Hi friends,
I have a problem running Pd 0.52-1 on Ubuntu Studio 20.04 using Jack. I compile the project the same way I always did, but in this version the DSP doesn't work (Test Audio and Midi), and the GUI partially works (on other patches). The last version I can run normally is 0.51.4 (0.52-0 tests didn't work either). Running Pd with the -verbose flag does not show any relevant information. I also didn't find similar reports here on the list and on the net.
(system info) $ uname -a Linux brunorohde-desktop 5.13.0-35-lowlatency #40~20.04.1-Ubuntu SMP PREEMPT Mon Mar 7 10:14:12 UTC 2022 x86_64 x86_64 x86_64 GNU/Linux
(compilation) ./autogen.sh ./configure --enable-jack ./make
Does anyone have any information about this problem? Any tips on how to investigate the cause and solution for it?
Thanks
Bruno
Bruno, which version of Pd is available on Ubuntu via the apt package manager? On Debian it is
$ apt --all-versions list puredata Listing... Done puredata/bullseye-backports,bullseye-backports,now 0.52.1+ds0-1~bpo11+1 all [installed] puredata/stable,stable 0.51.4-1 all puredata/oldstable,oldstable 0.49.0-3 all
Any reason you want to compile it yourself?
Does it work when you specify $ pd -nosound ? Does it work with a different audio driver architecture, eg. alsa?
Does a pd (any version) installed via apt work?
and so on and so forth... ;)
best, P
Hi friends,
I have a problem running Pd 0.52-1 on Ubuntu Studio 20.04 using Jack. I compile the project the same way I always did, but in this version the DSP doesn't work (Test Audio and Midi), and the GUI partially works (on other patches). The last version I can run normally is 0.51.4 (0.52-0 tests didn't work either). Running Pd with the -verbose flag does not show any relevant information. I also didn't find similar reports here on the list and on the net.
(system info) $ uname -a Linux brunorohde-desktop 5.13.0-35-lowlatency #40~20.04.1-Ubuntu SMP PREEMPT Mon Mar 7 10:14:12 UTC 2022 x86_64 x86_64 x86_64 GNU/Linux
(compilation) ./autogen.sh ./configure --enable-jack ./make
Does anyone have any information about this problem? Any tips on how to investigate the cause and solution for it?
Thanks
Bruno
-- Bruno Faria Rohde
Pd-list@lists.iem.at mailing list UNSUBSCRIBE and account-management -> https://lists.puredata.info/listinfo/pd-list
Hi Peter,
Bruno, which version of Pd is available on Ubuntu via the apt package manager?
puredata/focal-updates,focal-updates,now 0.50.2-3ubuntu0.1 all [installed]
Any reason you want to compile it yourself?
to use the latest version; compiling always worked, it is the first time I have this problem.
Does it work when you specify $ pd -nosound ?
GUI works as expected when using -nosound, no more freezing
Does it work with a different audio driver architecture, eg. alsa?
Yes, I tried ALSA and it works, so it must be something related to Jack
Does a pd (any version) installed via apt work?
Yes, 0.50.2 works normally, as any other version I compiled previous to 0.52-0
Thank you for the hints!
Bruno
Em seg., 21 de mar. de 2022 às 10:41, Peter P. peterparker@fastmail.com escreveu:
Bruno, which version of Pd is available on Ubuntu via the apt package manager? On Debian it is
$ apt --all-versions list puredata Listing... Done puredata/bullseye-backports,bullseye-backports,now 0.52.1+ds0-1~bpo11+1 all [installed] puredata/stable,stable 0.51.4-1 all puredata/oldstable,oldstable 0.49.0-3 all
Any reason you want to compile it yourself?
Does it work when you specify $ pd -nosound ? Does it work with a different audio driver architecture, eg. alsa?
Does a pd (any version) installed via apt work?
and so on and so forth... ;)
best, P
- Bruno Rohde brunorohde@gmail.com [2022-03-21 14:27]:
Hi friends,
I have a problem running Pd 0.52-1 on Ubuntu Studio 20.04 using Jack. I compile the project the same way I always did, but in this version the
DSP
doesn't work (Test Audio and Midi), and the GUI partially works (on other patches). The last version I can run normally is 0.51.4 (0.52-0 tests didn't work either). Running Pd with the -verbose flag does not show any relevant information. I also didn't find similar reports here on the list and on the net.
(system info) $ uname -a Linux brunorohde-desktop 5.13.0-35-lowlatency #40~20.04.1-Ubuntu SMP PREEMPT Mon Mar 7 10:14:12 UTC 2022 x86_64 x86_64 x86_64 GNU/Linux
(compilation) ./autogen.sh ./configure --enable-jack ./make
Does anyone have any information about this problem? Any tips on how to investigate the cause and solution for it?
Thanks
Bruno
-- Bruno Faria Rohde
Pd-list@lists.iem.at mailing list UNSUBSCRIBE and account-management ->
https://lists.puredata.info/listinfo/pd-list
Pd-list@lists.iem.at mailing list UNSUBSCRIBE and account-management -> https://lists.puredata.info/listinfo/pd-list
Hi Peter,
Bruno, which version of Pd is available on Ubuntu via the apt package manager?
puredata/focal-updates,focal-updates,now 0.50.2-3ubuntu0.1 all [installed]
Does Ubuntu have the latest and greatest Pd in its backports perhaps? https://help.ubuntu.com/community/UbuntuBackports
I just installed the latest version enabling the "jammy" repo (the one that has 0.52-1), and the result was the same, no DSP using Jack. No luck so far :(
Em seg., 21 de mar. de 2022 às 13:14, Peter P. peterparker@fastmail.com escreveu:
- Bruno Rohde brunorohde@gmail.com [2022-03-21 16:31]:
Hi Peter,
Bruno, which version of Pd is available on Ubuntu via the apt package manager?
puredata/focal-updates,focal-updates,now 0.50.2-3ubuntu0.1 all
[installed]
Does Ubuntu have the latest and greatest Pd in its backports perhaps? https://help.ubuntu.com/community/UbuntuBackports
Pd-list@lists.iem.at mailing list UNSUBSCRIBE and account-management -> https://lists.puredata.info/listinfo/pd-list
I just installed the latest version enabling the "jammy" repo (the one that has 0.52-1), and the result was the same, no DSP using Jack. No luck so far :(
But great that you tried it and hence gathered more information about your problem. You know that you can increase Pd's debugging level and/or start it with the -verbose flag to get more detailed error messages. Does Pd post anything there when you start DSP?
On the jackd side: Are you getting any messages from jackd, for example by looking at qjackctl's "Messages"?
And do you have a second alternative sound card you could try?
cheersz, P
You know that you can increase Pd's debugging level and/or start it with the -verbose flag to get more detailed error messages. Does Pd post anything there when you start DSP?
Yes, I know, but there's no info about the problem, that's the main reason I came to ask here (for the first time in 10 years using Pd, btw).
On the jackd side: Are you getting any messages from jackd, for example
by looking at qjackctl's "Messages"?
I see nothing special from jackd side, looks like everything is working properly, and Pd shows up in patchbays like qjackctl and catia. Starting jackd via terminal with -v(erbose) and comparing 0.51.4 (audio ok) and 0.52.1 (no audio) the output looks the same.
And do you have a second alternative sound card you could try?
Not now, but it doesn't seem it will make a difference in this case. When I have my other sound card in the next few days I'll try again.
Do you think it's worth opening an issue on pd's github?
Thank you for the support,
Cheers
Em seg., 21 de mar. de 2022 às 19:58, Peter P. peterparker@fastmail.com escreveu:
- Bruno Rohde brunorohde@gmail.com [2022-03-21 18:20]:
I just installed the latest version enabling the "jammy" repo (the one
that
has 0.52-1), and the result was the same, no DSP using Jack. No luck so far :(
But great that you tried it and hence gathered more information about your problem. You know that you can increase Pd's debugging level and/or start it with the -verbose flag to get more detailed error messages. Does Pd post anything there when you start DSP?
On the jackd side: Are you getting any messages from jackd, for example by looking at qjackctl's "Messages"?
And do you have a second alternative sound card you could try?
cheersz, P
Pd-list@lists.iem.at mailing list UNSUBSCRIBE and account-management -> https://lists.puredata.info/listinfo/pd-list
Am 21. März 2022 14:18:21 MEZ schrieb Bruno Rohde brunorohde@gmail.com: anyone have any information about this problem? Any tips on how to
investigate the cause and solution for it?
Turn OFF (or ON) the "use callbacks" checkbox in the audio dialog.
mfg.sfg.jfd IOhannes
Turn OFF (or ON) the "use callbacks" checkbox in the audio dialog.
That did the trick, with callbacks ON DSP works. I never used it before, and to be honest, I don't know what this does, but for now it solved the problem. Could you (or someone) explain briefly what this is for? Is there any impact elsewhere (maybe performance) by using it enabled?
Anyway, thank you so much for the hint.
Cheers
Em ter., 22 de mar. de 2022 às 03:46, IOhannes m zmölnig zmoelnig@iem.at escreveu:
Am 21. März 2022 14:18:21 MEZ schrieb Bruno Rohde brunorohde@gmail.com: anyone have any information about this problem? Any tips on how to
investigate the cause and solution for it?
Turn OFF (or ON) the "use callbacks" checkbox in the audio dialog.
mfg.sfg.jfd IOhannes
Pd-list@lists.iem.at mailing list UNSUBSCRIBE and account-management -> https://lists.puredata.info/listinfo/pd-list
Could you (or someone) explain briefly what this is for?
If "callbacks" is ON, Pd runs directly in the audio callback. This gives you lower latencies, but less "leeway" for CPU fluctuations - unless you increase the hardware/jack buffer size accordingly.
If "callbacks" is OFF, Pd uses a polling scheduler which runs independently from the audio callback and exchanges audio via a (lock-free) ringbuffer. This ringbuffer essentially adds some extra latency, so that Pd is allowed to "fall behind", as long as it can catch up again. You can control the latency with the "Delay" setting, independently from the actual hardware/jack buffer size, which can be quite useful.
As a rule of thumb: don't use "callbacks" unless you *know* that you need it :-)
Christof
On 22.03.2022 14:45, Bruno Rohde wrote:
Turn OFF (or ON) the "use callbacks" checkbox in the audio dialog.
That did the trick, with callbacks ON DSP works. I never used it before, and to be honest, I don't know what this does, but for now it solved the problem. Could you (or someone) explain briefly what this is for? Is there any impact elsewhere (maybe performance) by using it enabled?
Anyway, thank you so much for the hint.
Cheers
Em ter., 22 de mar. de 2022 às 03:46, IOhannes m zmölnig zmoelnig@iem.at escreveu:
Am 21. März 2022 14:18:21 MEZ schrieb Bruno Rohde <brunorohde@gmail.com>: anyone have any information about this problem? Any tips on how to >investigate the cause and solution for it? > Turn OFF (or ON) the "use callbacks" checkbox in the audio dialog. mfg.sfg.jfd IOhannes _______________________________________________ Pd-list@lists.iem.at mailing list UNSUBSCRIBE and account-management -> https://lists.puredata.info/listinfo/pd-list
-- Bruno Faria Rohde
Pd-list@lists.iem.at mailing list UNSUBSCRIBE and account-management ->https://lists.puredata.info/listinfo/pd-list
On Tue, 2022-03-22 at 10:45 -0300, Bruno Rohde wrote:
Turn OFF (or ON) the "use callbacks" checkbox in the audio dialog.
That did the trick, with callbacks ON DSP works. I never used it before, and to be honest, I don't know what this does, but for now it solved the problem. Could you (or someone) explain briefly what this is for? Is there any impact elsewhere (maybe performance) by using it enabled?
From what I understand, using callbacks means that Pd processes audio as soon as it is provided by JACK. This means, Pd doesn't add additional latency as JACK client. But it also means, that when Pd isn't quick enought with writing back processed data, the whole JACK process with all clients experiences a drop-out.
Without callbacks, Pd is less tightly coupled to JACK and uses a ringbuffer for input and output. The size of the ringbuffer is set in the 'delay(ms)' setting in the audio settings. From my experience, setting it too low causes Pd to not process audio at all. In my experience, this value is below 5ms, but this probably depends on JACK's blocksize setting.
Roman
On 22.03.2022 15:00, Roman Haefeli wrote:
From my experience, setting it too low causes Pd to not process audio at all. In my experience, this value is below 5ms, but this probably depends on JACK's blocksize setting.
This might actually explain the issue!
@Bruno: what is you current value of "Delay" in the audio settings and what is your Jack buffer size?
Actually, before Pd 0.52, the polling scheduler did not really work correctly together with the Jack backend, i.e. the "Delay" setting was basically ignored and it just did double buffering. This has been fixed.
Anyway, Pd should make sure that "Delay" is not smaller than the Jack buffer size, thus preventing silent output. If that's not the case, please file a bug report!
Thanks Christof and Roman for the clarification, now I understand it a little!
Actually, before Pd 0.52, the polling scheduler did not really work
correctly together with the Jack backend, i.e. the "Delay" setting was basically ignored and it just did double buffering. This has been fixed.
Anyway, Pd should make sure that "Delay" is not smaller than the Jack buffer size, thus preventing silent output. If that's not the case, please file a bug report!
You're right! I was using a 10ms delay with 512 buffer size / 3 periods on Jack. As soon as I changed to 12ms DSP worked normally. I used 10ms on Pd for a long time, but it was ignored as you said, and now it has turned into my "problem".
So, that is not a bug, but maybe this limit should be clearly stated somewhere to avoid confusion.
Thank you all so much!
Em ter., 22 de mar. de 2022 às 11:08, Christof Ressi info@christofressi.com escreveu:
On 22.03.2022 15:00, Roman Haefeli wrote:
From my experience, setting it too low causes Pd to not process audio at all. In my experience, this value is below 5ms, but this probably depends on JACK's blocksize setting.
This might actually explain the issue!
@Bruno: what is you current value of "Delay" in the audio settings and what is your Jack buffer size?
Actually, before Pd 0.52, the polling scheduler did not really work correctly together with the Jack backend, i.e. the "Delay" setting was basically ignored and it just did double buffering. This has been fixed.
Anyway, Pd should make sure that "Delay" is not smaller than the Jack buffer size, thus preventing silent output. If that's not the case, please file a bug report!
Pd-list@lists.iem.at mailing list UNSUBSCRIBE and account-management -> https://lists.puredata.info/listinfo/pd-list
good that we could figure out the problem!
So, that is not a bug, but maybe this limit should be clearly stated somewhere to avoid confusion.
Actually, there *is* a bug: Pd should not just output silence if the user accidentally sets a too small value for "Delay". It should either clamp the value or post a warning. It would be great if you could file a bug report.
Christof
Actually, there *is* a bug: Pd should not just output silence if the user accidentally sets a too small value for "Delay". It should either clamp the value or post a warning. It would be great if you could file a bug report.
Agreed, I'll file the bug report.
Em ter., 22 de mar. de 2022 às 11:34, Christof Ressi info@christofressi.com escreveu:
good that we could figure out the problem!
So, that is not a bug, but maybe this limit should be clearly stated somewhere to avoid confusion.
Actually, there *is* a bug: Pd should not just output silence if the user accidentally sets a too small value for "Delay". It should either clamp the value or post a warning. It would be great if you could file a bug report.
Christof _______________________________________________ Pd-list@lists.iem.at mailing list UNSUBSCRIBE and account-management -> https://lists.puredata.info/listinfo/pd-list