Assistance with HDX DSP Offline Bounce Crash

[fsavell]

Hello HDX users

I have an active case [Case: 04589385] that has been ongoing for a few months, without any solutions as yet. I'm hoping to get some assistance from anyone running a HDX system, that can spare the time to help test a session in DSP mode to see if you have similar problems, or to highlight that it is exclusive to my system.

The simple session linked below will result in a Hard-Crash-Quit every time Offline Bounce is used in DSP mode (Example of crash video 1; video 2; video 3), assuming all the plugins are active. The issue is not limited to this session, but it is a useful test session as it crashes, for me, almost every time I use Offline Bounce. It has been trimmed down to just the minimum plugins/routing to induce the crash with a single Offline bounce attempt. The session has no audio in it.

Download Test Session (.ptx file for testing)

Plugins used in the session (can still cause crash without all plugins active):
- Fabfilter ProQ3
- Fabfilter Saturn 2
- Oeksound Soothe 2
- Plugin Alliance bx_townhouse Buss Compressor
- Cranesong Phoenix II
- Ozone 10 Advanced (Ozone 9 Advanced will also suffice)


Some notable points:

• As it causes a hard-crash-quit, no Crashlogs are created
• the Bounce is always completed and the Bounced File is created successfully in the bounce folder. Crash occurs upon completion (or cancellation) of bounce.
• When testing I will repeatedly run up to 5 - 10 Offline Bounces of a 1 second selection to ensure I catch the issue. As mentioned below removing/inactivating some plugins reduces the frequency of the crash, so ~10 attempts at bouncing will likely induce the crash if its going to happen
• This issue does occur in Hybrid mode, but is too random and occasional to test, so I'm focusing on the DSP-mode crash for now, as its easy to reproduce (perhaps 1 in every 5 - 20 bounces will result in the crash in Hybrid mode; demonstrated with the test session in this video)

I have tested this for a few months with Avid Support and these are most of the parameters that so far have been tested and the result:

- Tested multiple HDX cards and run Digitest (successfully) on all cards

- Changed macOS
(issue occurs still on Catalina 10.15.7; Monterey 12.6.1 & 12.3.1

- Changed PT version
(occurs on 2022.12 & 2022.9)

- Changed HW Buffer size
(still crashes on any buffer size)

- Provided log files, kTrace files and OS Sample Process files (from Activity Monitor), videos monitoring Memory Tab in Activity monitor without success or insight into the issue

- Removed each individual plugin from AAX Plugins folder (Removing removing individual plugins lessens the frequency of crashes in some cases. Removing multiple plugins prevents crash)

- Removed Saturn 2 plugins (crashes after ~ 5 -10 bounces)

- Inactivate 1 Soothe plugin (crashes)

- Removed all Soothe Plugins (crashes - demonstrated in video 3)

- Changed all Soothe plugins to eco/1x oversampling (crashes)

- Testing with only Soothe plugins (no crash)

- Removed Phoenix II / bx_townhouse plugin (still crashes after ~ 5 - 10 bounces)

- Removed all FF ProQ3 (crashes)

- Inactivate one single Ozone module plugin, either Master Rebalance or Low End Focus
(crashes)

- Tested with only iZotope Plugins active
(no crash)

- Tested with iZotope Ozone 9 plugins instead of v10
(no effect, still crashes)

- Changed order of plugins
(crashes)

- tested with all plugins Native, or all (possible) plugins as DSP
(crashes)

- Removed Stereo Aux or Master fader
(crashes but reduced frequency)

- Removed all unused Busses from I/O
(crashes)

- Removed all Busses and created new ones
(crashes)

- Route Audio Tracks to different Buss than 'Stereo' Aux / change 'Stereo' Aux input to "No Input"
(prevents crash)

IN SUMMARY:
I feel like my testing clears any individual 3rd party plugin from causing the issue and it appears to be a ProTools bug, related to the number/memory/CPU use of cumulative plugins in the signal chain, irrespective of brand. This conclusion I feel is highlighted by the test where the 'Stereo Aux' input is set to No Input.

I would love to hear anyone else's results or insights, if you have time to test the session, or any suggestions on how to further test the issue. It is quite frustrating as it essentially prevents me from using Offline Bounce, which is essential for printing music mixes/stems with my workflow for which I do use both DSP and Hybrid depending on the session and its sample rate / CPU intensity. Even in Hybrid mode it still occurs frequently enough to disrupt workflow.

Thanks for reading the essay! Hope to hear from some of you.

Forrester

[Darryl Ramm]

- tested with all plugins Native, or all (possible) plugins as DSP
(crashes)

Yes but offline bounce actually uses native processing for plugins. If you take HDX out of this picture entirely what happens? Use built in output or whatever it's called on your Mac as the playback engine. What happens? Yes you might still have some HDX problem triggered by it using plugins in DSP mode once the bounce finishes, but I think you've maybe skilled over an important early troubleshooting step here.

[fsavell]

Originally Posted by Darryl Ramm

Yes but offline bounce actually uses native processing for plugins. If you take HDX out of this picture entirely what happens? Use built in output or whatever it's called on your Mac as the playback engine. What happens? Yes you might still have some HDX problem triggered by it using plugins in DSP mode once the bounce finishes, but I think you've maybe skilled over an important early troubleshooting step here.

Thanks for your suggestion Darryl . I'm aware that PT converts DSP to Native when using Offline bounce. The point of the test you've highlighted is to show that it's not the conversion of DSP to Native that causes the problem, as the crash occurs whether all plug-ins are native or not.

The crash doesn't happen when using native/Mac outputs, FWIW. All the evidence points to HDX hardware/cards/DSP mixer. Was there anything else you can deduce from it working/not working on the Mac Outputs?

Forrester

[fsavell]

Originally Posted by navim92152

Indeed, offline bounce utilizes native processing for plugins. However, if we exclude HDX from the equation completely, what would occur? You can utilize the built-in output or any other suitable option on your Mac as the playback engine. What would be the outcome? While it's possible that you may still encounter some HDX-related issues when the bounce completes due to the utilization of plugins in DSP mode, I believe you may have overlooked a crucial initial troubleshooting step. Cortexi Supplement

Thanks navim92152. Please see my reply to Darryl, above. Tested on Native and no problems. It is a HDX issue.

If you have a HDX system and can test, please let me know so I can relay any information back to Avid support.

[Darryl Ramm]

Originally Posted by fsavell

Thanks navim92152. Please see my reply to Darryl, above. Tested on Native and no problems. It is a HDX issue.

If you have a HDX system and can test, please let me know so I can relay any information back to Avid support.

navim92152 is a spammer/troll/LLM bot picking up other posts and reposting them. If they've not been kicked off DUC already they need to be.

[martthie_08]

any insights or progress on this issue?

I'm on HDX and about every third offline bounce crashes ProTools.

It ends up taking me a long time to bounce mix stems because of this.

[fsavell]

Originally Posted by martthie_08

any insights or progress on this issue?

I'm on HDX and about every third offline bounce crashes ProTools.

It ends up taking me a long time to bounce mix stems because of this.

Hey Martthie

This was resolved for me with an update to 2023.6

Forrester

[martthie_08]

interesting, good to know!
I might have to reinstate next year, didn't think this would get resolved, currently still on 2022.12

[fsavell]

Originally Posted by martthie_08

interesting, good to know!
I might have to reinstate next year, didn't think this would get resolved, currently still on 2022.12

I don't know that the bug was 'fixed' by active investigation and action on Avid's part. I suspect perhaps the HD Driver may have been involved, or was causing the issue in 2022.12 and earlier versions. It is ironic that the 2023.6 driver caused so many issues but possibly is the fix for this issue. Only guessing here though.

Forrester

[martthie_08]

ok thanks, good to know the 2023.6 driver has many issues, I will look into those before upgrading!