[directmusic] Re: Ciaran - you mentioned...

  • From: "Todor Fay" <todor@xxxxxxxxxxxxx>
  • To: <directmusic@xxxxxxxxxxxxx>
  • Date: Thu, 27 Feb 2003 18:20:54 -0800

So, it sounds to me that the autodownload feature I'm thinking of is not
Ciaran's problem. It's still worth checking to make sure the programmer
didn't stick that call in. It's a global parameter on the performance.
However, I don't think it would cause a performance blip if the data
were already downloaded by the script, so it's still unlikely.  But, I'd
recommend verifying anyway. 

Todor

-----Original Message-----
From: directmusic-bounce@xxxxxxxxxxxxx
[mailto:directmusic-bounce@xxxxxxxxxxxxx] On Behalf Of Scott Morgan
Sent: Thursday, February 27, 2003 8:47 AM
To: directmusic@xxxxxxxxxxxxx
Subject: [directmusic] Re: Ciaran - you mentioned...


Oh FYI, the point of the C++ Autodownload is to make life easier on the
programmer.  It means he doesn't have to load and unload every segment
he
calls play on.  It's all handled automatically.
-Scott Morgan
http://Morganstudios.com


----- Original Message -----
From: "Scott Morgan" <scott@xxxxxxxxxxxxxxxxx>
To: <directmusic@xxxxxxxxxxxxx>
Sent: Thursday, February 27, 2003 8:31 AM
Subject: [directmusic] Re: Ciaran - you mentioned...


>
> The Autodownload feature Todor is referring to is a C++ method that
has
> nothing to do with the download checkbox in scripts.  The programmer
would
> be the person who decides to use AutoDownload, not the composer.  Just
so
> you know DMDemo does not use Autodownload.  Typically if you are using
> scripting there is absolutely no need to use AutoDownload since the
script
> handles everything either through manual script commands or by loading
what
> it needs at start (only once).
>
> There are periodic CPU spikes in DMusic playback, but my understanding
is
> that it is the playback engine waking up every so often and writing
new
data
> to the sound buffer.
>
> -Scott Morgan
> http://Morganstudios.com
>
>
> ----- Original Message -----
> From: "Ciaran Walsh" <ciaran@xxxxxxxxxxxxxx>
> To: <directmusic@xxxxxxxxxxxxx>
> Sent: Thursday, February 27, 2003 7:34 AM
> Subject: [directmusic] Re: Ciaran - you mentioned...
>
>
> >
> > that would definitely explain it - does it really download the
instruments
> > again every time you play a band? that seems like a really weird way
for
> it
> > to work...
> >
> > I thought the point of AutoDownload was so that all content
referenced
by
> > the script was downloaded when the script loaded.
> >
> > -----Original Message-----
> > From: directmusic-bounce@xxxxxxxxxxxxx
> > [mailto:directmusic-bounce@xxxxxxxxxxxxx]On Behalf Of Todor Fay
> > Sent: 27 February 2003 15:18
> > To: directmusic@xxxxxxxxxxxxx
> > Subject: [directmusic] Re: Ciaran - you mentioned...
> >
> >
> >
> > Is there any chance you had the AutoDownload command turned on? It's
a
> > feature that would cause each band to automatically cause a download
of
> > its instruments when it played. Playback of a band should normally
not
> > cause any spike because all it does is send program change events to
the
> > synth which just stores them, so nothing active occurs.
> >
> > Todor
> >
> > -----Original Message-----
> > From: directmusic-bounce@xxxxxxxxxxxxx
> > [mailto:directmusic-bounce@xxxxxxxxxxxxx] On Behalf Of Ciaran Walsh
> > Sent: Thursday, February 27, 2003 5:32 AM
> > To: directmusic@xxxxxxxxxxxxx
> > Subject: [directmusic] Re: Ciaran - you mentioned...
> >
> >
> > The problem we were having was never pinned down 100% to one
particular
> > thing, but band events certainly seemed to be contributing
> > significantly.
> > Our performance tests showed big CPU spikes each time a segment was
> > triggered with a new band event, and somehow there was a cumulative
> > effect
> > which had a big impact on the overall frame rate.
> >
> > When we removed band events from the segments the spikes
disappeared.
> >
> > We defined the band globally by having a segment play right at the
start
> > containing just a band event and triggering the first Primary
Segment.
> > As
> > long as you don't recreate the audiopath or change any band settings
you
> > don't need to create the band again after that. I guess it's only a
> > problem
> > if you are re-assigning pchannels as you go along, but I've never
needed
> > to
> > do that.
> >
> > Hope that helps,
> > Ciaran
> >
> > -----Original Message-----
> > From: directmusic-bounce@xxxxxxxxxxxxx
> > [mailto:directmusic-bounce@xxxxxxxxxxxxx]On Behalf Of Paul Stroud
> > Sent: 27 February 2003 10:59
> > To: directmusic@xxxxxxxxxxxxx
> > Subject: [directmusic] Ciaran - you mentioned...
> >
> >
> >
> > ..that band tracks were giving you and your dev team a few problems
-
> > causing some glitch?
> >
> > If you don't mind going over the same subject again; what was the
> > problem?
> > And how did you sort it?
> >
> > I seem to remember that you globally declared a band instead of use
band
> > tracks on each segment.  Is this right?  Did you need to play a Band
> > track
> > at least once?  Is this how you globally declare a band?
> >
> > Anyone else had problems with this sort of thing?
> >
> >
> >
> >
> >
> >
>





Other related posts: