);
> >     subplot (n,m,i);
> >     plot (x(1,:),x(2,:));
> > end
> octave.exe:18>toc
> Elapsed time is 3.2813 seconds.
> 
> cputime is surely reduced on the computer in my university (cpu is 
> HT-pentium).
> 
> I will test again at home.
> 
> Regards
> 
> Tatsuro 
> I will check again at home
> 
> Regards.
> 
> 
> --- Tatsuro MATSUOKA wrote:
> 
> > Hello
> > 
> > --- Olli Saarela  wrote:
> > 
> > > I was looking for a CPU consuming loop in the C++ code, but there might
> > > actually be some in the script files. Does this help?
> > > 
> > > *** __gnuplot_get_var__.m~      Tue Jun  9 23:22:14 2009
> > > --- __gnuplot_get_var__.m       Mon Aug  3 13:52:35 2009
> > > ***************
> > > *** 125,135 ****
> > >         fflush (ostream);
> > > 
> > >         str = {};
> > >         while (isempty (str))
> > >           str = char (fread (istream)');
> > > !         if (! isempty (str))
> > >             str = regexp (str, "OCTAVE:.*", "match");
> > >             str = str{end}(8:end);
> > >           endif
> > >           fclear (istream);
> > >         endwhile
> > > --- 125,137 ----
> > >         fflush (ostream);
> > > 
> > >         str = {};
> > >         while (isempty (str))
> > >           str = char (fread (istream)');
> > > !         if (isempty (str))
> > > !           sleep(0.05);
> > > !         else
> > >             str = regexp (str, "OCTAVE:.*", "match");
> > >             str = str{end}(8:end);
> > >           endif
> > >           fclear (istream);
> > >         endwhile
> > > 
> > > 
> > > On my multicore machine the tic-toc timing is practically the same, but
> > > cputime() shows a significant speed-up.
> > > 
> > 
> > First 
> > > !         if (isempty (str))
> > > !           sleep(0.05);
> > > !         else
> > 
> > I have tested on computer in my home (cpu Celerron M 1.3GHz)
> > Why 'else' is here? 
> > With this 'else' plot was failled.
> > 
> > I remove 'else' and tested argument of sleep from 0.05 to 1 with task 
> > manager active.
> > Octave consumed 99% cpu time while loop of this test. 
> > N = 100;
> > n = 2;
> > m = 2;
> > tic
> > for i = 1:n*m
> >     x = rand (2,N);
> >     subplot (n,m,i);
> >     plot (x(1,:),x(2,:)); 
> > end
> > toc
> > 
> > 
> > Regards
> > 
> > Tatsuro
> > 
> > --------------------------------------
> > Power up the Internet with Yahoo! Toolbar.
> > http://pr.mail.yahoo.co.jp/toolbar/
> > _______________________________________________
> > Help-octave mailing list
> > address@hidden
> > https://www-old.cae.wisc.edu/mailman/listinfo/help-octave
> > 
> 
> 
> --------------------------------------
> Power up the Internet with Yahoo! Toolbar.
> http://pr.mail.yahoo.co.jp/toolbar/
> 


--------------------------------------
Power up the Internet with Yahoo! Toolbar.
http://pr.mail.yahoo.co.jp/toolbar/