Charles, Do you mean from a terminal startup pd and jack with sudo or to actually log on to the system as root from startup? I don't know how to do that besides running the recovery startup. Linux has changed in the last 15 years. I did try changing the runlevel of the timers and sound card IRQ as well as adding myself to the sound group on another installation and still had the same problem.
Bernard, I guess I should have said performance is better on vista than puredyne with said latency settings. I did not actually measure latency. The audio does not drop out with latency set at 50ms on vista but does with puredyne. 2.6.31-9-rt is the kernel i'm using. Yes 50ms is bad for rt performance that is why I'm trying to get this working with linux. It is impossible it seems with vista to get the latency any lower. It is crazy 10 years ago latency was not an issue with 10 times slower machines. The current processor is an AMD Sempron LE 1250 at 2.2GHz.
Thanks and Thanks
Billy Stiltner wrote:
Charles, Do you mean from a terminal startup pd and jack with sudo
yes sudo will do.
I also would recommend trying Archlinux, which is really up-to-date. But as they say that's no picnic.
It is impossible it seems with vista to get the latency any lower.
Try using another driver. It seems in some cases the directX driver gives better results than Asio. Just try every driver in turn. That's what I did on a old Pentium4 running XP with a low-end terratec sound card: Asio gives 40ms, directX gives 12ms. Go figure...
good luck !
Charlot
Some general and useful hints:
==> Install up-to-date kernel (2.6.38.1) or the last rt-patch (still 2.6.33)
==> Install up-to-date JACK (jack1 or jack2, try both and see -- jack2 has support for multi-core)
==> Edit the /etc/security/limits.conf to allow you (user) to have rt priviledges
@audio soft cpu unlimited @audio - rtprio 100 @audio - memlock unlimited
And add yourself to the audio group.
==> Use this script to help you check other realtime related configurations:
http://code.google.com/p/realtimeconfigquickscan/
Hope that helps!
2011/3/25 Bernardo Barros bernardobarros2@gmail.com:
==> Edit the /etc/security/limits.conf to allow you (user) to have rt priviledges
@audio soft cpu unlimited @audio - rtprio 100 @audio - memlock unlimited
And add yourself to the audio group.
Forgot to say: NEVER start anything (jack, pd etc...) as root, it's not safe. This configuration above is enough.
==> Edit the /etc/security/limits.conf to allow you (user) to have rt priviledges
@audio soft cpu unlimited @audio - rtprio 100 @audio - memlock unlimited
And add yourself to the audio group.
Forgot to say: NEVER start anything (jack, pd etc...) as root, it's not safe. This configuration above is enough.
I did try something similar but with cpu at 95 I think and memlock at 30% of ram. still same glitching. once I get the glitching to stop when pd window is open I will mess around with that. again. I guess If I set cpu to unlimited I will have to catch the cpu overload in pd and stop dsp.
2011/3/25 Billy Stiltner billy.stiltner@gmail.com:
I did try something similar but with cpu at 95 I think and memlock at 30% of ram. still same glitching. once I get the glitching to stop when pd window is open I will mess around with that. again. I guess If I set cpu to unlimited I will have to catch the cpu overload in pd and stop dsp.
Ooops. Recent Ubuntu changed that, it's not in limits.conf anymore. When you install JACK Ubuntu creates a "audio.conf" file. I think if you use Ubuntu 10.10 just installing jack does the cofiguration for you.
Try using another driver. It seems in some cases the directX driver gives better results than Asio. Just try every driver in turn. That's what I did on a old Pentium4 running XP with a low-end terratec sound card: Asio gives 40ms, directX gives 12ms. Go figure...
good luck !
Charlot
With the directx drivers selected audio drops out till latency is set to 150ms. even midi glitches till then. I didn't think that Vista used directx drivers but evidently there is some kind of directx layer in there.
Does the glitches occur with other software ? Since you have a Windows handy, you could check with a trial version of maxmsp or vvvv. If you can reproduce the same/lame results, then pd is out of cause.
Also maybe the nvidia soundcard is just a piece of sh*t :) ?
Thanks Bernardo
Charles Possibly so.
I wonder if I install the SB live if it will be autodetected. and the drivers are allready a part of the kernel or will I have to patch the kernel or something then reinstall all the audio software. At least I would determine if it is the soundcard or sound card drivers or not.
From Vista I was just able to set the latency to 12ms in pd 042.5
vanilla without loading any extra libraries besides the moog~filter from ggee and this did not have any audio dropouts. This is without GEM
I will try this same thing with the same patch in Uuntu Studio 10.10.
Earlier I noticed from puredyne that whenever the audio glitches the cpu% jumps to 1% instead of 0 for soundcard IRQ process. I tried setting the priority of pd and pdgui to -10 and this seemed to help a bit. Maybe I need to adjust the buffers for ALSA to match the buffers of ASIO for ALL. which is set at 512bytes with a buffer offset of 4ms.
On 3/25/11, Billy Stiltner billy.stiltner@gmail.com wrote:
Thanks Bernardo
Charles Possibly so.
I wonder if I install the SB live if it will be autodetected. and the drivers are allready a part of the kernel or will I have to patch the kernel or something then reinstall all the audio software. At least I would determine if it is the soundcard or sound card drivers or not.
Well with the stock install of Ubuntu studio 10.10 except for the graphics accelerator upgrade the audio was still glitching the same as with 9.10 puredyne. So I done all the upgrades possible and that seemed to get the patch running without gem with a 50ms buffer. There is no glitching while messing with the guis in pure data like there is in 9.10 so I'm guessing it's either a window drawing routine or something to do with alsa or the audio codecs that got upgraded. Now maybe I can figure out how to tinker with the alsa drivers to lower the latency.
On 3/25/11, Billy Stiltner billy.stiltner@gmail.com wrote:
From Vista I was just able to set the latency to 12ms in pd 042.5 vanilla without loading any extra libraries besides the moog~filter from ggee and this did not have any audio dropouts. This is without GEM
I will try this same thing with the same patch in Uuntu Studio 10.10.
Earlier I noticed from puredyne that whenever the audio glitches the cpu% jumps to 1% instead of 0 for soundcard IRQ process. I tried setting the priority of pd and pdgui to -10 and this seemed to help a bit. Maybe I need to adjust the buffers for ALSA to match the buffers of ASIO for ALL. which is set at 512bytes with a buffer offset of 4ms.
On 3/25/11, Billy Stiltner billy.stiltner@gmail.com wrote:
Thanks Bernardo
Charles Possibly so.
I wonder if I install the SB live if it will be autodetected. and the drivers are allready a part of the kernel or will I have to patch the kernel or something then reinstall all the audio software. At least I would determine if it is the soundcard or sound card drivers or not.
Sounds like you have an interesting problem with your nvidia sound device. Is the audio device a part of a graphic card by any chance? Sounds like the graphics is interferring with the audio device, though on puredyne the rtirq-init script together with the realtime kernel should give higher priority to the audio device even when sharing irq with other devices, if this is the source of your problem. Perhaps Alsa would be interested to know of this problem? You could subscribe to their mail list and let them know.
Here are the Creative Labs devices supported by Alsa: http://www.alsa-project.org/main/index.php/Matrix:Vendor-Creative_Labs
On 03/25/2011 07:42 PM, Billy Stiltner wrote:
Thanks Bernardo
Charles Possibly so.
I wonder if I install the SB live if it will be autodetected. and the drivers are allready a part of the kernel or will I have to patch the kernel or something then reinstall all the audio software. At least I would determine if it is the soundcard or sound card drivers or not.
Pd-list@iem.at mailing list UNSUBSCRIBE and account-management -> http://lists.puredata.info/listinfo/pd-list
On 3/26/11, ailo ailo.at@gmail.com wrote:
Sounds like you have an interesting problem with your nvidia sound device. Is the audio device a part of a graphic card by any chance? Sounds like the graphics is interferring with the audio device, though on puredyne the rtirq-init script together with the realtime kernel should give higher priority to the audio device even when sharing irq with other devices, if this is the source of your problem. Perhaps Alsa would be interested to know of this problem? You could subscribe to their mail list and let them know.
Here are the Creative Labs devices supported by Alsa: http://www.alsa-project.org/main/index.php/Matrix:Vendor-Creative_Labs
Thanks ailo for the link. I'm not sure if the audio card is seperate from the video and ethernet chip or not but they are all built into the motherboard. I'm going to look into the machine later. I could not find any info about the sound device from the computer manufacturere. Windows just lists the device as HD audio.
Yes I will check the alsa forums. I thought someone else on this list might have encountered the problem before. It is odd that only interaction with pd's gui objects causes the glitching and interaction with other windows does not. I still haven't tracked down the problem.
I think the sound device has it's own irq and is not shared with anything.
I'll have to check the device confiuration in vista and somehow match it on linux. I noticed in the pd config file there are settings for which audio device to use.
On another note I/O latency measured with MAX on Vista is 30 ms. I guess that is due to the 256 byte framebuffer and ASIO matching it's buffer to max's. I'm sure it is less with pd. It sure doesn't sound like 30ms though there is a definite delay when playing with the example fm synth.
perhaps a rewriting of tcl/tks gui rendering to use hardware acceleration would help.
If the device is not sharing irq with another device, then there should be no problems in that regard. Haven't ever heard of your symptoms.
On Linux, if you use jack, PD's ms setting won't make any difference on audio, only midi. I wouldn't use any lower setting on PD than the lowest you would use with jack. To avoid glitches with jack, try using a high frames/period. Also, make sure you are realtime capable, which should already be prepared on Puredyne. Could be someone already had you're problem and discussed it on another mail list. Alsa would be my strongest bet.
On 03/27/2011 12:19 AM, Billy Stiltner wrote:
On 3/26/11, ailo ailo.at@gmail.com wrote:
Sounds like you have an interesting problem with your nvidia sound device. Is the audio device a part of a graphic card by any chance? Sounds like the graphics is interferring with the audio device, though on puredyne the rtirq-init script together with the realtime kernel should give higher priority to the audio device even when sharing irq with other devices, if this is the source of your problem. Perhaps Alsa would be interested to know of this problem? You could subscribe to their mail list and let them know.
Here are the Creative Labs devices supported by Alsa: http://www.alsa-project.org/main/index.php/Matrix:Vendor-Creative_Labs
Thanks ailo for the link. I'm not sure if the audio card is seperate from the video and ethernet chip or not but they are all built into the motherboard. I'm going to look into the machine later. I could not find any info about the sound device from the computer manufacturere. Windows just lists the device as HD audio.
Yes I will check the alsa forums. I thought someone else on this list might have encountered the problem before. It is odd that only interaction with pd's gui objects causes the glitching and interaction with other windows does not. I still haven't tracked down the problem.
I think the sound device has it's own irq and is not shared with anything.
I'll have to check the device confiuration in vista and somehow match it on linux. I noticed in the pd config file there are settings for which audio device to use.
On another note I/O latency measured with MAX on Vista is 30 ms. I guess that is due to the 256 byte framebuffer and ASIO matching it's buffer to max's. I'm sure it is less with pd. It sure doesn't sound like 30ms though there is a definite delay when playing with the example fm synth.
perhaps a rewriting of tcl/tks gui rendering to use hardware acceleration would help.
Pd-list@iem.at mailing list UNSUBSCRIBE and account-management -> http://lists.puredata.info/listinfo/pd-list
I found this to test if it is indeed a tcl/tk problem
l2ork http://l2ork.music.vt.edu/main/?page_id=56
I was looking into doing the same thing myself and found juce http://www.rawmaterialsoftware.com/juce.php
I don't see what the problem with c++ is.
-
ailo -
Bernardo Barros -
Billy Stiltner -
Charles Goyard