Push 3 audio interface - Give us Direct monitoring - Roundtrip latency is bad

[TheEntireRobot]

Just adding to this thread, if I set up a new audio track in Push 3 standalone and turn on monitoring, the latency feels to be around 10ms, which is enough that I really feel it when playing. It's just low enough that I can deal with it (most of the time), but any higher, and it would feel unusable. Still, it would be nice to reduce the monitoring latency or allow direct monitoring, because the current monitoring latency makes for a pretty miserable playing experience.

[braduro]

Again, Dough, to honor what the author is illuminating, it's not a pragmatic question of whether of whether push's performance is good or bad: the latency is workable, until it isn't.
It's the fact that the actual best practices of direct monitoring is literally unavailable to push as an interface. This is remedied by direct monitoring from an ADAT device with direct monitoring. Some preamps have this, some small mixers have this, and many rack mounts have this, but not much of anything that you'd necessarily want to throw in the gig bag along with the push, unless you were doing a more robust on-site install.

[jlgrimes]

Some people are way more sensitive to latency issues than others.

For most people anything under 10ms is usually unnoticeable.

I can't tell you the last time I used direct monitoring. It's been several years as one Rapper was very sensitive to small amounts of latency. This guy had superb timing by the way.

But the other 90% of folks were fine.

I hope it is possible for them to implement.


That said maybe try 64 samples (if you haven't tried), that should reduce the latency even further.

You could even try doubling your sampling rate and going to 64. That should be extremely low by that point (but your CPU would probably be through the roof).

[jlgrimes]

There must be some type of background overhead causing the latency. This shouldn't be an issue in anymore in 2023. I think the push 3 is trying to do too much.

IMO latency has seemed to actually gotten worse over the years.



Focusrite Scarletts were pretty decent but I think after an upgrade they added some safety buffers which increase the latency.

I think namely cheap USB audio chipsets are to blame. Push 3 is probably using something similar.

Thunderbolt usually is much lower. RME USB audio is about the only USB interfaces that seems comparable to Thunderbolt but they are pretty pricey.

[braduro]

Yup!
I still use thunderbolt. Why not? Is a free port waiting to get in the action. And USB transfers in packets, albeit the performance must be getting better. Interesting article published by Ableton on how the M1 chip actually performs better under some lower buffer sizes (presumedly due to the idea that the efficiency cores at lower buffers weren’t going to come into play. Kind of like the Parkinson’s Law applied to processors)

But again, something that is contingent on latency isn’t going to behave itself and magically hover at 6ms for our convenience. And, yeah, I can hear it. It doesn’t sound like the performance someone executed.

But really, I appreciate why you’re defending the current architecture. And I’m sure everyone on here wants to be a sport about it. But you can’t invalidate the author’s point. Again. If you have an adat and monitor from the adat. Your monitor latency won’t accrue with your playback latency. That’s a solution for now