I played with settings for awhile and found something in the manual
(Using->Realtime Audio), which I had missed previously:

"If -iadc is used, ksmps should be set to an integer power of two, and
should be less than or equal to -b."

I tried many combinations of settings, but got best results with:
ksmps=128, -b 128, -B256.  Having ksmps be a power of 2, equal to the
value passed in -b was the ONLY way I could get rid of clicks.

I have always set ksmps to either 10 or 100 (habit I guess) so it
didn't occur to me to play so much with this value.

Could we add this bit about ksmps to the documentation of ksmps
please?  Thanks so much for the help.  Bill: have you tried this?  I
wonder if the fix might not be the same for you.

Maybe an interesting sidenote:  I had to go into the control panel and
manually set the DMA buffer size to match -B.  I assumed -B would
override/force the device driver to the -B setting I provided, but it
didn't.  btw - I am aware of the settings in .csoundrc and removed -b
and -B for testing.


Thanks again - I'm excited to use some of my rt instruments in csound5!
b


On 2/10/06, Victor Lazzarini <victor.lazzarini@nuim.ie> wrote:
> It seems to be with any setting of -b that is an integer
> multiple of ksmps. I originally tested with -b400, coming
> down from -b512 which was giving me clicks. But the
> clicks are there with -b1024 and -b2048, as well as -b4096.
>
> I was thinking about why this is the case, since it does not
> make much sense as you said. Without looking at the code
> (I was on my way home), I speculated whether this is caused
> by the lock mechanism.
>
> The interesting thing (and this is why I thought), is that
> while
> the clicks are there on input, they're not on signals
> generated
> by Csound. On the same performance, an instrument with RT
> input has clicks, while an instrument with say, an
> oscillator,
> does not. So the clicks are being generated at the input,
> possibly at the buffering, and that's why I think the locks
> are involved. Somehow,  the buffering needs to be 'in sync'.
> What do you think?
>
> Victor
>
> >
> > On Friday 10 February 2006 19:07, Victor Lazzarini wrote:
> >
> > > But hey, I found out why!
> > >
> > > It just happens that -b has to be a multiple of ksmps. I
> > > have not checked the code, but it's simple. With -b512
> > > and ksmps=100, clicks are there, but with -b100 (or
> > -b200), the sound is perfect.
> >
> > Are you sure the clicks go away because of the -b/ksmps
> > ratio, and not simply because -b is lower ? What happens
> > if you use ksmps=64 or ksmps=128 with -b 512 ? Also, what
> > is -B set to ? The rtpa plugin does not know about ksmps,
> > and if there is a bug in libsnd.c (where the buffering
> > from -b to ksmps samples is done), it should also affect
> > file input. --
> > 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
>