netjack -C shows 3 times the cpu o'head of -H

[Geoff Beasley-2]

is there any news on this issue? Ticket #136. I'm hoping to open the new  
studio in a couple of weeks and I really need to know if there is a fix or  
work-around apart from not running in duplex mode.

best,

g.
_______________________________________________
Jack-Devel mailing list
Jack-Devel@...
http://lists.jackaudio.org/listinfo.cgi/jack-devel-jackaudio.org

[Stéphane Letz]

Le 15 sept. 2009 à 07:01, Geoff Beasley a écrit :

> is there any news on this issue? Ticket #136. I'm hoping to open the  
> new studio in a couple of weeks and I really need to know if there  
> is a fix or work-around apart from not running in duplex mode.
>
> best,
>
> g.


I'll try to investigate (although not much time for that stuff right  
now..)

Stéphane
_______________________________________________
Jack-Devel mailing list
Jack-Devel@...
http://lists.jackaudio.org/listinfo.cgi/jack-devel-jackaudio.org

[Stéphane Letz]

Le 15 sept. 2009 à 07:01, Geoff Beasley a écrit :

> is there any news on this issue? Ticket #136. I'm hoping to open the  
> new studio in a couple of weeks and I really need to know if there  
> is a fix or work-around apart from not running in duplex mode.
>
> best,
>
> g.
> _______________________________________________


OK, actually the network latency was (incorrectly) taken in account in  
CPU load measurement.  Should be fixed in r3643.

Can you test and report?

Thanks

Stephane
_______________________________________________
Jack-Devel mailing list
Jack-Devel@...
http://lists.jackaudio.org/listinfo.cgi/jack-devel-jackaudio.org

[Geoff Beasley-2]

On Tue, 15 Sep 2009 22:58:37 +1000, Stéphane Letz <letz@...> wrote:

>
> Le 15 sept. 2009 à 07:01, Geoff Beasley a écrit :
>
>> is there any news on this issue? Ticket #136. I'm hoping to open the  
>> new studio in a couple of weeks and I really need to know if there is a  
>> fix or work-around apart from not running in duplex mode.
>>
>> best,
>>
>> g.
>> _______________________________________________
>
>
> OK, actually the network latency was (incorrectly) taken in account in  
> CPU load measurement.  Should be fixed in r3643.
>
> Can you test and report?
>
> Thanks
>
> Stephane

that did it; -C 40 -P40 =6%cpu. thanks so much for looking at this  
Stephane. however,transport is broken too as i reported yesterday. the  
master jack merely increments => with successive play/stops from ardour  
and/or muse. slave behaviour is odd as well... ardours' playhead jumps  
when  moved with the mouse rather than moving smoothly when 'time master'  
but will move smoothly when not time master.

best,

g.
--
_______________________________________________
Jack-Devel mailing list
Jack-Devel@...
http://lists.jackaudio.org/listinfo.cgi/jack-devel-jackaudio.org

[Stéphane Letz]

>
> that did it; -C 40 -P40 =6%cpu. thanks so much for looking at this  
> Stephane. however,transport is broken too as i reported yesterday.  
> the master jack merely increments => with successive play/stops  
> from ardour and/or muse. slave behaviour is odd as well... ardours'  
> playhead jumps  when  moved with the mouse rather than moving  
> smoothly when 'time master' but will move smoothly when not time  
> master.
>
> best,
>
> g.
> --

Yep, I did a fix for transport yesterday, but yes, locate does not  
yet work corectly, and I don't know when "timebase master" is used.
I"ll try to have another look at end of this week.

Stéphane
_______________________________________________
Jack-Devel mailing list
Jack-Devel@...
http://lists.jackaudio.org/listinfo.cgi/jack-devel-jackaudio.org

[Geoff Beasley-2]

On Fri, 18 Sep 2009 23:15:36 +1000, Stéphane Letz <letz@...> wrote:


> Ok, I did a simple change that "deactivate" transport control from the  
> slave, now the master only can control the transport. Note also that  
> locate on the master when stopped is not reflected on the slave, until  
> master start again.
>
> Get the latest SVN.
>
> Stephane

great Stephane. built and testing; will give report in 24 hours

thanks again so much for taking the time to look at this so quickly.

best,

g.

--
_______________________________________________
Jack-Devel mailing list
Jack-Devel@...
http://lists.jackaudio.org/listinfo.cgi/jack-devel-jackaudio.org

[Fons Adriaensen-2]

On Sat, Sep 19, 2009 at 06:10:44PM +1000, Geoff Beasley wrote:

> On Fri, 18 Sep 2009 23:15:36 +1000, Stéphane Letz <letz@...> wrote:
>
>
>> Ok, I did a simple change that "deactivate" transport control from the
>> slave, now the master only can control the transport. Note also that
>> locate on the master when stopped is not reflected on the slave, until
>> master start again.

I missed the start of this thread. Are you changing the way Jack
transport works, or is this something else ?

Ciao,

--
FA

Io lo dico sempre: l'Italia è troppo stretta e lunga.

_______________________________________________
Jack-Devel mailing list
Jack-Devel@...
http://lists.jackaudio.org/listinfo.cgi/jack-devel-jackaudio.org

[Stéphane Letz]

Le 19 sept. 09 à 00:50, Fons Adriaensen a écrit :

> On Sat, Sep 19, 2009 at 06:10:44PM +1000, Geoff Beasley wrote:
>
>> On Fri, 18 Sep 2009 23:15:36 +1000, Stéphane Letz <letz@...>  
>> wrote:
>>
>>
>>> Ok, I did a simple change that "deactivate" transport control  
>>> from the
>>> slave, now the master only can control the transport. Note also that
>>> locate on the master when stopped is not reflected on the slave,  
>>> until
>>> master start again.
>
> I missed the start of this thread. Are you changing the way Jack
> transport works, or is this something else ?
>
> Ciao,
>

I'm try to simplify (so hoping having something working..) the way  
transport works between the master and slaves when NetJack2 is used.  
First NetJack2 version tried to have either master or slaves be able  
control everything. This did not work correctly, so I'm just getting  
back to a simpler model.

Stephane
_______________________________________________
Jack-Devel mailing list
Jack-Devel@...
http://lists.jackaudio.org/listinfo.cgi/jack-devel-jackaudio.org

[Geoff Beasley-2]

Stephane;
so far so good. jack2 seems more stable now for me ..certainly fewer xruns  
so far. Muse has no way of selecting/deselecting jack transport yet, like  
ardour has, however i can just stop and restart with -t0 on that slave and  
all will be as it was (always had to do that anyway with netjack1 as you  
described). and that is working fine.

thanks again for this;

best,

g.
_______________________________________________
Jack-Devel mailing list
Jack-Devel@...
http://lists.jackaudio.org/listinfo.cgi/jack-devel-jackaudio.org

[Geoff Beasley-2]

is that a possibility at some stage?  there are not that many jackmidi  
aware apps yet... i suppose it's a big problem, oui??

best,

g.
_______________________________________________
Jack-Devel mailing list
Jack-Devel@...
http://lists.jackaudio.org/listinfo.cgi/jack-devel-jackaudio.org

[Stéphane Letz]

Le 20 sept. 2009 à 21:50, Geoff Beasley a écrit :

> is that a possibility at some stage?  there are not that many  
> jackmidi aware apps yet... i suppose it's a big problem, oui??
>
> best,
>
> g.

You can define JACK MIDI ports in NetJack2. What is actually missing?

Thanks

Stéphane

_______________________________________________
Jack-Devel mailing list
Jack-Devel@...
http://lists.jackaudio.org/listinfo.cgi/jack-devel-jackaudio.org

[Geoff Beasley-2]

OK, now to fill you in fully. 4 machines on a gigabit network, master @  
2/128/48k on 4 C-Ports (ice1712) using alsa-multi; slave 1 with Ardour  
loaded with a project running 16 tracks and about 5 LV2 plugs (and opera  
open) netjack2 with 40 I/O, another with fst/Kontakt2 with 3 instruments  
(big) loaded netjack2 with 24 outputs, another with Linuxsampler with 3  
instruments loaded (big) and Muse running and looped netjack2 with 24  
outputs ; all running for 15 1/2 hours with 1 single xrunlet (1). Couldn't  
do that before ;)


looking very good indeed...


best,

g.
_______________________________________________
Jack-Devel mailing list
Jack-Devel@...
http://lists.jackaudio.org/listinfo.cgi/jack-devel-jackaudio.org

[Geoff Beasley-2]

on the down side, slaves don't 'chase' to the jack master time. they  
restart @ 0:0:1:0 (on both ardour and muse on separate machines) when  
rewind is pressed on the master. jumping along the master timeline  
produces no equivalent jump in the slaves; only a 'reset' to
0:0:1:0.

hth.

best,

g.
_______________________________________________
Jack-Devel mailing list
Jack-Devel@...
http://lists.jackaudio.org/listinfo.cgi/jack-devel-jackaudio.org

[Geoff Beasley-2]

On Mon, 21 Sep 2009 16:08:28 +1000, Patrick Shirky  
<pshirkey@...> wrote:

> Hi,
>
>
> That's very cool.
>
> Would you be interested in making a quicktoot about this setup?
>
> I would be happy to build out the screenshots and format it etc if you  
> wanted to do a quick brain dump while it is all still fresh for you.
>
> I would also be some free marketing for your studio and I could link  
> back to your site to increase your web search page ranking.
>
>
> Cheers.
>
> Patrick Shirkey
> Boost Hardware Ltd

i'd like that. once transport is working well, i intend to. thanks for the  
offer of help; i'd like that too ;) there are 6 production machines in the  
current studio array; LinAS is the audio server: hosting the 40 channels  
of I/O via netjack2 to the rest of the system. LinDAW is obviously the  
daw, Ardour 2 and plugs. LinSampler is for Kontakt2, Zynaddsubfx,Pahsex  
etc, LinSeq is for Muse 1.0-rc3-cvs (which is mighty btw ) and  
LinuxSampler/Fantasia, LinFX which basically hosts Jconv/Zita and the Pori  
(and other) IR's back into the desk in realtime and LinVS is a video  
server that can  handle video playback sync using Xjadeo (piped through  
the building) and some simple video nle apps for adding media in and out  
of the studio. The last machine is the last one to be assembled;  
everything else is working beautifully; just add synchronization :)

I'll keep you posted on developments Patrick,

best,

g.
_______________________________________________
Jack-Devel mailing list
Jack-Devel@...
http://lists.jackaudio.org/listinfo.cgi/jack-devel-jackaudio.org

[James Warden]

Hi Stephane,

I wonder whether you changed something in the jack2 transport. I updated my svn version to benefit from the recent netjack2 developments. And I now have some serious issues sync'ing rosegarden and ardour in a simple jack transport setup. It used to work fine, I haven't changed either ardour or RG so I can only suspect the jack2 update.

The symptom is: I have to reset the jack transport to 0 in order to have ardour and RG syncing nice. If I start the transport from a random time point (whether from ardour or RG), there's a substantial delay between both apps, which makes the whole sync'ing via jack completely useless. I have not tried yet with jack1 as I was running out of time.

I cannot give you more info as I am at work right now. But maybe you already know something about this ?

J.


     
_______________________________________________
Jack-Devel mailing list
Jack-Devel@...
http://lists.jackaudio.org/listinfo.cgi/jack-devel-jackaudio.org

[Stéphane Letz]

Le 22 sept. 2009 à 10:10, James Warden a écrit :

> Hi Stephane,
>
> I wonder whether you changed something in the jack2 transport. I  
> updated my svn version to benefit from the recent netjack2  
> developments. And I now have some serious issues sync'ing rosegarden  
> and ardour in a simple jack transport setup. It used to work fine, I  
> haven't changed either ardour or RG so I can only suspect the jack2  
> update.
>
> The symptom is: I have to reset the jack transport to 0 in order to  
> have ardour and RG syncing nice. If I start the transport from a  
> random time point (whether from ardour or RG), there's a substantial  
> delay between both apps, which makes the whole sync'ing via jack  
> completely useless. I have not tried yet with jack1 as I was running  
> out of time.
>
> I cannot give you more info as I am at work right now. But maybe you  
> already know something about this ?
>
> J.
>

Nothing has been changed in the jack2 transport. Some changes have  
been done only at NetJack2 level. Was it working with 1.9.3 version?

Thanks

Stephane
_______________________________________________
Jack-Devel mailing list
Jack-Devel@...
http://lists.jackaudio.org/listinfo.cgi/jack-devel-jackaudio.org

[James Warden]

It worked until last week (I had not used RG and ardour via jack transport for a few days so I cannot tell you exactly when it started to screw up). I was also regularly updating jack2 from svn. I will try some more testing tonight. Note that I also switched to kernel 2.6.31-rt11. Not sure if that can have an impact but that's a change that happened at the same time I last updated jack2.

I'll get back to you when I have some more substantial info.

J.
--- On Tue, 9/22/09, Stéphane Letz <letz@...> wrote:

> From: Stéphane Letz <letz@...>
> Subject: Re: [Jack-Devel] Jack transport: time location messed up ?
> To: "James Warden" <warjamy@...>
> Cc: "jack-devel@..." <jack-devel@...>
> Date: Tuesday, September 22, 2009, 4:27 AM
>
> Le 22 sept. 2009 à 10:10, James Warden a écrit :
>
> > Hi Stephane,
> >
> > I wonder whether you changed something in the jack2
> transport. I updated my svn version to benefit from the
> recent netjack2 developments. And I now have some serious
> issues sync'ing rosegarden and ardour in a simple jack
> transport setup. It used to work fine, I haven't changed
> either ardour or RG so I can only suspect the jack2 update.
> >
> > The symptom is: I have to reset the jack transport to
> 0 in order to have ardour and RG syncing nice. If I start
> the transport from a random time point (whether from ardour
> or RG), there's a substantial delay between both apps, which
> makes the whole sync'ing via jack completely useless. I have
> not tried yet with jack1 as I was running out of time.
> >
> > I cannot give you more info as I am at work right now.
> But maybe you already know something about this ?
> >
> > J.
> >
>
> Nothing has been changed in the jack2 transport. Some
> changes have been done only at NetJack2 level. Was it
> working with 1.9.3 version?
>
> Thanks
>
> Stephane

     
_______________________________________________
Jack-Devel mailing list
Jack-Devel@...
http://lists.jackaudio.org/listinfo.cgi/jack-devel-jackaudio.org

[Geoff Beasley-2]

Stephane,

I have had to regress to jack1 as netjack2 and jack2 transport is utterly
useless for now. let me know if/when you get a chance to look at it again and
i'll eagerly rebuild ;)

jack1/netjack transport is solid for me.


best,

g.
_______________________________________________
Jack-Devel mailing list
Jack-Devel@...
http://lists.jackaudio.org/listinfo.cgi/jack-devel-jackaudio.org

[Geoff Beasley-2]

Sorry Stephane if it seemed I was complaining, I wasn't truely.. ;)

 As i said on Tuesday,transport with netjack2 is broken in that if you start
transport on the master, the slaves simply start from 1 sec instead of 0 sec .
there is no sync between the slaves and master at all. they only start (always
from 1 sec not 0 sec) and stop.This is as you described to me ,that 'locate'
doesn't work yet. Let me know if that isn't clear enough.

best,

g.
_______________________________________________
Jack-Devel mailing list
Jack-Devel@...
http://lists.jackaudio.org/listinfo.cgi/jack-devel-jackaudio.org

[James Warden]

Hi Stephane,

I tried with jack1 and the problem is still visible.
however, I also tried another ardour / rosegarden session and it worked fine.

So I suspect either ardour or RG to have introduced a synchronization issue.

The session starts with one time signature (4/4) at BPM 167 and switches to a new time signature (15/4) at the same BPM after one measure at 4/4. What happens is that if I place the transport shuttle to a certain bar in RG, the ardour transport shuttle is moved at about 1 second too early (I counted the number of frames but I don't remember it now). This delay is consistent everytime you move the shuttle to another bar in RG: ardour's shuttle will follow but misses the bar.

If I reset the shuttle back to 0, RG and ardour are sync'ed OK. I don't think this is a jack bug any longer.

J.




--- On Tue, 9/22/09, James Warden <warjamy@...> wrote:

> From: James Warden <warjamy@...>
> Subject: Re: [Jack-Devel] Jack transport: time location messed up ?
> To: "jack-devel@..." <jack-devel@...>
> Date: Tuesday, September 22, 2009, 5:28 AM
> It worked until last week (I had not
> used RG and ardour via jack transport for a few days so I
> cannot tell you exactly when it started to screw up). I was
> also regularly updating jack2 from svn. I will try some more
> testing tonight. Note that I also switched to kernel
> 2.6.31-rt11. Not sure if that can have an impact but that's
> a change that happened at the same time I last updated
> jack2.
>
> I'll get back to you when I have some more substantial
> info.
>
> J.
> --- On Tue, 9/22/09, Stéphane Letz <letz@...>
> wrote:
>
> > From: Stéphane Letz <letz@...>
> > Subject: Re: [Jack-Devel] Jack transport: time
> location messed up ?
> > To: "James Warden" <warjamy@...>
> > Cc: "jack-devel@..."
> <jack-devel@...>
> > Date: Tuesday, September 22, 2009, 4:27 AM
> >
> > Le 22 sept. 2009 à 10:10, James Warden a écrit :
> >
> > > Hi Stephane,
> > >
> > > I wonder whether you changed something in the
> jack2
> > transport. I updated my svn version to benefit from
> the
> > recent netjack2 developments. And I now have some
> serious
> > issues sync'ing rosegarden and ardour in a simple
> jack
> > transport setup. It used to work fine, I haven't
> changed
> > either ardour or RG so I can only suspect the jack2
> update.
> > >
> > > The symptom is: I have to reset the jack
> transport to
> > 0 in order to have ardour and RG syncing nice. If I
> start
> > the transport from a random time point (whether from
> ardour
> > or RG), there's a substantial delay between both apps,
> which
> > makes the whole sync'ing via jack completely useless.
> I have
> > not tried yet with jack1 as I was running out of
> time.
> > >
> > > I cannot give you more info as I am at work right
> now.
> > But maybe you already know something about this ?
> > >
> > > J.
> > >
> >
> > Nothing has been changed in the jack2 transport. Some
> > changes have been done only at NetJack2 level. Was it
> > working with 1.9.3 version?
> >
> > Thanks
> >
> > Stephane
>
>
>      
> _______________________________________________
> Jack-Devel mailing list
> Jack-Devel@...
> http://lists.jackaudio.org/listinfo.cgi/jack-devel-jackaudio.org
>

     
_______________________________________________
Jack-Devel mailing list
Jack-Devel@...
http://lists.jackaudio.org/listinfo.cgi/jack-devel-jackaudio.org