Here are test results. All are under as comparable conditions as humanly
possible. All timings done manually as Csound5 apparently no longer reports
render time (??): (all tests repeated and averages taken)

csound32.exe 3600" file - render time = 130" average
csound.exe 3600" file - render time = 143.5" average
  Conclusion: 64-bit is 10%+ slower
(Note: the only MIDI activity here was an "always on" instrument reading 8
ctrl7's which did not move; thus no "schedkwhen" MIDI events were
triggered.)

csound32.exe as above adding 60" of MIDI ctrl activity -
  render time = 180" average
csound.exe as above adding 60" of MIDI ctrl activity -
  render time = 211" average
  Conclusion: 64-bit is 17%+ slower

In the MIDI ctrl activity test, the same sequence of MIDI activity was
repeated for nearly the same length of time in each trial. Six intense and
sustained MIDI events were triggered. I expected csound.exe to do less well
on the second test because this was one where I knew that the multiple
(concurrent) MIDI events caused overload and stuttering, and my
system/csound.exe could not keep up.

My test system is an Athlon XP 1600+ (32-bit) running WinME, with 224MB RAM.

Art Hunkins

----- Original Message -----
From: "Michael Gogins" 
To: 
Sent: Wednesday, March 01, 2006 9:50 AM
Subject: Re: [Csnd] Realtime Csound5


> Could you please re-run your test, rendering both times to soundfile, and
measure the times taken? This will give a more precise idea of the speedup
of csound32.
>
> Regards,
> Mike
>
> -----Original Message-----
> >From: Victor Lazzarini 
> >Sent: Mar 1, 2006 5:38 AM
> >To: csound@lists.bath.ac.uk
> >Subject: Re: [Csnd] Realtime Csound5
> >
> >Thanks for the report (I always use the floats version anyway, never
bothered
> >with the doubles... even for 'tape' pieces. But it's good to know).
> >
> >Victor
> >
> >At 03:54 01/03/2006, you wrote:
> >>I've been comparing csound.exe and csound32.exe with respect to realtime
> >>performance. (Csound.exe is the 64-bit, "higher precision" version.)
> >>
> >>I can report that csound32.exe has a small but significant advantage as
to
> >>what it can handle without breakup compared with csound.exe. I'm
currently
> >>working on a .csd which runs fine at 44100SR with csound32.exe, but
breaks
> >>up with csound.exe. (I imagine the difference is comparable to the
> >>difference in render times for the two versions.)
> >>
> >>That being the case, I'd recommend csound32.exe for all realtime use (on
> >>Windows anyway). This means you need to install from the .zip
distribution
> >>(Varga) rather than the Gogins installer .exe, as the Gogins distro only
> >>contains csound.exe - not csound32.exe.
> >>
> >>Art Hunkins
> >>
> >>--
> >>Send bugs reports to this list.
> >>To unsubscribe, send email to csound-unsubscribe@lists.bath.ac.uk
> >
> >Victor Lazzarini
> >Music Technology Laboratory
> >Music Department
> >National University of Ireland, Maynooth
> >
> >--
> >Send bugs reports to this list.
> >To unsubscribe, send email to csound-unsubscribe@lists.bath.ac.uk
>
>
>
> --
> Send bugs reports to this list.
> To unsubscribe, send email to csound-unsubscribe@lists.bath.ac.uk