Unfortunately, CsoundAV is built only with Microsoft C++, while Csound 5 is
built on Windows only with MinGW GCC. Steinberg bungled the calling
conventions for the ASIO callbacks, which work only with Microsoft C++, so
PortAudio needs a thunking layer for GCC that only started working in
PortAudio v19-devel recently. 

What this means is that there does not appear to be any straightforward way
to get Gabriel's audio code ported into Csound 5.

Original Message:
-----------------
From: Art Hunkins abhunkin@uncg.edu
Date: Tue, 24 Aug 2004 00:32:08 -0400
To: csound-dev@eartha.mills.edu
Subject: [CSOUND-DEV:5268] Re: New CsoundVST file release -- again.


I went back and tried the ASIO on your Win dist prior to the current one,
and yes, ASIO does work comparable to AV's ASIO.

I still get occasional "zipper noise" in CS5's ASIO, where I don't in AV's.
No adjustment of ksmps (-b and -B) seems to help. (And I have a bare minimum
of stuff running in the background.) Could the additional Csound "buffer
layer" you mentioned be responsible for this?

I note that ASIO for both Csound5 and CsoundAV are limited to SR of 44100
and above, so I assume this is a limitation of ASIO. Is this correct?

Your reluctance to further pursue other drivers than ASIO is understandable.
I'm looking forward to your perfecting absolutely the best ASIO performance
possible (including for slow systems like mine). Perhaps someone else with
interest and expertise in Portaudio could take up the project of providing
other drivers; there are several reasons why I'd really like to see this
happen:

1) Only multi-channel soundcards seem to come with ASIO drivers. I know that
my basic stereo Avance (CompUSA) card is limited to MME and DirectSound. (I
am planning, however, to try out the generic ASIO driver that Andres put me
on to, and will report on how that may or may not work. URL:
http://www.asio2ks.de/download.html .)

2) A number of my realtime Csound pieces use SR's of 22050 - this to allow
performance with MME (and DirectSound) drivers and improve polyphony, etc.
Moving them to SR=44100 would often preclude playing them on non-ASIO cards
and slower systems.

Gabriel has managed to get really good MME, DirectSound *and* ASIO
performance (your choice!) within the CsoundAV context. Gab, can you offer
us your secrets? And are they adaptable to any platform besides Windows? (I
fail to see a Portaudio file in the AV distribution - but then I understand
nothing of these technical things.)

Art Hunkins

----- Original Message -----
From: "Michael Gogins" 
To: "Csound Developers Discussion List" 
Sent: Sunday, August 22, 2004 8:12 PM
Subject: [CSOUND-DEV:5252] Re: New CsoundVST file release -- again.


> Something happened when I built for not just ASIO, but also DirectX and
> WMME. But I didn't change any sources.
>
> I don't want to be a PortAudio developer, so I'm going back to just ASIO
to
> see if the problems go away again, and if so, that's what I'll distribute.
I
> will move the DLLs to the main directory, you're right there's no reason
to
> get fancy with this.
>
> With just ASIO, realtime was _excellent_, it worked as you say CsoundAV
> does.
>
> I'd guess with the multidriver build, the PortAudio code is opening more
> than one stream, or something.
>
> ----- Original Message -----
> From: "Art Hunkins" 
> To: "Csound Developers Discussion List" 
> Sent: Sunday, August 22, 2004 5:26 PM
> Subject: [CSOUND-DEV:5248] Re: New CsoundVST file release -- again.
>
>
> > Michael,
> >
> > Here's some feedback on the latest CS5 win .exe. (I've had major system
> > problems the last 3-4 days, and am just now back operational.)
> >
> > 1) Ctrl-C and Ctrl-Break are now working properly for me - as long as I
> > remember to select the correct window first. Somehow, on my system CS5
> > closes the (FLTK) performance window without selecting the "run" window.
> >
> > 2) Any reason, in the distribution, not to have the windows dll's in the
> > main folder? Otherwise you have to move them (as the .pdf says); they
work
> > fine in the "main."
> >
> > 3) With regard to ASIO: unfortunately, I can't test this with my Gina
> card.
> > It seems the program doesn't accept SR's less than 44100; did you mean
> this?
> > At 44100, I can't get anything - even very cut up.
> >
> > 4) With regard to MME: for my sound card to put out anything even
> generally
> > chopped up, my SR (on both my Gina card and a garden-variety CompUSA
> stereo
> > card) must lower to 8000 - and I'm talking mono here. (Please see the
> > dirt-simple test file I've been using to test - a switched on/off
> 10-second
> > performance sine wave.)
> >
> > Here a very odd thing happens: with SR=8000, if ksmps=1000 (along
with -B
> > and -b), the actual performance is 8 seconds long (instead of 10). When
> > ksamps=2000, the performance lasts 4 seconds; when 500, performance 17
> > seconds (and interrupted/sporadic); when 100, performance 83 seconds
(and
> > vastly distorted).
> >
> > Funny thing: elapsed time, as indicated in the performance messages, is
> all
> > proportional to the time warp; i.e., the messages always indicate a
> > performance duration of 10 seconds.
> >
> > Again, these results apply both to my Gina card and the ordinary stereo
> card
> > I have.
> >
> > I've also run this file in CsoundAV. Both with ASIO and the DirectX
> drivers
> > it has, this can run at normal speeds (44100) without problems (it has a
> bit
> > more problem with MME). AV also doesn't have the SR limitation to 44100
of
> > CS5.
> >
> > AV has got to be handling audio differently from CS5. If this has
> something
> > to do with the Csound extra buffer layer you've been talking about, for
> the
> > sake of all realtimers, let's get that out of here!
> >
> > Might be good to talk with Gab to come to some idea of what the realtime
> > performance difference (radical) is due to. For me at least, CS5 isn't
> close
> > to being practical for realtime - even with ASIO.
> >
> > Have others put the latest Win .exe through any realtime test?
> >
> > Art Hunkins
> >
> > ----- Original Message -----
> > From: "Michael Gogins" 
> > To: "Csound Developers Discussion List" 
> > Sent: Monday, August 16, 2004 10:59 PM
> > Subject: [CSOUND-DEV:5224] New CsoundVST file release -- again.
> >
> >
> > > This file release is in the csoundvst package and is named
> > > csound5-mingw-2004-08-16.zip. It is for Csound 5 and CsoundVST with
> > prebuilt
> > > binaries for Windows built with MinGW, and sources for all platforms,
> and
> > > contains the following changes:
> > >
> > > The portaudio.dll file is now named portaudio.dll instead of
> > > portaudio.dll.0.0.19. It is built with Dev-C++/MinGW instead of
> autotools
> > > and contains ASIO drivers, DirectSound (DirectX) drivers, and Windows
> > > Multimedia Extensions (WMME) drivers.
> > >
> > > Top/main.c and InOut/pa_blocking.c contain minor changes in an effort
to
> > > improve the ability to cancel a performance with real-time audio using
> > > Control-C (console Csound only). It works for me, but your mileage may
> > vary.
> > >
> >
>


--------------------------------------------------------------------
mail2web - Check your email from the web at
http://mail2web.com/ .