With NTP and no tricks, you can get 1-2ms of accuracy, if you buy a
external clock sync device, you can get 10 microseconds. I don't have
a specific software package in mind, its more a technique.
.hc
On Oct 10, 2009, at 5:03 AM, Brian FG Katz wrote:
Dear Mans-Christian,
This is an interesting idea. I first pose a question, to see if it is suitable for the problem as I understand it. First, you assume that
the CPU clocks of all machines are synchronized at a sub-sample level. Now,
if you send a message for an event to occur at t0+20msec, how precise is this timing executed? Can it execute the function in mid-buffer?Thanks for your help.
-Brian
Brian FG Katz, Ph.D Audio & Acoustique LIMSI-CNRS BP 133 F91403 Orsay France tel. (+33) 01 69 85 81 55 fax. (+33) 01.69.85.80.88 e-mail Brian.Katz@limsi.fr web_theme: http://www.limsi.fr/Scientifique/aa/thmsonesp/ web_group: http://www.limsi.fr/Scientifique/aa/
-----Message d'origine----- De : Hans-Christoph Steiner [mailto:hans@at.or.at] Envoyé : jeudi 8 octobre 2009 19:16 À : Brian FG Katz Cc : pd-list@iem.at Objet : Re: [PD] multi-pc synchrony
If you use some kind of time-tagging, which some implementations of OSC support, then you add add a bit of latency to the whole system and use that to reduce the inter-machine latency. The basic idea is that each message has a time tag that marks when that message should take effect. Then you put that time tag 20ms in the future when you send it, every machine should have it within 20ms, and then they'll all execute the message at the same time.
.hc
On Oct 8, 2009, at 6:27 AM, Brian FG Katz wrote:
Dear PD-ers,
We are working on an installation with 4 machines running PD to feed 157 loudspeakers. We are interested in reducing latency to a minimum between channels, and especially between machines. The inter-channel latency in less than a sample, so all is fine there. For inter-machine latency, we arrive at differences on the order of 10msec, close to our minimum audio- buffer length of 11msec. Any small audio-buffer and we get audio artifacts.
We are using a word clock synchronizer (Nanosyncs HD; Rosendahl), but I don't think that does latency synchronization.
My question, is there another means to improve inter-machine latency performance other than reducing the audio-buffer?
-Brian
Brian FG Katz, Ph.D Audio & Acoustique LIMSI-CNRS BP 133 F91403 Orsay France tel. (+33) 01 69 85 81 55 fax. (+33) 01.69.85.80.88 e-mail Brian.Katz@limsi.fr web_theme: http://www.limsi.fr/Scientifique/aa/thmsonesp/ web_group: http://www.limsi.fr/Scientifique/aa/
Pd-list@iem.at mailing list UNSUBSCRIBE and account-management ->
http://lists.puredata.info/listinfo/pd-list
Information wants to be free. -Stewart Brand
Looking at things from a more basic level, you can come up with a more
direct solution... It may sound small in theory, but it in practice,
it can change entire economies. - Amy Smith
-
Hans-Christoph Steiner