HD Native Card Input Latency

[Siegfried Meier]

Hey guys,

Can I get some confirmation on how the native cards react to input latency? I've read that the playback latency is not affected by the buffer settings in 11HD and 12HD. I was messing with the card, and I had to have the buffer pretty low in order to be able to track instruments comfortably. I was afraid that this would cause CPU dropout issues in larger sessions, and I assumed that the card would be no good for tracking.

If indeed the 2 latencies are independent of each other, that would actually change this scenario quite a bit...

I'd test it, but the card has been pulled from the system and is now at another location.

Any advice much appreciated.
Thanks!
Sig

[Darryl Ramm]

What exact interface are you talking about?

What instrument?

What buffer size are you using to get acceptable latency? if that is extremely low any ideas on why it needs to be do low? Can you improve setup, e.g. to avoid comb filtering effects e.g. don't combine local/live sound with software monitoring.

Are you talking about recording an audio or midi/instrument track?

Recording audio through plugins (what ones) or not? LLM enabled?




Sent from my iPhone using Tapatalk

[Siegfried Meier]

No plugins, 192 or Rosetta 800, 128 is the most comfortable, recording live guitar - standard issue stuff. I only tested it with a single track, and since it's an AVID card, there's no other monitoring happening other than inside the PT mixer.

We have 5 Accel TDM rigs here at the studio, I'm used to the workflow and latency that those provide me. I was, foolishly, under the assumption that a native card would also give me that sort of zero latency feel, despite what settings the buffers were at.

However, if I can keep the buffers low on input, and the output buffer is always set to high (as I've read on here), then indeed it could be a useable system.

[albee1952]

I don't think the card has any impact on latency. Latency is a combination of conversion(A>D and D>A), plus buffer setting(the amount of time you give the system to process audio). Having said that(and assuming no plugins adding latency), then a 128 buffer setting (96K session) should feel like nothing for most anyone. I work at 48K so I set buffer to 64 for all tracking and have yet to hear any mention of latency(unless I screw something up).

[Darryl Ramm]

Well the card's impact on latency is that it is not a Core/TDM or HDX card and therefore always forces data through an IO buffer. But yes the card per-se adds no significant overhead above the latency in the interface box. Then there is the software latency caused by the IO buffer(s) we are really talking about here.

Siegfried -- How the split buffers work in Pro Tools 11 and later, and limitations of that, causes folks confusion, there is useful stuff scattered in old threads here, with unfortunately also lots of confusion and noise (with posts by one forum troll in particular grrr..), here is a good thread to start on, and a link to the Avid video talking about this stuff is in that thread.

http://duc.avid.com/showthread.php?t=392221

Hope that helps.

Sent from my iPad using Tapatalk

[Siegfried Meier]

Thanks for the info Darryl - that video confirmed exactly what I had read, so it does indeed make it rather useful. I'm messing with it now, and the cpu adjusts whether you have a track recording or if it's just being played back in mixing. It's a different workflow than I'm used to from our TDM rig, but it's more than useable. I'm amazed that I can monitor through plugins with near zero latency in 32 or 64 sample buffer - not something I recall ever being able to do in HD10 and under in native mode with a 3rd party interface. The Apollo has it's place for sure, and I use that on another rig, but it's really nice being able to just use the Pro Tools mixer for all our monitoring during a tracking session.

Thanks for clearing this up man!
Sig

[TNM]

Originally Posted by Siegfried Meier

No plugins, 192 or Rosetta 800, 128 is the most comfortable, recording live guitar - standard issue stuff. I only tested it with a single track, and since it's an AVID card, there's no other monitoring happening other than inside the PT mixer.

We have 5 Accel TDM rigs here at the studio, I'm used to the workflow and latency that those provide me. I was, foolishly, under the assumption that a native card would also give me that sort of zero latency feel, despite what settings the buffers were at.

However, if I can keep the buffers low on input, and the output buffer is always set to high (as I've read on here), then indeed it could be a useable system.

output latency definitely has something to do with it, as what you hear is the combination of total adda delay, input and output latency, IF you are monitoring through the pro tools DAW.

Once you are in the daw, only the output latency determines your virtual instrument monitoring time, i.e real time when you hit a midi key to when you hear the sound.

If you have any outboard instruments going through the pro tools mixer, then it will be the total combination of delay as above.. in many cases way too high to feel natural.

From what I understand, at lower sample rates, PT HD native has a total i/o latency of 4.3 ish milliseconds even at 32 sample buffer, which is way too high..at 64 and 128, forget it. To me, the only acceptable figure is the 96k one of 1.9(?) i think it is, ms, and that's at 32 samples. I suppose 64 samples at 96 would be ok too.

But of course, this puts strain on any system. I have seen how my own recent macbook with up to date intel chip reacts when i put the latency to 64 (lowest available) and arm a track (core audio here).. 128 is where everything settles to comfortable.
But that's why I am happy to use the apollo for all input monitoring, totally bypassing PT mixer.

AFAIK there are solutions for HD native to fix this.. you can get an expensive Antelope audio HD interface which has a dsp mixer with tons of effects inc guitar amps from overloud, and never have to monitor through the DAW.. it then somehow, with a propriety driver, syncs sample accurate when recording through the HD native it is attached to via digilink.

I am thinking of some options myself, to lose the 32 input limit of the apollo and to upgrade to HD software.