Page 1 of 1

send/receive data between devices, what's the deal with latency ?

Posted: Mon Sep 05, 2022 11:46 am
by chapelier fou
Is there any official info about the possible latency when using send/receive between m4l devices ?

I used to think that it was not reliable, but my recent tests seem to show that there is no latency, so I'm a bit perplex.
(I'm talking about non-critical data like MIDI note, monome grid presses, and stuff like that).

Thanks !

Re: send/receive data between devices, what's the deal with latency ?

Posted: Mon Sep 05, 2022 12:46 pm
by [jur]
AFAIK it's: nothing's been done to make it reliable, so use at your own risk.
In practice though it can be totally appropriate and somehow reliable for non time-critical stuff.

Re: send/receive data between devices, what's the deal with latency ?

Posted: Mon Sep 05, 2022 2:00 pm
by chapelier fou
Thanks. Since I got your attention, would you check the bottom of this topic and share your thoughts about my testing method ?
https://cycling74.com/forums/latency-of ... redictable

Re: send/receive data between devices, what's the deal with latency ?

Posted: Fri Sep 09, 2022 5:27 pm
by iainduncan
I can confirm I get a fair bit of jitter/slop when using message sends and receives between devices and tracks. I recently had a device that was using a loop of a send to a separate track's device and back to the main one, and then outputting notes. I compared against a redesign that eliminated this send/receive loop, and my jitter went away. I was getting an average of 2-3ms jitter with occasional outliers of 6-9ms. So fine for UI activity, but not good for note onsets. (this was all in the scheduler thread btw, I am sure that this is not coming from accidentally using the low thread - unless Live is deferring to the low thread on any send/receive unbeknownst to me!)

It would be fantastic to get some improved way of sending Max messages between tracks and devices, but I can see that this could be a very tall order at the code level. In standalone Max it works fine, but that's quite a different scenario.