I'm hitting a brick wall realizing how big and unpredictable API calls via M4L are.
Can we hope (massive) improvements in the future ?
Meanwhile, is this latency documented somewhere ?
Thanks.
Latency of API calls
-
- Posts: 6025
- Joined: Mon May 15, 2006 12:15 pm
Latency of API calls
MacBook Pro 13" Retina i7 2.8 GHz OS 10.13, L10.0.1, M4L.
MacStudio M1Max 32Go OS 12.3.1
MacStudio M1Max 32Go OS 12.3.1
Re: Latency of API calls
Could it be related to some specific objects you use to move the data? Like (udp)send/receive?
Events in Max are limited to one buffer size, can you use signals instead?
What kind of API calls are you doing in instance?
Events in Max are limited to one buffer size, can you use signals instead?
What kind of API calls are you doing in instance?
Ableton Forum Moderator
-
- Posts: 6025
- Joined: Mon May 15, 2006 12:15 pm
Re: Latency of API calls
I made a simple test :
a device that listens for a midi note. If it receives a note, it just fires a clip (no launch quant) that contains one note at the very beginning.
Then I placed a few notes in the triggering clip, recorded the result of the "called" clip in a third track and compared the two clips.
So, in my experience, firing a clip has a random latency, from like 5ms to 100ms.
a device that listens for a midi note. If it receives a note, it just fires a clip (no launch quant) that contains one note at the very beginning.
Then I placed a few notes in the triggering clip, recorded the result of the "called" clip in a third track and compared the two clips.
So, in my experience, firing a clip has a random latency, from like 5ms to 100ms.
MacBook Pro 13" Retina i7 2.8 GHz OS 10.13, L10.0.1, M4L.
MacStudio M1Max 32Go OS 12.3.1
MacStudio M1Max 32Go OS 12.3.1
Re: Latency of API calls
I see. Unfortunately it's not possible to get sample accurate here.
I myself spent a lot of time and questions around, years ago, to do something like what you're describing and the hard reality is that it's not possible to do this accurately... and as you've noticed, you don't just get inaccuracy but randomness too!
I convinced myself that this should be seen and accepted just like "the beauty of analog imperfection" right there in our digital world
Another report to support@ableton.com wouldn't hurt, though
If it can't be sample accurate, at least the latency should be fixed and predictable.
I myself spent a lot of time and questions around, years ago, to do something like what you're describing and the hard reality is that it's not possible to do this accurately... and as you've noticed, you don't just get inaccuracy but randomness too!
I convinced myself that this should be seen and accepted just like "the beauty of analog imperfection" right there in our digital world
Another report to support@ableton.com wouldn't hurt, though
If it can't be sample accurate, at least the latency should be fixed and predictable.
Ableton Forum Moderator
-
- Posts: 6025
- Joined: Mon May 15, 2006 12:15 pm
Re: Latency of API calls
Except that clocking in the analog domain is the tightest thing in the world.
MacBook Pro 13" Retina i7 2.8 GHz OS 10.13, L10.0.1, M4L.
MacStudio M1Max 32Go OS 12.3.1
MacStudio M1Max 32Go OS 12.3.1
Re: Latency of API calls
You're right!
That doesn't solve our problem, though...
That doesn't solve our problem, though...
Ableton Forum Moderator