Hi, I have a problem with my sound card, which is a m-audio firewire audiophile. I can't manage reaching a low latency in pure data : When I use shareware like fruity loops or Live Lite, the shareware automatically detects the buffer size i have specified in the control panel of the sound card and tells me what latency it gets, with a 256k buffer size i have a 5 ms latency (which is good for me).But in pure data, using the M-Audio FW Asio driver, as soon as I try putting a latency less than 60 ms, the test tone becomes very dirty, with a lot of discontinuities. I have tried a lot of combinations with -audiobuf -blocksize... but i simply don't manage to do it. If you have some ideas i would be very glad. Bye.
P.S. : i have tried with pd 0.37, 0.38 and 0.39 on windows XP
Retrouvez tout, instantanément, sur votre PC, avec Windows Desktop Search ! http://desktop.msn.fr/
this is a known problem with pure data's portaudio interface. pd is not able to run with low latencies on osx/windoze ... using jack / alsa the situation is better.
the developer branch contains some changes in the scheduler that allows pd to run with low latencies with portaudio. i'd suggest you give this a try ... there are some binaries on sourceforge for windows ... no 0.39 binaries, though ...
cheers ... tim
thanks answering that fast Tim. I put the pd-0.38-4-devel-2 version on my computer and i managed reaching a very low latency. I'm now wondering about the stability of this pd version, do you think i can make live with this version? (i've already got some fatal errors...) And there is a thing that i don't understand : on my other computer i have a terratec MT88 with asio drivers that perfectly works with pd 0.37.3 (i have a latency of 5ms on this computer). Do you know why? And a last question : when do you think this work on the portaudio and asio driver that have been down in this "devel pd" will be integrated in common pure data releases. Again, thanks a lot for your help. Cheers.
Retrouvez tout, instantanément, sur votre PC, avec Windows Desktop Search ! http://desktop.msn.fr/
well, devel doesn't mean unstable ... there are parts that are more stable in miller's branch and there are parts that are more stable in devel... since devel has some radical changes in the scheduler i can't promise that it's bug-free, but personally, i trust devel more than i trust stable... if you can send me detailed descriptions about the errors that you experience, i can try to figure out, if they are devel bugs and if so try to fix them ...
i must admit, i haven't followed the pd development before 0.37, especially not on windoze ...
short answer: never
long answer: miller is very conservative in accepting changes to pd. and pd is designed from scratch as a single-threaded application. all modern audio apis (asio, jack, portaudio, possible also alsa and coreaudio) are running a callback at the time of the audio hardware interrupt to reduce latency. thus pd has to be changed from a single-threaded to a multi-threaded program, which is not exactly a minor change... thus i don't think that miller will ever accept these changes ...
cheers ... tim
Tim Blechmann wrote:
Can I infer from this statement that using JACK on OSX is a low-latency option?
d.
it will probably give better latencies on osx than portaudio ... in miller's branch, although i'm not sure, if it's possible to compile pd/jack on osx... (iirc hans has been trying this ... didn't really follow this, though)
in devel, it shouldn't be a big difference if you're running pd with the changed scheduler, that's computing dsp in the audio driver thread.
cheers ... tim