Latency and I/O
 

Is there any advantage to using Avid I/O or Omni versus ones by Apogee, UA, MOTU etc. as far as latency goes?

Thanks.

Re: Latency and I/O
 

I was going to open a thread about latency myself.
Hope Michael don't mind me for hopping in his head...:-)

I have PT HD (12.3.1), with no Avid HW. I believe however that Taking advantage of Avid HW would give me much better (2ms?), latency. I use non-Avid, AES connected converters, through RME interfaces. How could I move to a AVID interface like HDN and maintaining my converters and have about 2ms latency?

Help appreciated

Re: Latency and I/O
 

To use non AVID converters with either HDX or HDN, your converters need to have a digilink connection. Don´t know which converters you have at the moment running through AES from the RME, but if they don´t have a slot for a digilink option card, you´re kind of stuck.
A workaround would be to buy a digital only interface that has both digilink and AES, so that you can keep using you´re present converters. Basically, it would work by connecting the digilink cable from either HDN or HDX card to the converter and then from it´s AES I/O to your existing converters. But it´s a costly option.
I can suggest 4 converters that can do just that (digital conversion between AES and digilink) but also offer optional analogue I/O and AoIP (Ravenna or Dante) on top of that, in case you opt to for an integrated solution and decide to give up your present converters for the sake of avoiding redundancy and save some money:

http://merging.com/products/networked-audio (Horus and Hapi)

http://www.digitalaudio.dk/page1613.aspx (digital only, no analogue option)

http://www.digitalaudio.dk/page1503.aspx (DAD AX32)

Hope it helps.

Re: Latency and I/O
 

With Mergings Horus and Hapi, you can choose to emulate the latency of any of the protools interfaces. The lowest emulated latency is achieved when emulating the HD Madi (6 samples in and 5 out in 1FS and 2FS)
Lowest latency is when you don't emulate an interface, you then get 2 samples in and 2 out at all samplerates.
Analog/Digital modules latencies are already included in these figures.

Re: Latency and I/O
 

For the folks asking questions here, interested in conversion time, what IO buffer size and sample rate are you running at?

Re: Latency and I/O
 

Latency is a general term. In Audio recording we usually are concerned with what many call Round Trip Latency (RTL) which is defined as the time it takes audio to do a roundtrip from analog input --> Recording System --> analog output. To do this loop entails many individual "latencies":

A/D time,
transfer time from data bus to DAW, (TB, PCIe, FW, USB, Ethernet etc...)
DAW buffer in,
Any plug in delay
DAW buffer out,
transfer time back to data bus (TB, PCIe, FW, USB, Ethernet etc...)
D/A

This is the general case and there are many interfaces that intercept the incoming signal before it gets to the DAW buffer and feed that back to the performer for low latency recording.

So the whole concept of reducing latency is many faceted. The AVID converters are middle of the road (~0.47ms (A/D + D/A) at 96k). I wish all manufacturers would publish all the specs (they almost always tell you the 96k one cause its the fastest). From these (AFAIK) the MOTU, Apollo and Apogee converters are all fairly similar at 96kHz. Certainly within 0.25ms at 96kHz.

The Data bus transfer times are usually very significant. PCIe, TB and the DigiLink ones (AVID) are all near enough to zero to be ignored. FW and USB are huge in comparison (1 - 2 ms and at times much more).

The DAW buffer itself for in and out is a pretty big player: 64 samples X 2 /96kHz = 1.33ms.

So - the AvID I/O's may reduce your latency (assuming you have the right hardware to plug them into). In the Pro Tools 12 case there are only three options for using these these. HD Native or HDX for PT12HD. DiGiGrid DLS or (DLI + SoundGrid server) for PT12 Vanilla.

Re: Latency and I/O
 

These have excellent reputations and are some of the fastest but what you wrote above does not include A/D or D/A times.

From the Horus and HAPI spec page
Latencies

Input latency at 44.1/48kHz 16 samples
Input latency at 88.2/96kHz 16 samples
Input latency at 176.4/192kHz 13 samples
Output latency at 44.1/48kHz 9* samples
Output latency at 88.2/96kHz 9* samples
Output latency at 176.4/192kHz 9* samples

Some of the fastest converters there are (RME ADI8-QS is faster still)!
A/D + D/A at 44.1 = 25/44100 = 0.57ms
A/D + D/A at 96 = 25/96000 = 0.26ms

Re: Latency and I/O
 

While focusing on shaving down numbers, don't forget the old ~1ms/foot (at all sample rates :-)) for sound.

I suspect like Propower I do get concerned about systems with higher overall latency caused by all the things mentioned, and folks picking up on some tecnical single figure of merit that is only part of that and over worrying about beating that number down.

It can be interesting to pull out a microphone and another DAW (or even the one you are using) and measure actual total clap/click latency round trip times.

Re: Latency and I/O
 

We run at 48k with a 128 buffer on a TDM system. Musicians and singers have never had issues with that setting.

As we move more than likely to a Native HD system this year, I was wondering if there is any added advantage to staying with an Avid interface vs other manufacturers strictly on the latency issue and not sound quality.

Re: Latency and I/O
 

If you track with all TDM plugins on the record enabled channels (no native on these - elsewhere is fine) your latency is ~2.4ms today to the artist (no dependence on buffer). If you moved to HD Native and track at 128 with an AVID I/O you will have ~7.2ms. Other converters won't help much (as I already detailed above). The DAW buffer is already 256/48k = 5.33ms. Whether this is acceptable or not for you is all up to you.

I will add that at 44.1kHz the AVID I/O is 1.9ms A/D + D/A time. Many other I/Os are actually much faster at 44.1 and are more like 1ms (lynx Aurora comes to mind). This will certainly help. But if your buffer is truly 128 then 6ms - 7 ms RTL is not much difference to me (both pretty high).