HDSP 9632 driver and expansion cards

HDSP 9632 driver and expansion cards

[Florin Andrei]

Anyone using the latest RME HDSP 9632 driver in ALSA, while also using
the expansion cards? I mean, the newer AI4S-192 and/or AO4S-192 which
are required to get the 192kHz sampling on all analogs.

Do the latest expansion cards (192kHz) work with the latest ALSA driver?

In general, are the ALSA drivers ready to support any kind of expansion
cards on the HDSP 9632 / HDSP 9652?

-- 
Florin Andrei

http://florin.myip.org/



-------------------------------------------------------
The SF.Net email is sponsored by EclipseCon 2004
Premiere Conference on Open Tools Development and Integration
See the breadth of Eclipse activity. February 3-5 in Anaheim, CA.
http://www.eclipsecon.org/osdn

Re: HDSP 9632 driver and expansion cards

[Thomas Charbonnel]

Florin Andrei wrote :
> Anyone using the latest RME HDSP 9632 driver in ALSA, while also using
> the expansion cards? I mean, the newer AI4S-192 and/or AO4S-192 which
> are required to get the 192kHz sampling on all analogs.
> 
> Do the latest expansion cards (192kHz) work with the latest ALSA driver?
> 
> In general, are the ALSA drivers ready to support any kind of expansion
> cards on the HDSP 9632 / HDSP 9652?
> 

Hi Florin,

Support is there, but as of today and as far as I know this is untested.

Thomas




-------------------------------------------------------
The SF.Net email is sponsored by EclipseCon 2004
Premiere Conference on Open Tools Development and Integration
See the breadth of Eclipse activity. February 3-5 in Anaheim, CA.
http://www.eclipsecon.org/osdn

Re: HDSP 9632 driver and expansion cards

[Paul Davis]

>Anyone using the latest RME HDSP 9632 driver in ALSA, while also using
>the expansion cards? I mean, the newer AI4S-192 and/or AO4S-192 which
>are required to get the 192kHz sampling on all analogs.
>
>Do the latest expansion cards (192kHz) work with the latest ALSA driver?

i doubt it. RME have not given us any information about this new
hardware. i suggest that you write to them and ask that they give me
and/or thomas charbonnel the required information.

>In general, are the ALSA drivers ready to support any kind of expansion
>cards on the HDSP 9632 / HDSP 9652?

they are "ready to support" in the sense that it probably isn't that
much work, but they do not currently do so.

--p


-------------------------------------------------------
The SF.Net email is sponsored by EclipseCon 2004
Premiere Conference on Open Tools Development and Integration
See the breadth of Eclipse activity. February 3-5 in Anaheim, CA.
http://www.eclipsecon.org/osdn

Re: HDSP 9632 driver and expansion cards

[Thomas Charbonnel]

Paul Davis a écrit :
>>Anyone using the latest RME HDSP 9632 driver in ALSA, while also using
>>the expansion cards? I mean, the newer AI4S-192 and/or AO4S-192 which
>>are required to get the 192kHz sampling on all analogs.
>>
>>Do the latest expansion cards (192kHz) work with the latest ALSA driver?
> 
> 
> i doubt it. RME have not given us any information about this new
> hardware. i suggest that you write to them and ask that they give me
> and/or thomas charbonnel the required information.
> 
> 
>>In general, are the ALSA drivers ready to support any kind of expansion
>>cards on the HDSP 9632 / HDSP 9652?
> 
> 
> they are "ready to support" in the sense that it probably isn't that
> much work, but they do not currently do so.
> 
> --p
> 


To clarify the situation : I already asked for and got this information 
while working on the H9632 driver. The code to support those expansion 
boards is already in the driver, and also in hdspmixer and hdspconf, the 
only thing is that this is untested. I also added traditional (old) 
expansion boards support for H96xx cards (you have to activate the board 
using hdspconf or amixer).

Thomas



-------------------------------------------------------
The SF.Net email is sponsored by EclipseCon 2004
Premiere Conference on Open Tools Development and Integration
See the breadth of Eclipse activity. February 3-5 in Anaheim, CA.
http://www.eclipsecon.org/osdn

Re: HDSP 9632 driver and expansion cards

[Ed Wildgoose]

> Florin Andrei wrote :
> > Anyone using the latest RME HDSP 9632 driver in ALSA, while also using
> > the expansion cards? I mean, the newer AI4S-192 and/or AO4S-192 which
> > are required to get the 192kHz sampling on all analogs.
> >
> > Do the latest expansion cards (192kHz) work with the latest ALSA driver?
> >
> > In general, are the ALSA drivers ready to support any kind of expansion
> > cards on the HDSP 9632 / HDSP 9652?
> >
>
> Hi Florin,
>
> Support is there, but as of today and as far as I know this is untested.
>
> Thomas

Hi Thomas,

Thanks so much for your RME 9632 driver and your fantastic mixer utilities.
I have a 9632 now and also the 4 port analogue expansion card.

I haven't had time to try and get any sound out of it yet!  However, it is
detected correctly in the mixer, and the driver thinks that there are 4
extra channels.  I need to get some jack socket convertors to get some sound
out, but will let you know.

I have a few usability comments with regards to the drivers that I would
like to share, will draft a fuller email later, but basically I am using it
as a hifi input for music, nothing else, but I'm frequently getting very
jittery & broken up audio (guessing that the default alsa buffer/latency
size is really tiny?) and not many input formats are supported which means
that the plug driver is usually required (not a huge problem though).  It
would also be extremely nice to have some simple mixer controls available
for compatibility with older/simpler apps.  In particular a fake master
mixer which reduces all channels would be extremely desirable (I'm hoping to
use the card for a surround sound output)

Thanks for your work on this, it is very much appreciated.

Ed W



-------------------------------------------------------
The SF.Net email is sponsored by EclipseCon 2004
Premiere Conference on Open Tools Development and Integration
See the breadth of Eclipse activity. February 3-5 in Anaheim, CA.
http://www.eclipsecon.org/osdn

Re: HDSP 9632 driver and expansion cards

[Florin Andrei]

On Wed, 2004-01-21 at 10:41, Ed Wildgoose wrote:
> > Florin Andrei wrote :
> > > Anyone using the latest RME HDSP 9632 driver in ALSA, while also using
> > > the expansion cards? I mean, the newer AI4S-192 and/or AO4S-192 which
> > > are required to get the 192kHz sampling on all analogs.

> I have a 9632 now and also the 4 port analogue expansion card.

The new, 192kHz ones?

> I haven't had time to try and get any sound out of it yet!  However, it is
> detected correctly in the mixer, and the driver thinks that there are 4
> extra channels.  I need to get some jack socket convertors to get some sound
> out, but will let you know.

Please send feedback to the list when you're done.

I was contemplating the purchase of a HDSP9632 plus at least the input
card, 192kHz version. If the driver is not working until i get it (which
might be weeks or months away), i'm willing to do testing on behalf of
Thomas or whoever is doing the development (or even provide SSH access
to the system).

> Thanks for your work on this, it is very much appreciated.

Indeed, the RME driver is a great feature of ALSA.

-- 
Florin Andrei

http://florin.myip.org/



-------------------------------------------------------
The SF.Net email is sponsored by EclipseCon 2004
Premiere Conference on Open Tools Development and Integration
See the breadth of Eclipse activity. February 3-5 in Anaheim, CA.
http://www.eclipsecon.org/osdn

Re: HDSP 9632 driver and expansion cards

[Paul Davis]

>I have a few usability comments with regards to the drivers that I would
>like to share, will draft a fuller email later, but basically I am using it
>as a hifi input for music, nothing else, but I'm frequently getting very
>jittery & broken up audio (guessing that the default alsa buffer/latency
>size is really tiny?) and not many input formats are supported which means
>that the plug driver is usually required (not a huge problem though).  It

its the philosophy of ALSA that the "hw:N" drivers export the h/w
functionality, which in this case indeed means "not many input formats
are supported" (i.e. the h/w doesn't do this). this is precisely what
the "plughw:N" devices are for.



-------------------------------------------------------
The SF.Net email is sponsored by EclipseCon 2004
Premiere Conference on Open Tools Development and Integration
See the breadth of Eclipse activity. February 3-5 in Anaheim, CA.
http://www.eclipsecon.org/osdn

Re: HDSP 9632 driver and expansion cards

[Thomas Charbonnel]

Ed Wildgoose wrote :
>>Florin Andrei wrote :
>>
>>>Anyone using the latest RME HDSP 9632 driver in ALSA, while also using
>>>the expansion cards? I mean, the newer AI4S-192 and/or AO4S-192 which
>>>are required to get the 192kHz sampling on all analogs.
>>>
>>>Do the latest expansion cards (192kHz) work with the latest ALSA driver?
>>>
>>>In general, are the ALSA drivers ready to support any kind of expansion
>>>cards on the HDSP 9632 / HDSP 9652?
>>>
>>
>>Hi Florin,
>>
>>Support is there, but as of today and as far as I know this is untested.
>>
>>Thomas
> 
> 
> Hi Thomas,
> 
> Thanks so much for your RME 9632 driver and your fantastic mixer utilities.
> I have a 9632 now and also the 4 port analogue expansion card.
> 
> I haven't had time to try and get any sound out of it yet!  However, it is
> detected correctly in the mixer, and the driver thinks that there are 4
> extra channels.  I need to get some jack socket convertors to get some sound
> out, but will let you know.
> 

Please do so, thanks.

> I have a few usability comments with regards to the drivers that I would
> like to share, will draft a fuller email later, but basically I am using it
> as a hifi input for music, nothing else, but I'm frequently getting very
> jittery & broken up audio (guessing that the default alsa buffer/latency
> size is really tiny?) and not many input formats are supported which means
> that the plug driver is usually required (not a huge problem though).  It
> would also be extremely nice to have some simple mixer controls available
> for compatibility with older/simpler apps.  In particular a fake master
> mixer which reduces all channels would be extremely desirable (I'm hoping to
> use the card for a surround sound output)
> 

For the "jittery & broken up audio" problem, are you running an 
optimized audio kernel (preempt + lowlatency patched) ?
How frequent is frequently ? Can you find any relation between the audio 
problems and external factors (disk access, heavy load, etc...) ?

For the supported formats, this is a limitation of the hardware. Plughw 
is indeed your friend here, or try to find a player that handles this 
natively (alsaplayer ?).

For the mixer controls : there were some in the past, but I removed them 
because I couldn't find any clean solution to deal with them while 
changing speed mode (because their number changes at this occasion). 
Maybe Jaroslav or Takashi have an idea on the subject ? When I last 
worked on this, the only solution I could find would have been to mess 
up with a semaphore held by the alsa-lib layer and although it would 
have worked, it would have been plain ugly.
Anyway I plan a massive redesign of the way the driver and userspace 
apps interact regarding the matrix mixer, mostly to enable several 
application to access the mixer at the same time in a cooperative way 
(how about an alsaseq client or a ladspa plugin accessing the matrix 
with the changes reflected in hdspmixer ? :). This is the prerequisite 
of some work I want to do with jack and ardour to handle hardware 
monitoring in a better way. I could indeed add a master volume control 
at this occasion.

> Thanks for your work on this, it is very much appreciated.
> 
Thanks for your kind words, they're very much appreciated these days...

Thomas




-------------------------------------------------------
The SF.Net email is sponsored by EclipseCon 2004
Premiere Conference on Open Tools Development and Integration
See the breadth of Eclipse activity. February 3-5 in Anaheim, CA.
http://www.eclipsecon.org/osdn

Re: HDSP 9632 driver and expansion cards

[Ed Wildgoose]

> > > Florin Andrei wrote :
> > > > Anyone using the latest RME HDSP 9632 driver in ALSA, while also
using
> > > > the expansion cards? I mean, the newer AI4S-192 and/or AO4S-192
which
> > > > are required to get the 192kHz sampling on all analogs.
>
> > I have a 9632 now and also the 4 port analogue expansion card.
>
> The new, 192kHz ones?

Hi, tested this very briefly last night (I only have enough plugs to wire up
a single channel...), and got sound out of it!!  Wahey!

Yes, it is the new 192Khz one, and the card has balanced, large 1/4" phono
sockets out of the back (which is useful, yet inconvenient).

For what it is worth, I have an RME 96/8PAD card, and the sound quality
(analogue out) is basically identical.  Test done using a fairly decent hifi
system with speakers.  Perhaps with headphones you could discern a
difference?

I would suggest that you could save money and stick with the 96khz 96/8PAD
card unless you need some of the 9632 features

Both however, have excellent sound quality and are easily on a par with my
Meridian 508.24 CD player.  I can hear practically no difference on the
analogue output of each.  I would very much recommend the card to anyone
looking to implement a high end hifi system and not wanting to have to feed
CD's into the machine (rip 'em all to the HD and play them all from the
couch with a remote control...)

However, as a negative to the 9632 card, I find that I can't use it with
many apps such as Mythtv, mplayer, etc.  The sound just stutters.  I suspect
that the driver defaults to a short period size and some of these apps don't
request anything other than the default?  Is there any way to use the
asound.conf file to force a specific period size?  Since this is just doing
home music duties, I don't really need or care for the ultra-low latency
capabilities of the 9632...

It would also be extremely cool if there was a simple master mixer control
in the alsa mixer.  This would reduce all channels simultaneously and allow
easy use as a 5.1 surround component.

Thanks very much to Thomas for writing this driver!

Ed W



-------------------------------------------------------
The SF.Net email is sponsored by EclipseCon 2004
Premiere Conference on Open Tools Development and Integration
See the breadth of Eclipse activity. February 3-5 in Anaheim, CA.
http://www.eclipsecon.org/osdn

Re: HDSP 9632 driver and expansion cards

[Ed Wildgoose]

> > I haven't had time to try and get any sound out of it yet!  However, it
is
> > detected correctly in the mixer, and the driver thinks that there are 4
> > extra channels.  I need to get some jack socket convertors to get some
sound
> > out, but will let you know.
> >
>
> Please do so, thanks.

Hi Thomas,  I tested this, and it does seem to work ok!  I literally just
squirted some sound through a single channel briefly, so this is nothing
more than a "it works" test.  I will get some plugs over the next few days
for better trials.  Thanks so much for your work on this!

> For the "jittery & broken up audio" problem, are you running an
> optimized audio kernel (preempt + lowlatency patched) ?
> How frequent is frequently ? Can you find any relation between the audio
> problems and external factors (disk access, heavy load, etc...) ?

I'm using 2.6.1 mm-1 with the pre-empt option switched on.  With low system
load I seem to be able to get JACK to work for short periods with 1.5ms
latency!  3ms and up latency works with JACK, reasonably despite some
processing with Brutefir going on as well...

I only tried it a couple of times, and I forget now whether I used both OSS
emulation and ALSA devices, or just one?  However, the effect was something
like when you are feeding an audio file to the card at a fixed rate, and
it's playing it at a faster rate, you get continuous snatches of sound and
underflows leaving a short gap before sound restarts (multiple times per
second).

I will try it again and document exactly which apps and what my setup is.
Sorry for the vague problem report...  In the meantime I'm actually using
the PAD card for general duties (ie watching TV)


> For the mixer controls : there were some in the past, but I removed them
> because I couldn't find any clean solution to deal with them while
> changing speed mode (because their number changes at this occasion).
> Maybe Jaroslav or Takashi have an idea on the subject ? When I last
> worked on this, the only solution I could find would have been to mess
> up with a semaphore held by the alsa-lib layer and although it would
> have worked, it would have been plain ugly.
> Anyway I plan a massive redesign of the way the driver and userspace
> apps interact regarding the matrix mixer, mostly to enable several
> application to access the mixer at the same time in a cooperative way
> (how about an alsaseq client or a ladspa plugin accessing the matrix
> with the changes reflected in hdspmixer ? :). This is the prerequisite
> of some work I want to do with jack and ardour to handle hardware
> monitoring in a better way. I could indeed add a master volume control
> at this occasion.

I hear what you are saying.  Would it be possible to have a very badly
behaved master mixer control which possibly has a rather confused view of
min and max levels even?

The situation I would rather like to get to is that I could use the output
mixers on each channel to set a maximum amplitude for that channel.  The
physical analogue outputs are then likely to feed directly into power amps
with no preamp (perhaps an attenuator in the way though).  The master
control (lets call it a fader) will operate on all channels individually,
and adjust them equally (in dB, not linearly), so that I effectively have a
home theatre 5.1 output directly into the power amps.  The master fader
control is effectively applying a 0-100% scaling on the output mixer,
bearing in mind that each output channel may well have different absolute
levels (depends how close each speaker is).

Does this sound easier?  Is it feasible?

It would be very exciting if this could be made to work.  I suspect this
feature alone would sell RME some sound cards... (not that you care, just an
aside!)

For what it is worth, the card will likely be fed by JACK via a Brutefir
filter to do room correction based on a filter designed using Brutefir. (A
very interesting project that is well worth experimenting with...)

Thanks so much for taking the time to listen - much appreciated

Ed W



-------------------------------------------------------
The SF.Net email is sponsored by EclipseCon 2004
Premiere Conference on Open Tools Development and Integration
See the breadth of Eclipse activity. February 3-5 in Anaheim, CA.
http://www.eclipsecon.org/osdn

Re: HDSP 9632 driver and expansion cards

[Edward Wildgoose]

> I will try it again and document exactly which apps and what my setup is.
> Sorry for the vague problem report...  In the meantime I'm actually using
> the PAD card for general duties (ie watching TV)

Hi Thomas,

I successfully watched a surround 5.1 film tonight using the 9632!! Wahey!  Thanks a bunch!

I had to use Xine and oss emulation (kernel 2.6.1 mm-1 and alsa patched to around 1.01 I think?).  Alsa and xine gave errors about
device surround51 not being defined, but I think this is likely my setup of Xine..?

Mplayer completely refuses to play ball.  In asla mode (using plug), it scrolls a load of:
alsa-space: xrun of at least: 11.501 msecs, resetting stream

type messages.   (always 10-12ms, with occasionaly 18ms thrown in)  Sound just stutters constantly.

It is no better in OSS mode, same bad audio, but no messages on the screen.

I'm fairly sure that mplayer is at least trying to set the period size (since it fails to work under alsa with my RME 96/8 PAD card,
moaning about unable to set period size).

Any ideas what might be happening?  My guess would have been that the period size was too small and the player was not servicing the
audio frequently enough...  However, I'm not sure...

Mythtv has a similar issue, and this is the main app I want to get working.  Any ideas would be appreciated!

Thanks

Ed W



-------------------------------------------------------
The SF.Net email is sponsored by EclipseCon 2004
Premiere Conference on Open Tools Development and Integration
See the breadth of Eclipse activity. February 3-5 in Anaheim, CA.
http://www.eclipsecon.org/osdn

Re: HDSP 9632 driver and expansion cards

[Florin Andrei]

On Thu, 2004-01-22 at 07:06, Ed Wildgoose wrote:
> > > > Florin Andrei wrote :
> > > > > Anyone using the latest RME HDSP 9632 driver in ALSA, while also
> using
> > > > > the expansion cards? I mean, the newer AI4S-192 and/or AO4S-192
> which
> > > > > are required to get the 192kHz sampling on all analogs.
> >
> > > I have a 9632 now and also the 4 port analogue expansion card.
> >
> > The new, 192kHz ones?
> 
> Hi, tested this very briefly last night (I only have enough plugs to wire up
> a single channel...), and got sound out of it!!  Wahey!
> Yes, it is the new 192Khz one, and the card has balanced, large 1/4" phono
> sockets out of the back (which is useful, yet inconvenient).

Balanced? Are you sure?
Wow! That's not specified on their website.
Very, very good news.

I was kinda feeling inclined to save a bucketful of bucks and get the
M-Audio 1010LT instead, which is supposed to be pretty good as well, but
now that you said the Hammerfall has balanced connectors... i don't care
about money anymore. :-)

> I would suggest that you could save money and stick with the 96khz 96/8PAD
> card unless you need some of the 9632 features

(see below)

> However, as a negative to the 9632 card, I find that I can't use it with
> many apps such as Mythtv, mplayer, etc.  The sound just stutters.  I suspect
> that the driver defaults to a short period size and some of these apps don't
> request anything other than the default?

Whoops, not good.
Which application does work for you?

The main reason why i'm planning to get the Hammerfall is to use it with
JACK and music apps. But of course i'd like to use it with XMMS, Xine,
etc as well.
So that's why what you noticed is kinda scary to me.

As for the 192-vs-96kHz thing...
I want to use the RME in a small music studio (hence the JACK thing).
Digital processing, if done entirely at 44.1 (for CD) or 48kHz (for DAT)
has an ugly "lifeless" flavour that's disliked by most musicians. One
way to work around it is to simply bump up the frequency of the entire
digital chain, to 88.2 or 96kHz. But i've heard recently about some
famous british composer who simply went all the way up to 176.4/192kHz,
with results that amazed even the dyed-in-the-wool analogue fans.
To get back to 44.1 before burning the master, you have to do an
interpolating downsample (do NOT simply discard 3 samples out of 4!); it
may seem counterintuitive, but apparently it sounds a lot better this
way than if you process all along at 44.1

So... balanced connectors + 192kHz = i want the RME

-- 
Florin Andrei

http://florin.myip.org/



-------------------------------------------------------
The SF.Net email is sponsored by EclipseCon 2004
Premiere Conference on Open Tools Development and Integration
See the breadth of Eclipse activity. February 3-5 in Anaheim, CA.
http://www.eclipsecon.org/osdn

Re: HDSP 9632 driver and expansion cards

[Thomas Charbonnel]

Ed Wildgoose wrote :
>>>I haven't had time to try and get any sound out of it yet!  However, it
> 
> is
> 
>>>detected correctly in the mixer, and the driver thinks that there are 4
>>>extra channels.  I need to get some jack socket convertors to get some
> 
> sound
> 
>>>out, but will let you know.
>>>
>>
>>Please do so, thanks.
> 
> 
> Hi Thomas,  I tested this, and it does seem to work ok!  I literally just
> squirted some sound through a single channel briefly, so this is nothing
> more than a "it works" test.  I will get some plugs over the next few days
> for better trials.  Thanks so much for your work on this!
> 
> 

Thanks, this is good news. Could you also please test hdspmixer with the 
expansion board and confirm there's no problem here either ?

>>For the "jittery & broken up audio" problem, are you running an
>>optimized audio kernel (preempt + lowlatency patched) ?
>>How frequent is frequently ? Can you find any relation between the audio
>>problems and external factors (disk access, heavy load, etc...) ?
> 
> 
> I'm using 2.6.1 mm-1 with the pre-empt option switched on.  With low system
> load I seem to be able to get JACK to work for short periods with 1.5ms
> latency!  3ms and up latency works with JACK, reasonably despite some
> processing with Brutefir going on as well...
> 
> I only tried it a couple of times, and I forget now whether I used both OSS
> emulation and ALSA devices, or just one?  However, the effect was something
> like when you are feeding an audio file to the card at a fixed rate, and
> it's playing it at a faster rate, you get continuous snatches of sound and
> underflows leaving a short gap before sound restarts (multiple times per
> second).
> 

Just to make it clear that this is a software problem and not the 
problem that Tim, Paul and Mark reported, could you please try to route 
an incoming signal (through analog or digital in) to your amp using 
hdspmixer and tell me if the routed sound suffers from the same audio 
corruption as the one being played back ? (this test should be done 
while you're playing back a stuttering software stream)

> I will try it again and document exactly which apps and what my setup is.
> Sorry for the vague problem report...  In the meantime I'm actually using
> the PAD card for general duties (ie watching TV)
> 
> 
> 
>>For the mixer controls : there were some in the past, but I removed them
>>because I couldn't find any clean solution to deal with them while
>>changing speed mode (because their number changes at this occasion).
>>Maybe Jaroslav or Takashi have an idea on the subject ? When I last
>>worked on this, the only solution I could find would have been to mess
>>up with a semaphore held by the alsa-lib layer and although it would
>>have worked, it would have been plain ugly.
>>Anyway I plan a massive redesign of the way the driver and userspace
>>apps interact regarding the matrix mixer, mostly to enable several
>>application to access the mixer at the same time in a cooperative way
>>(how about an alsaseq client or a ladspa plugin accessing the matrix
>>with the changes reflected in hdspmixer ? :). This is the prerequisite
>>of some work I want to do with jack and ardour to handle hardware
>>monitoring in a better way. I could indeed add a master volume control
>>at this occasion.
> 
> 
> I hear what you are saying.  Would it be possible to have a very badly
> behaved master mixer control which possibly has a rather confused view of
> min and max levels even?
> 
> The situation I would rather like to get to is that I could use the output
> mixers on each channel to set a maximum amplitude for that channel.  The
> physical analogue outputs are then likely to feed directly into power amps
> with no preamp (perhaps an attenuator in the way though).  The master
> control (lets call it a fader) will operate on all channels individually,
> and adjust them equally (in dB, not linearly), so that I effectively have a
> home theatre 5.1 output directly into the power amps.  The master fader
> control is effectively applying a 0-100% scaling on the output mixer,
> bearing in mind that each output channel may well have different absolute
> levels (depends how close each speaker is).
> 
> Does this sound easier?  Is it feasible?
> 
> 

This is more or less my plan. What you're asking for actually exists but 
is a totalmix/hdspmixer only software abstraction (the third row, i.e. 
"Outputs", one of the few features of the HDSP that is not directly 
reflecting the hardware). Any change on this third row is interpreted by 
hdspmixer and translated into standard matrix mixer calls. This prevents 
other applications to play nicely with hdspmixer as any change done to 
the matrix mixer's state by those applications will shortcut hdspmixer's 
representation and lead to an unsynchronised state. The idea is thus to 
implement this output layer in the driver and setup specific controls 
for it. My main concern at this point is performance, particularly with 
the new MADI driver Winfried Ritsch has been working on (64 channels...).

It would be very exciting if this could be made to work.  I suspect this
> feature alone would sell RME some sound cards... (not that you care, just an
> aside!)
> 

I think the nicest feature you'd get with this design is the ability to 
manipulate totalmix/hdspmixer's enhanced mixer (with this output stage) 
from several applications in a coherent way (hdspmixer, ladpsa plugin, 
script, midi automation, etc...), a feature that people constantly ask 
for on the RME forum. This could also very well "sell" a few linux 
installs to RME customers :)

> For what it is worth, the card will likely be fed by JACK via a Brutefir
> filter to do room correction based on a filter designed using Brutefir. (A
> very interesting project that is well worth experimenting with...)
> 
> Thanks so much for taking the time to listen - much appreciated
> 
> Ed W
> 

Thanks for testing and reporting !

Thomas





-------------------------------------------------------
The SF.Net email is sponsored by EclipseCon 2004
Premiere Conference on Open Tools Development and Integration
See the breadth of Eclipse activity. February 3-5 in Anaheim, CA.
http://www.eclipsecon.org/osdn

Re: HDSP 9632 driver and expansion cards

[Ed Wildgoose]

> Thanks, this is good news. Could you also please test hdspmixer with the
> expansion board and confirm there's no problem here either ?

Yes, hdspmixer does all the right things with the extra channels.

There appears to be a minor bug in that the input channels 8 and 9 (I
think?), ie the top right two inputs (right of the spdif input), do not draw
properly initially.  The numbers, and bits of the outline appear, but the
sliders and bars don't draw.  At some later stage it all appears properly.

Since this is done using a remote xsession to a windows PC and cygwin I
can't be sure that this isn't just an X problem on win32...  It's a minor
issue anyway

Actually more of an issue is that there is a ton of X traffic when stuff is
playing which overwhelms my 802.11a network.  PC gets maxed out with a high
cpu load just trying to squirt all the traffic over SSH.  I wonder if we
could have variable refresh rates on the bouncing level meters?  Or perhaps
you could investigate whether less of the screen could be updated perhaps
(optimised drawing of boxes might reduce X traffic?)

Anyway, just a nice to have.  Thanks for what we have already though!


> Just to make it clear that this is a software problem and not the
> problem that Tim, Paul and Mark reported, could you please try to route
> an incoming signal (through analog or digital in) to your amp using
> hdspmixer and tell me if the routed sound suffers from the same audio
> corruption as the one being played back ? (this test should be done
> while you're playing back a stuttering software stream)

Sure, will do.  For what it's worth the same film works under xine, but not
mplayer.

Alsaplayer also works fine I think (on different files), as does aplay.  I
would be pretty sure that for a given FLAC file, alsaplayer would play it
fine and mplayer would stutter.

I will try and grab the mplayer source and have a peer at the audio output
code.


> I think the nicest feature you'd get with this design is the ability to
> manipulate totalmix/hdspmixer's enhanced mixer (with this output stage)
> from several applications in a coherent way (hdspmixer, ladpsa plugin,
> script, midi automation, etc...), a feature that people constantly ask
> for on the RME forum. This could also very well "sell" a few linux
> installs to RME customers :)

Please consider dropping me a line if you do any work on this.  I would love
to test it as you go along.

Thanks again



-------------------------------------------------------
The SF.Net email is sponsored by EclipseCon 2004
Premiere Conference on Open Tools Development and Integration
See the breadth of Eclipse activity. February 3-5 in Anaheim, CA.
http://www.eclipsecon.org/osdn

Re: HDSP 9632 driver and expansion cards

[Edward Wildgoose]

> > Just to make it clear that this is a software problem and not the
> > problem that Tim, Paul and Mark reported, could you please try to route
> > an incoming signal (through analog or digital in) to your amp using
> > hdspmixer and tell me if the routed sound suffers from the same audio
> > corruption as the one being played back ? (this test should be done
> > while you're playing back a stuttering software stream)
>

OK, I'm sure that I misunderstand what you are asking, but here is what I did:

Set mplayer playing with something like "mplayer file.mpg -ao alsa9:rme9632plug" - It sutters like crazy and mplayer reports "xruns
of at least 10ms".

At the same time I spun a disk in the CD player, and changed the hdspmixer so that spdif mapped through to the front speakers
instead of the mplayer audio.  CD output was fine, mplayer continued spewing errors up the screen.

I have just had a peek at the mplayer alsa output code, and compared with the verbose logs.  It does seem that mplayer has managed
to set "chunksize" to 1024 and asked for 2 fragments - so this should be quite reasonable.  Switching to mmap or noblock mode makes
no difference (neither does switching to the OSS device...)

Any ideas what could be happening?  As I said, the same problem happens with Mythtv under OSS, and I think also under alsa.  Some
other apps are fine, alsaplayer for example.

Quite peculiar...

I just checked and I actually have alsa 1.00 rc2.  Am I missing any driver updates?

asound.conf is pretty empty in case it matters.  Just an entry for the hardware device and a plug device for the same.

Thanks

Ed W



-------------------------------------------------------
The SF.Net email is sponsored by EclipseCon 2004
Premiere Conference on Open Tools Development and Integration
See the breadth of Eclipse activity. February 3-5 in Anaheim, CA.
http://www.eclipsecon.org/osdn

Re: HDSP 9632 driver and expansion cards

[Thomas Charbonnel]

Ed Wildgoose wrote :
>>Thanks, this is good news. Could you also please test hdspmixer with the
>>expansion board and confirm there's no problem here either ?
> 
> 
> Yes, hdspmixer does all the right things with the extra channels.
> 

Good news too.

> There appears to be a minor bug in that the input channels 8 and 9 (I
> think?), ie the top right two inputs (right of the spdif input), do not draw
> properly initially.  The numbers, and bits of the outline appear, but the
> sliders and bars don't draw.  At some later stage it all appears properly.
> 

Thanks for the report, I'll check this.

> Since this is done using a remote xsession to a windows PC and cygwin I
> can't be sure that this isn't just an X problem on win32...  It's a minor
> issue anyway
> 
> Actually more of an issue is that there is a ton of X traffic when stuff is
> playing which overwhelms my 802.11a network.  PC gets maxed out with a high
> cpu load just trying to squirt all the traffic over SSH.  I wonder if we
> could have variable refresh rates on the bouncing level meters?  Or perhaps
> you could investigate whether less of the screen could be updated perhaps
> (optimised drawing of boxes might reduce X traffic?)
> 

It is already supposed to be optimized to minimize redraws (but it's 
been a long time since I last worked on this part of the code, maybe it 
could be optimized further). Anyway redraws are indeed frequent. I have 
to say that my primary concern was cpu load with this. One plan I have 
for the future is to use OpenGL to reduce this load, but it'll be no 
help in your case.
This would just be a workaround, but maybe you could try tunneling VNC 
instead of X through SSH. You'll get de facto a smaller refresh rate and 
a reduced bandwidth consumption.

> Anyway, just a nice to have.  Thanks for what we have already though!
> 
> 
> 
>>Just to make it clear that this is a software problem and not the
>>problem that Tim, Paul and Mark reported, could you please try to route
>>an incoming signal (through analog or digital in) to your amp using
>>hdspmixer and tell me if the routed sound suffers from the same audio
>>corruption as the one being played back ? (this test should be done
>>while you're playing back a stuttering software stream)
> 
> 
> Sure, will do.  For what it's worth the same film works under xine, but not
> mplayer.
> 
> Alsaplayer also works fine I think (on different files), as does aplay.  I
> would be pretty sure that for a given FLAC file, alsaplayer would play it
> fine and mplayer would stutter.
> 
> I will try and grab the mplayer source and have a peer at the audio output
> code.
> 
> 

It sounds like an application design issue, then.

> 
>>I think the nicest feature you'd get with this design is the ability to
>>manipulate totalmix/hdspmixer's enhanced mixer (with this output stage)
>>from several applications in a coherent way (hdspmixer, ladpsa plugin,
>>script, midi automation, etc...), a feature that people constantly ask
>>for on the RME forum. This could also very well "sell" a few linux
>>installs to RME customers :)
> 
> 
> Please consider dropping me a line if you do any work on this.  I would love
> to test it as you go along.
> 

Sure, no problem.

Thanks,
Thomas




-------------------------------------------------------
The SF.Net email is sponsored by EclipseCon 2004
Premiere Conference on Open Tools Development and Integration
See the breadth of Eclipse activity. February 3-5 in Anaheim, CA.
http://www.eclipsecon.org/osdn

Re: HDSP 9632 driver and expansion cards

[Thomas Charbonnel]

Edward Wildgoose wrote :
>>>Just to make it clear that this is a software problem and not the
>>>problem that Tim, Paul and Mark reported, could you please try to route
>>>an incoming signal (through analog or digital in) to your amp using
>>>hdspmixer and tell me if the routed sound suffers from the same audio
>>>corruption as the one being played back ? (this test should be done
>>>while you're playing back a stuttering software stream)
>>
> 
> OK, I'm sure that I misunderstand what you are asking, but here is what I did:
> 
> Set mplayer playing with something like "mplayer file.mpg -ao alsa9:rme9632plug" - It sutters like crazy and mplayer reports "xruns
> of at least 10ms".
> 
> At the same time I spun a disk in the CD player, and changed the hdspmixer so that spdif mapped through to the front speakers
> instead of the mplayer audio.  CD output was fine, mplayer continued spewing errors up the screen.
>

That's exactly what I was asking for, thanks. This is also good news 
because you seem to be safe from the problem Tim, Paul and Mark are 
experiencing.

> I have just had a peek at the mplayer alsa output code, and compared with the verbose logs.  It does seem that mplayer has managed
> to set "chunksize" to 1024 and asked for 2 fragments - so this should be quite reasonable.  Switching to mmap or noblock mode makes
> no difference (neither does switching to the OSS device...)
> 
> Any ideas what could be happening?  As I said, the same problem happens with Mythtv under OSS, and I think also under alsa.  Some
> other apps are fine, alsaplayer for example.
> 

Still looks like a design issue for the misbehaving apps, that may also 
be emphasised by the system. Since I updated to the 2.6.x series, disk 
accesses are a major cause of audio dropouts for me as long as the audio 
apps do not run with realtime priviledges. SCHED_FIFOed apps on the 
other hand work as expected even with low latency settings.

> Quite peculiar...
> 
> I just checked and I actually have alsa 1.00 rc2.  Am I missing any driver updates?
> 

Nothing hdsp specific.

> asound.conf is pretty empty in case it matters.  Just an entry for the hardware device and a plug device for the same.
> 

There should be no problem here.

Thomas




-------------------------------------------------------
The SF.Net email is sponsored by EclipseCon 2004
Premiere Conference on Open Tools Development and Integration
See the breadth of Eclipse activity. February 3-5 in Anaheim, CA.
http://www.eclipsecon.org/osdn

Re: HDSP 9632 driver and expansion cards

[Edward Wildgoose]

> > Alsaplayer also works fine I think (on different files), as does aplay.  I
> > would be pretty sure that for a given FLAC file, alsaplayer would play it
> > fine and mplayer would stutter.
> >
> > I will try and grab the mplayer source and have a peer at the audio output
> > code.
> >
> >
>
> It sounds like an application design issue, then.

I have had a look at mplayers audio code.  I think I made some notes on a seperate email?  It looks fine actually, and it does seem
to be trying to set the buffer size to 1024 with 2 fragments.  This should sort out any latency problems I think.

Actually this system seems to work quite well with buffer sizes as low as 64 so I think it isn't latency issues with the system (at
least as far as the system being capable of running that fast, of course the app may not be written in a way that can feed that data
that quickly...)

It's interesting that it doesn't work under both OSS and Alsa with mplayer.

Obvious question really.  If you have your card in a machine, can you please try mplayer 1.00rc3 and see if it works for you, either
alsa or oss? (Or anyone else with one?)

The same setup works if I change nothing other than using either the RME 96/8 or the onboard intel8x0 card, so I don't think that
it's an app design issue, as such.  I suspect that the app might be doing something innocent which is perhaps even causing the
latency to be reset?  Sound plausible?

By the way, some other things which might be just my misunderstanding: "amixer cset numid=5" doesn't seem to do anything for me,
perhaps I have fumbled the syntax though?  Is it supposed to work?  Secondly, everytime I load hdspmixer it resets the card to
default settings.  Pressing the save button and saving to one of the card defaults doesn't seem to stay changed - should it do?
Perhaps I misunderstood how I should save changes to the mixer?


Thanks Thomas.  I think I'm slowly getting it all hooked up now.  Seems to be a very impressive card anyway.  Thanks again for your
work on the driver - much appreciated

Ed W



-------------------------------------------------------
The SF.Net email is sponsored by EclipseCon 2004
Premiere Conference on Open Tools Development and Integration
See the breadth of Eclipse activity. February 3-5 in Anaheim, CA.
http://www.eclipsecon.org/osdn