Avid Pro Audio Community

Avid Pro Audio Community

How to Join & Post  •  Community Terms of Use  •  Help Us Help You

Knowledge Base Search  •  Community Search  •  Learn & Support


Avid Home Page

Go Back   Avid Pro Audio Community > Legacy Products > Pro Tools 12

Reply
 
Thread Tools Search this Thread Display Modes
  #1  
Old 05-18-2017, 10:58 AM
Marsdy Marsdy is offline
Member
 
Join Date: Oct 2009
Location: England
Posts: 1,707
Default PT's dual low latency buffer, 110 instances of Kontakt 'n stuff.

It seems that some of us, including me don't fully understand or are even aware of how Pro Tools' dual low latency buffer introduced with PT11 works. I'm wondering if people are being fooled into using lower buffers than their system or PT session can cope with because of this?

In another thread, ( http://duc.avid.com/showthread.php?t=390774 ,) there is mention of an Avid video showing 99 instances of Mini Grand running. It definitely rings a bell with me. In the same thread there was talk of how Pro Tools dual low latency buffer works. I couldn't find any reference to the 99 Mini Grand instances but I did find an old video by Avid explaining exactly how the dual low latency buffer works. Since this is OT in the original thread I've started this one.
The video is here: https://youtu.be/6JzwngQbgHE - The info is 2'19" in.

I got it completely the wrong way round in the original thread linked to above. It seems that tracks in playback mode are in a permanent high buffer state which the user has no control over. Only Record enabled or input monitor enabled tracks use the buffer the user sets in Playback Engine. This obviously explains why I often get BIG jumps in CPU usage when I record enable a Kontakt instrument, especially ones with high polyphony counts and complex scripting.

Getting back to the 99 instances of Mini Grand. I've just run a test and got 110 individual instances of Kontakt, (each using one instrument - the factory library Grand Piano.) With a buffer of 128 in PTHD 12.6.1 and OSX 10.12.3, CPU usage was reasonably stable hovering at the lower range of 48 to 60%. Each of the 110 instances maxed at 30 voices or 3300 voices of polyphony. It seemed stable too. (I got up to 130 instances but then started getting a few -9171 and -9173 errors.) The weird thing is, there wasn't much apparent difference in CPU usage when a track was recorded to and I went nuts bashing away on a MIDI keyboard. Maybe the instrument in question wasn't CPU demanding enough but I was running a heap of instances.

(BTW, I tried the exact same thing in Logic. Again, 110 instances although Logic's CPU meter looked to be running higher, (more stable admittedly.) OSX's Activity Monitor also showed surprisingly similar results. Both Pro Tools and Logic used very similar amounts of RAM - around 11GB.)

SO... In theory and according to Avid's video, playback is always at a high buffer setting internally preset by Pro Tools. It's only the record enabled/input monitor enabled tracks that are affected by the user buffer setting in Playback Engine. The thing is, to me it often doesn't feel like that is actually what is happening. I sometimes have to increase the buffer when VI and VE Pro polyphony and CPU usage gets above a certain level to stop pops and clicks. Why do I need to do that if PT is always running at a high buffer during playback?

As an aside - I felt that with PT11, given the same buffer settings as pre-AAX versions of PT, latency was better at higher buffer settings. Personally, I find a buffer of 256 or even 512 is quite playable for most VI's (not Maschine!) I didn't in previous versions. Overall system delay is usually more of a concern for me. (I can't get system delay as low with HDX than TDM. I'd like to be proved wrong but it seems to me the cause is that I have to run a lot more plug-ins natively because there's no HDX version.) ...But I digress.
__________________
Dave Marsden
UK
Avid HDX1, PT12 HD, HD I/O 16x16, Avid S6 M10 24-5, 2013 12 Core Mac Pro. Robby The Robot meter bridge ornament.
Reply With Quote
  #2  
Old 05-18-2017, 02:29 PM
john1192 john1192 is offline
Member
 
Join Date: Nov 2001
Location: Woodland Hills
Posts: 5,694
Default Re: PT's dual low latency buffer, 110 instances of Kontakt 'n stuff.

Quote:
Originally Posted by Marsdy View Post
It seems that some of us, including me don't fully understand or are even aware of how Pro Tools' dual low latency buffer introduced with PT11 works. I'm wondering if people are being fooled into using lower buffers than their system or PT session can cope with because of this?

In another thread, ( http://duc.avid.com/showthread.php?t=390774 ,) there is mention of an Avid video showing 99 instances of Mini Grand running. It definitely rings a bell with me. In the same thread there was talk of how Pro Tools dual low latency buffer works. I couldn't find any reference to the 99 Mini Grand instances but I did find an old video by Avid explaining exactly how the dual low latency buffer works. Since this is OT in the original thread I've started this one.
The video is here: https://youtu.be/6JzwngQbgHE - The info is 2'19" in.

I got it completely the wrong way round in the original thread linked to above. It seems that tracks in playback mode are in a permanent high buffer state which the user has no control over. Only Record enabled or input monitor enabled tracks use the buffer the user sets in Playback Engine. This obviously explains why I often get BIG jumps in CPU usage when I record enable a Kontakt instrument, especially ones with high polyphony counts and complex scripting.

Getting back to the 99 instances of Mini Grand. I've just run a test and got 110 individual instances of Kontakt, (each using one instrument - the factory library Grand Piano.) With a buffer of 128 in PTHD 12.6.1 and OSX 10.12.3, CPU usage was reasonably stable hovering at the lower range of 48 to 60%. Each of the 110 instances maxed at 30 voices or 3300 voices of polyphony. It seemed stable too. (I got up to 130 instances but then started getting a few -9171 and -9173 errors.) The weird thing is, there wasn't much apparent difference in CPU usage when a track was recorded to and I went nuts bashing away on a MIDI keyboard. Maybe the instrument in question wasn't CPU demanding enough but I was running a heap of instances.

(BTW, I tried the exact same thing in Logic. Again, 110 instances although Logic's CPU meter looked to be running higher, (more stable admittedly.) OSX's Activity Monitor also showed surprisingly similar results. Both Pro Tools and Logic used very similar amounts of RAM - around 11GB.)

SO... In theory and according to Avid's video, playback is always at a high buffer setting internally preset by Pro Tools. It's only the record enabled/input monitor enabled tracks that are affected by the user buffer setting in Playback Engine. The thing is, to me it often doesn't feel like that is actually what is happening. I sometimes have to increase the buffer when VI and VE Pro polyphony and CPU usage gets above a certain level to stop pops and clicks. Why do I need to do that if PT is always running at a high buffer during playback?

As an aside - I felt that with PT11, given the same buffer settings as pre-AAX versions of PT, latency was better at higher buffer settings. Personally, I find a buffer of 256 or even 512 is quite playable for most VI's (not Maschine!) I didn't in previous versions. Overall system delay is usually more of a concern for me. (I can't get system delay as low with HDX than TDM. I'd like to be proved wrong but it seems to me the cause is that I have to run a lot more plug-ins natively because there's no HDX version.) ...But I digress.
"Why do I need to do that if PT is always running at a high buffer during playback?"

exactly ... no tracks in record but a low BUffer results in Pops and Clicks .. Why ??? i thought it was only when you had tracks in record or input monitoring .. ???
__________________
OS 10.12.6 for PT 11.3.2 & 12.4 & above
Mac Pro 6-core 3.33ghz (2012) 24gb ram 256gb SSD
Macbook Pro july 2014 2.5ghz 16gb ram 512gb SSD - OS10.12.6 - PT 12.4 & above
lynx AES16e-SRC & Aurora 16
Artist Transport & iPad - Pro |Control - EuCon 3.7 (older Euphonix Branded MC Transport)
Reply With Quote
  #3  
Old 05-18-2017, 04:20 PM
Carl Lie Carl Lie is offline
Member
 
Join Date: Nov 2010
Location: New York City
Posts: 1,034
Default Re: PT's dual low latency buffer, 110 instances of Kontakt 'n stuff.

Dave:

Thanks for the follow through! You beat me to it. I'm going to run some tests this weekend.

I suppose then that once other processes are brought in (plugins, elastic audio on audio tracks) that number comes down substantially. I'd like to then try to quantify those effects on native processing.

Agreed that it doesn't feel like we have this power in practice.

You rock!!

Carl
__________________
PT 2018.7 HDX OMNI+HDX8x16 -Mac Pro 12 Cores 48 GB RAM OS 10.12.5 - API 16 channel - AMS Neve 16 channel, AMS-Neve-SSL-GML -Pres/Processing, Bock Audio, BLUE Bottle, Neumann, Josephson -Mics, Bogner, Kemper- Guitar, Chandler Zener-Curve Bender Outboard/Master. UAudio, Waves, Plugin Alliance etc.
Reply With Quote
  #4  
Old 05-18-2017, 04:49 PM
jjnssn jjnssn is offline
Member
 
Join Date: May 2011
Location: Minneapolis
Posts: 527
Default PT's dual low latency buffer, 110 instances of Kontakt 'n stuff.

Quote:
Originally Posted by john1192 View Post
"Why do I need to do that if PT is always running at a high buffer during playback?"



exactly ... no tracks in record but a low BUffer results in Pops and Clicks .. Why ??? i thought it was only when you had tracks in record or input monitoring .. ???


This my question as well. If no tracks are record enabled why would changing buffer size lower change the CPU usage? Other than the CPU usage is taking into account something other than the plugins.


Sent from my iPhone using Tapatalk Pro
__________________
____________________________________________

W10 Pro / X99 / E5-1650v3 / 64GB ECC / nVidia P1000 / ASUS TB3 AIC
PT HDX1 / OMNI / HD IO 16x16 / 11R / Kontakt Kontrol S49 / Artist Transport and Mix / UAD OCTO Satellite, Apollo Twin MKII &, x8P Thunderbolt

[Late 2013 10.14 MacBook Pro]
Reply With Quote
  #5  
Old 05-18-2017, 10:05 PM
clausiii's Avatar
clausiii clausiii is offline
Member
 
Join Date: Jul 2001
Location: Vienna, Austria
Posts: 1,800
Default Re: PT's dual low latency buffer, 110 instances of Kontakt 'n stuff.

That's something that i want to know too!
__________________
My ears! My ears!
clausiii
-----------------
MacBook Pro i7 Quad 2.6GHz 16GB RAM
ProTools 2019.5
Digidesign Command|8
macOS Mojave
MOTU 4pre
MOTU Track16
Axiom Pro 25
LinnStrument 128
Roli Seaboard Block

tons of stomp boxes
tube amps
guitars
noise
Reply With Quote
  #6  
Old 05-18-2017, 10:25 PM
jesele jesele is offline
Member
 
Join Date: Oct 2011
Location: Sweden
Posts: 183
Default Re: PT's dual low latency buffer, 110 instances of Kontakt 'n stuff.

Me too.

Jesper
__________________
Mac Pro 5,1 8-core, 24 GB, OS X.12.6, HDX, HD i/o 8x8x8 + HD i/o 16 digital + Metric Halo ULN 8 + RME Fireface 800
http://vicc.se
http://jesperelen.com
Reply With Quote
  #7  
Old 05-18-2017, 11:29 PM
Jon_Atkinson Jon_Atkinson is offline
Member
 
Join Date: Apr 2005
Location: London
Posts: 830
Default Re: PT's dual low latency buffer, 110 instances of Kontakt 'n stuff.

Is it not the case that this super-low buffer thing only applies when all MIDI is kept within PT...?
I don't think it works when the VI is held within VEPro.

I'm out of the studio so can't easily check, but I think that's correct




Sent from my iPhone using Tapatalk
__________________
Mac Westmere 6 core 3.3Ghz, 24GB RAM, OS 10.13.6, Protools 2018.7 HD, Antelope Orion 32, Vienna Ensemble Pro 5, UAD Quad PCIe
AJA I/O Express
Reply With Quote
  #8  
Old 05-19-2017, 12:45 AM
Marsdy Marsdy is offline
Member
 
Join Date: Oct 2009
Location: England
Posts: 1,707
Default Re: PT's dual low latency buffer, 110 instances of Kontakt 'n stuff.

Quote:
Originally Posted by Carl Lie View Post
Dave:

Thanks for the follow through! You beat me to it. I'm going to run some tests this weekend.

I suppose then that once other processes are brought in (plugins, elastic audio on audio tracks) that number comes down substantially. I'd like to then try to quantify those effects on native processing.

Agreed that it doesn't feel like we have this power in practice.

You rock!!

Carl
Thanks Carl!

I'd be very interested to hear what you come with on your tests!
__________________
Dave Marsden
UK
Avid HDX1, PT12 HD, HD I/O 16x16, Avid S6 M10 24-5, 2013 12 Core Mac Pro. Robby The Robot meter bridge ornament.
Reply With Quote
  #9  
Old 05-19-2017, 12:58 AM
Marsdy Marsdy is offline
Member
 
Join Date: Oct 2009
Location: England
Posts: 1,707
Default Re: PT's dual low latency buffer, 110 instances of Kontakt 'n stuff.

Quote:
Originally Posted by Jon_Atkinson View Post
Is it not the case that this super-low buffer thing only applies when all MIDI is kept within PT...?
I don't think it works when the VI is held within VEPro.

I'm out of the studio so can't easily check, but I think that's correct

Sent from my iPhone using Tapatalk
Hey Jon

I'm pretty sure VE Pro and the VI's it hosts are treated just like any other VI in Pro Tools. I'm sure I'm getting the same apparent latency when playing the same instruments in Pro Tools as when it's hosted in VE Pro. I need to do some testing myself though!!!

As I said, it seems the super-low buffer is what you set in Pro Tools' Playback Engine. I've got it the wrong way round in the past. I always assumed that the Playback Engine buffer setting affected playback and that Pro Tools set a magical super-low buffer in the background. It's the other way round according to the Avid video! But given that the Playback Engine buffer actually does affect playback... weird!
__________________
Dave Marsden
UK
Avid HDX1, PT12 HD, HD I/O 16x16, Avid S6 M10 24-5, 2013 12 Core Mac Pro. Robby The Robot meter bridge ornament.
Reply With Quote
  #10  
Old 05-19-2017, 01:04 AM
Sardi Sardi is online now
Member
 
Join Date: Aug 2002
Location: Perth
Posts: 1,631
Default Re: PT's dual low latency buffer, 110 instances of Kontakt 'n stuff.

IIRC correctly, there's 2 stages of buffer lengths dependent on what you set it too.

So for example, if you set it to 64 it has a playback buffer of 1024. If you set it to 512 it has a playback buffer of 2048.

I'm making those numbers up, but I recall reading something along those lines when 11 was launched.


Sent from my iPhone using Tapatalk
Reply With Quote
Reply

Thread Tools Search this Thread
Search this Thread:

Advanced Search
Display Modes

Posting Rules
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is On
HTML code is Off

Forum Jump

Similar Threads
Thread Thread Starter Forum Replies Last Post
How many instances of Kontakt can you run in the new pro tools HD 11? Meccanizm Pro Tools 11 5 04-10-2013 11:42 AM
Anyone using Kontakt 3 or NI Stuff With PT8 And Getting By? epu 003, Mbox 2, Digi 002, original Mbox, Digi 001 (Win) 16 12-17-2009 01:53 PM
This buffer stuff is making me mad. meltdown5 003, Mbox 2, Digi 002, original Mbox, Digi 001 (Win) 9 07-25-2006 10:45 PM
Buffer Issues and other stuff sainsomnia 003, Mbox 2, Digi 002, original Mbox, Digi 001 (Mac) 0 01-19-2006 12:44 PM
New section for Dual monitor stuff Ray Fabi Pro Tools TDM Systems (Mac) 1 04-18-2001 10:41 AM


All times are GMT -7. The time now is 03:54 AM.


Powered by: vBulletin, Copyright ©2000 - 2008, Jelsoft Enterprises Limited. Forum Hosted By: URLJet.com