DP 9.1 is here

[dbender]

I'm just loving these favorable reports. Thanks guys.

So far, none have mentioned any problems with plug-ins.

is anybody having any problems with 9.1 and plug-ins, and if so, which plug-ins and what are the problems?

Everybody, have a great weekend!

[Robert Randolph]

Can you explain in more detail about what is actually producing the sound being measured and how it is being measured? The DP Click? Does that mean the DP metronome click? How do you generate the "click" in the other applications? "Scope Tap?" -- what is that? What is a "DSO"?

It seems to me that in order to measure something that mirrors actual virtual instruments as we use them, you need MIDI and a sample playback instrument involved (e.g., Kontakt). Latency as it relates to buffers and virtual instruments is about the delay between when I play a note and when I hear it.

Is there a way you can do a comparison test using actual music, musical notes and an instrument, piano or drums or something?

Ok, so let me take some time and flesh this out for you. I understand this can be difficult to understand from a 'musician point of view'.

Pre-req

First we have to understand that modern software moves audio around in what we call buffers. These are chunks of samples, and a reference to that chunk of data's location in memory is passed around the software. We'll use 128 samples for this discussion.

So this whole time our audio is MOVING. 44,100 samples (in this example) need to come out of the speakers per second. If we want to do something with that 128 sample buffer, we have about ~2.9ms (128 / 44100) * 1000) before we need to send the 129-256th sample buffer.

We need to both fill this buffer (take it off disc, generate sound, whatever), and empty the buffer (send it to the audio device for DA). So there's going to be 2 full buffers worth of time that need to expire before we can hear what happened.

128 samples filled. That takes 2.9ms
128 samples dumped. That takes 2.9ms.

So right now ANYthing that happens will take ~5.8ms at minimum before we can hear it.

On top of that latency, our hardware also has to convert audio to digital, and digital to audio. Due to the processes necessary for that, the hardware will usually penalize us with from 2-10ms of extra latency that is fixed. Let's assume 4ms for now.

128 sample buffer, 44.1khz sample rate, 4ms hardware latency.

Test Prep

So to test what the latency of DP is, all we really need to do is send audio from in->out and measure the distance in time between them. That is called "Round Trip Latency", or RTL for short. This operation is commutative, meaning we can also test it by measuring out->in.

So the idea is to play a sound, capture that sound (a), have it run through the interface, back out and capture that sound (b). The time difference (delta) between A and B is our RTL.

There is no need to test this 'with real music' since we simply need to know the distance between those 2 signals. If the latency drifted, that would be really bad. You would hear glitches, wow, flutter and all sorts of nasty artifacts. All this digital audio stuff works because the latency is fixed during playback.

All we need to do is have a very clear signal. For this I rendered a click track. Just a very simple waveform with a clear initial transient.

The Real Test

So took my Digital Sampling Oscilliscope (DSO), which allows me to capture 2 signals and measure/compare them.

The routing is as follows:

Main 1->(a)->Input 8->DAW->Output 6->(b)

I place the DSO probe between Main 1 and Input 8. This lets me capture the signal coming out of the card. That is the reference signal.

Then the signal passes from Input 8 to the DAW then to the output. Since we have an Input and Output stage, that gives us our hardware latency. The DAW gives us our 2 buffers of latency. Then I put a DSO probe at the end there.

Now in the scope, I can overlay the signals and place 2 cursors: 1 at the beginning of (a) and 1 at the beginning of (b). Then I hit a button and it tells me the delta, or the distance between the signals.

In this case, the expected value would be ((2.9*2) + 4) = 9.8ms of latency. Any other value I see, I can subtract 4ms for the fixed hardware latency and determine how many buffers are being used. For instance DP9.02 would give us a value of 15.6ms of latency in this scenario. Subtract 4, giving us 11.6ms. That just happens to be 2.9 * 4... meaning it's using 4 total buffers rather than 2. That's exactly what happened in my testing.

No matter what the signals were (or how you created them! the latency is fixed remember!), the delta would be the same. The click was used simply to have a clear transient that I could easily place the (a) and (b) cursors on for an accurate delta measurement.

Other stuff

So for some signals, you will only get half the hardware latency. Such as when you use a MIDI keyboard. However! there is usb latency (up to a disgusting 6ms), MIDI latency, keyboard latency ...

The real point of this test is that it shows that DP 9.02 was underperforming. If you do the math, you can easily see that it was adding 1 extra buffer for input and output. DP 9.1 now performs as expected.

THIS IS ONLY FOR RTL. This has no bearing on efficiency, what size projects you can load or anything of the like. How much you can get done in that 128 samples worth of time is a different topic. That's what efficiency is.

Hopefully that helps you understand, if you have any questions please let me know.

[Shooshie]

Maybe someone can confirm this as a bug or not. When the transport is running, hitting the numeric 4 key used to go in reverse until you stopped holding it. Now it creates a little burp in the timing and the transport immediately continues forward. I used that fairly often in cueing and editing on the fly... in fact, the reverse and slow reverse keys are non functional even when the transport is stopped.

This has cropped up again today. Restarting DP did't fix it. It never happened in earlier version of DP and reassigning the key to something else doesn't help. The keyboard is reading correctly in the OS, DP 9.1 is simply loosing the ability to act on the key command.

Techlink to MOTU

That's not happening here. Yet.
Is it intermittent? Did it work for a while, then fail?
I just tested it, and it's working for now.

Shooshie

[MIDI Life Crisis]

Yeah. Startup is fine then it goes away. Restarting the Mac fixes so it may be an OS thing. Still, started w/9.1

[djwizprod]

Couple of issues today...

Checked out the OMF AAF import, same as it always was, no luck importing embedded OMFs or AAFs! However, if I import the OMF to PT10, export as AAF with associated Audio Files folder, works no problem every time. Really wish this would have been fixed by now!?

Bounced a bunch of mixes and stems today, everything went as expected, except during one bounce the progress window locked up and... DP 9.1 (not responding) force quit. After a restart relaunch I could not replicate it again, hopefully that ghost in the machine has moved out.

Other than that, I'm showing huge performance gains no matter what I'm doing, still really happy about that!

[Tidwells@aol.com]

Can anyone report if DP 9.1 is working well on Mavericks? Thanks!

Doug

[SMS]

Am I weird to be wondering why MOTU doesn't email me to let me know there's a new release? I've been a registered user since 1985 or so...

[toodamnhip]

I'm just loving these favorable reports. Thanks guys.

So far, none have mentioned any problems with plug-ins.

is anybody having any problems with 9.1 and toplug-ins, and if so, which plug-ins and what are the problems?

Everybody, have a great weekend!

Cannot snap shot automate correctly without manually typing in all parameters

[Shooshie]

Sample Modeling's "The Cello" is overloading the performance meter, even at 1024 buffer. Nothing else does that. Just The Cello. (Maybe all Sample Modeling instruments; I haven't tried the flutes yet.)

It's basically unusable. In 9.02, it ran beautifully at 128 buffer.

Shooshie

[SMS]

It's like putting your head at the keyboard of a 6" Steinway. Pretty natural.

A 6 inch Steinway? That reminds me of an old joke...

[Shooshie]

It's like putting your head at the keyboard of a 6" Steinway. Pretty natural.

A 6 inch Steinway? That reminds me of an old joke...

My, my... such a tiny piano would require a tiny person who plays it. That piano must have a 6" piano player.

Shooshie

[MIDI Life Crisis]

It's like putting your head at the keyboard of a 6" Steinway. Pretty natural.

A 6 inch Steinway? That reminds me of an old joke...

My, my... such a tiny piano would require a tiny person who plays it. That piano must have a 6" piano player.

Shooshie

...nah, I better not...

[stevenew]

@Robert Randolph,

Just wanted to add my thanks for your detailed explanation (buffers, latency etc) in your post above, really really useful in helping my tired old brain understand these things:-)

[Sean Kenny]

Can anyone report if DP 9.1 is working well on Mavericks? Thanks!

Doug

All good..... so far ..... here Doug. Been tracking and editing and mixing here for a couple of days and no problems yet. I'm on 10.9.5 on a trash can Mac Pro 8 core, 32GB Ram

[jim2b]

Hi folks,

Is anyone running 9.1 with Mountain Lion, and if so, what are your results?

Thanks,

Jim