View Single Post
  #1  
Old 10-20-2020, 02:44 PM
Stephen333 Stephen333 is offline
Member
 
Join Date: Jun 2014
Location: Los Angeles, CA
Posts: 16
Default Pro Tools and 9093 Errors

Hey guys, this is my first post, but I've been using this community for a long time. I wanted to reintroduce a topic that I know has been covered at some length in this community, and provide a bit of backstory on my journey to fix these issues.

To start, let’s go over the computers I'm going to be bringing up.

The main rig here is a 2019 Mac Pro, 128gb of ram, 3.2GHz 16-Core Intel Xeon W processor, and 4TB SSD startup disc on 10.15.3 Catalina. This rig operates primarily as an HDX2 rig, connected to two Burl Motherships, although all tests have both been performed via HDX and with no HDX connected running exclusively native.

Pro Tools Ultimate version 2020.5
Basically every major plugin manufacturer is represented here, and certainly the entire product line of most of the big guys; UA, Waves, Fab-Filter, Plugin-Alliance, Izotope, McDsp, and of course Acustica.


The second engineer's rig is a 2018 Mac Mini, 32gb of ram, 3.2GHz 6-Core Intel i7 processor, and a 1TB SSD startup disc on 10.14.6 Mojave. This rig is native.

Pro Tools Ultimate version 2020.9
Basically every major plugin manufacturer is also represented here, and certainly the entire product line of most of the big guys; UA, Waves, Fab-Filter, Plugin-Alliance, Izotope, McDsp, and of course Acustica.



My issue is simple and is easily repeatable: on any session utilizing a larger amount of plugins there is a great risk of running into continuous 9093 (or 9173 if native) errors. Pro Tools system meters show the all-too-common peaking around the center cpu core, and meter performance is incredibly chaotic, jumping around indiscriminately. This most often happens in mix sessions with long plugin chains and lots of delay comp being used, but it also happens in basic tracking sessions with delay compensation barely being utilized. Acustica plugins send Pro Tools meters into disarray (plenty of forum posts about this), but these issues will happen with no Acustica plugins in a session at all as well.

I have elevated support tickets currently with both Avid and Apple, and Acustica promptly closed the support ticket we had with them a few months back and were largely not helpful in the matter, though I don't think they are solely to blame here, their plugins are just better at revealing some of the issues at play.

Both systems perform roughly the same with the sessions, despite the fact that CPU, RAM, etc... usage on both isn't even approaching the limits of the systems. On the Mac Mini we usually hit 30% usage before the errors can't be avoided, on the Mac Pro we often can't even get to anything past 15% of our available resources before a session is deemed unplayable by Pro Tools.

Here are some of the various fixes we have tried to no avail, if I am missing anything that can help, please let me know:
  • Trash Preferences (thanks Pete Gates)
  • Reset SMC/PRAM/NVRAM
  • Always operate on highest buffer
  • we dropped all of our mix sessions to 48kHz just to get a bit more plugin usage
  • Create fresh new admin account
  • Fresh installs all around
  • used CPU setter to disable hyperthreading
  • manually checked RAM as well as using Memtest86
  • checked all issues on both HDX and Native
  • checked for any clocking issues in setup
  • updated plugins/removed potentially outdated or hazardous plugins
  • spread plugins over multiple auxes in high count plugin chains
  • computer optimized by every optimization guide we could find and is used only for PT

I understand that there are a lot of standard workarounds for working in Pro Tools but the fact of the matter is that both of our systems in use, despite being dramatically different in available system resources, operate at roughly the same level.

Has anyone figured out a solution to this that doesn't involve moving to another DAW or utilizing Vienna or something similar? It feels like Pro Tools has a self-imposed ceiling on CPU usage that doesn't allow it to scale across the cores properly.

Any ideas?
Reply With Quote