Virtual instrument latency -- not!

[DAWgEAR]

This is an odd question, but I am experimenting with various buffer sizes in a session with virtual instruments and I am finding that setting the buffer to 64 or 2048 (at 96k) seems to make no difference in terms of time from when I press a key on my controller keyboard to when I hear the sound triggered in the VI: it is nearly instantaneous regardless of the buffer. (This is good!)

I have tried recording midi to a click and there is no latency or timing problem in the recorded events.

I am not complaining about this! But I am wondering why this would be the case. My prior experiences with other DAWs is that high buffers tend to create so much latency as to render virtual instruments unplayable. What is going on here? Is there a setting or preference that would influence this? Is this specific to HD Native? Am I having a senior moment?

Any thoughts?

I would like to understand what is happening because, if it is some setting and I inadvertently change it, I want to be able to get back to what I am experiencing!

[nst7]

I think this is a well kept secret about HD Native, and I'm surprised Avid haven't focused on promoting this more.

I don't know exactly what goes on under the hood, but here is my theory:

We know that these new Avid HD interfaces actually convert faster, being a new design with a simplified path to the converters. They've promoted that fact heavily and is part of the reason for lower latency in general, even at the same buffer settings vs. other setups.

So for this reason, even audio recording is lower latency than it otherwise might be. But, with VI's, that is cut in half again because VI's are only using D to A conversion, not the full round trip of A/D and D/A that recording audio uses. So that helps a lot.

Also, it may be that the D/A portion converts even faster than the A/D portion (from what I understand about converter design, D/A is a simpler design).

And, since your sessions are at 96k, that cuts the latency down even more (vs. 44/48), for both audio and VI's.

So this makes perfect sense. I haven't really used 88/96k much yet, but even at 44k, I've noticed not much latency with VI's even at the 1024 buffer. In fact I've sometimes just messed around with a VI in a session, and forgot that I was at a 1024 buffer, because it just wasn't that noticeable.

And as mentioned, if you're at 96k, I'm pretty sure that cuts the latency in half yet again.


So this is my explanation. I'd be interested to know if there was anything else going on under the hood that we don't know about.


When PT11 comes out, with it's more efficient processing and handling of VI's and RAM, combined with a powerful computer, HD Native could be a killer rig for a composer.

[DAWgEAR]

nst7, as always, your reply is well thought out. Thanks for taking the time to write it.

That sounds like a reasonable explanation. If it is specific to the hardware, I too am surprised that Avid has not promoted it more.

I agree with your final comment. Between HD Native + Omni, Vienna Ensemble Pro 5, and PT 11, this will be a dream rig for someone like myself who does a lot of MIDI/VI work and records only one mono or stereo track (vocals, guitar, bass) at a time (sans inserts). I am getting near-zero latency monitoring analog inputs directly through the Omni Mixer and playing VIs is essentially latency-free ... and all this at 96k and at any buffer size.