12+ versions performance

[Ivan Bergamo]

I am thinking about moving to the newest PT version, if there are improvements in the bouncing performance. I am rendering some huge 10 hours audiobook projects, with quite of processing going on. My off-line bouncing times are at 2x speed. I was wondering if there will be any improvement if I jump to the newest version. I know there have been some engine optimizations in the past, but these were related to playback smoothness only, if I recall correctly. Any clues on this? Thanks!

[JFreak]

Software upgrade will not give you performance boost, hardware upgrade will. But with hardware upgrade you will also need software upgrade.

[its2loud]

It also entirely depends on what type of computer you’re coming from and planning to go to.

If I were you, I ask some colleagues with newer computers to run some tests for you with one of your sessions.

It is true that a faster, newer computer will offline bounce faster but plugins are also a factor. What types of plugins and how many you have in your session will impact the speed of a bounce.

[Darryl Ramm]

Originally Posted by Ivan Bergamo

I am thinking about moving to the newest PT version, if there are improvements in the bouncing performance. I am rendering some huge 10 hours audiobook projects, with quite of processing going on. My off-line bouncing times are at 2x speed. I was wondering if there will be any improvement if I jump to the newest version. I know there have been some engine optimizations in the past, but these were related to playback smoothness only, if I recall correctly. Any clues on this? Thanks!

Nobody knows exactly what the bottleneck is in your offline bounce is so nobody can answer this question with confidence. These questions are unfortunately frequently a waste of time asking expecting some intelligent answer, so I'll give you a semi intelligent non-answer.

The off-line bounce performance is likely bottlenecked on plugin processing, but the details we cannot see matter, in pathological cases you might get increases in performance simply by updating the plugins doing the most processing... or maybe the plugins got worse and decreases performance. The off-line bounce performance may be related to the structure of signal flow in sessions, you may be able to reduce times by optimizing that signal flow, reducing the length of plugin chains, etc. Pro Tools did include significant improvements to AAX processing in a release in 2018 IIRC, I would not want to guess what change that might make to offline bounce performance for you.

For heavy plugin processing you may be able to make significant increases in off line bounce speed by changing plugin settings to reduce workload, or switch plugins to more effective/lower overhead alternatives. It's an audio book so I would be curios where all the processing power is being spent, is there heavy duty deessibg on every input? Heavy weight look ahead compression and limiting on seperate tracks etc? You might just find you have a key plugin that is fairly single threaded and prevents scaling.

Performance stats collected during a bounce might give some useful clues. Starting with CPU usage and looking st memory to confirm there is not a problem there.

You very much are likely to be able to reduce off-line bounce times by using modern workflows with track commit. Your profile says you are on 12.5, track commit was introduced in 12.4, are you using it? There were bug fixes in later versions... that may make sense to upgrade to later versions for. I forget all the history there.

We have no idea how good or bad your system setup is, if for example you have a large disk cache, and how effective that is at caching your session, or if one is even enabled at all. etc.

So instead of asking other folks to guess for you it is much more useful for you to try and see. You are on Windows 10 and presumably could install recent versions of Pro Tools, either Pro Tools Ultimate free trial, or who knows, maybe a month subscription of Pro Tools Studio... or maybe even free Intro might be enough for that test.Pay attention to the compatibility matrix, likely it may be safest to leave your current install alone and to do a clean new Windows 10 or 11 install on a spare partition for the testing. And worse case you end up with showing the software upgrades did not help but you have a clean OS install with the latest Pro Tools install and maybe you will also be enticed to play with some workflow or plugin changes, and/or you can clone that to a USB drive and then if you do want to test on other hardware you can clone that there, sort out any driver needs, and measure the performance. I sure think it makes sense to be on a recent version of Pro Tools and then try out workflow and plugin changes. ... and maybe there is a path here that makes sense and kind of keeps the overall effort in reasonable check.

[Ivan Bergamo]

Originally Posted by its2loud

It also entirely depends on what type of computer you’re coming from and planning to go to.

If I were you, I ask some colleagues with newer computers to run some tests for you with one of your sessions.

It is true that a faster, newer computer will offline bounce faster but plugins are also a factor. What types of plugins and how many you have in your season will impact the speed of a bounce.

Thanks for your response its2loud. I've been using Ryzen 9 5950x, with some older version of Windows 10, which is compatible with PT 12.6. Yes, I understand there are a lot of factors (including OS), and the question is hard. Anyway, I think I'll be the one to test it as soon as we get the Izotope ARA support. :)

Thanks again.

[Ivan Bergamo]

Originally Posted by Darryl Ramm

Nobody knows exactly what the bottleneck is in your offline bounce is so nobody can answer this question with confidence. These questions are unfortunately frequently a waste of time asking expecting some intelligent answer, so I'll give you a semi intelligent non-answer.

The off-line bounce performance is likely bottlenecked on plugin processing, but the details we cannot see matter, in pathological cases you might get increases in performance simply by updating the plugins doing the most processing... or maybe the plugins got worse and decreases performance. The off-line bounce performance may be related to the structure of signal flow in sessions, you may be able to reduce times by optimizing that signal flow, reducing the length of plugin chains, etc. Pro Tools did include significant improvements to AAX processing in a release in 2018 IIRC, I would not want to guess what change that might make to offline bounce performance for you.

For heavy plugin processing you may be able to make significant increases in off line bounce speed by changing plugin settings to reduce workload, or switch plugins to more effective/lower overhead alternatives. It's an audio book so I would be curios where all the processing power is being spent, is there heavy duty deessibg on every input? Heavy weight look ahead compression and limiting on seperate tracks etc? You might just find you have a key plugin that is fairly single threaded and prevents scaling.

You very much are likely to be able to reduce off-line bounce tines by using modern workflows with track commit. Your profile says you are on 12.5, track commit was introduced in 12.4, are you using it? There were bug fixes in later versions... that may make sense to upgrade to later versions for. I forget all the history there.

We have no idea how good or bad your system setup is, if for example you have a large disk cache, and how effective that is at caching your session, or if one is even enabled at all. etc.

So instead of asking other folks to guess for you it is much more useful for you to try and see. You are on Windows 10 and presumably could install recent versions of Pro Tools, either Pro Tools Ultimate free trial, or who knows, maybe a month subscription of Pro Tools Studio... or maybe even free Intro might be enough for that test.Pay attention to the compatibility matrix, likely it may be safest to leave your current install alone and to do a clean new Windows 10 or 11 install on a spare partition for the testing. And worse case you end up with showing the software upgrades did not help but you have a clean OS install with the latest Pro Tools install and maybe you will also be enticed to play with some workflow or plugin changes, and/or you can clone that to a USB drive and then if you do want to test on other hardware you can clone that there, sort out and driver needs, and measure the performance. I sure think it makes sense to be on a recent version of Pro Tools and then try out workflow and plugin changes.

Thank you so much Darryl for your exhaustive answer. Yes, I was thinking about that specific 2018 version which introduced some optimizations. And also in recent times the Hyper-threading became supported too. I absolutely agree, the best way is to test it, instead of guessing, and I'll do my best to do it by the end of the year.

Regarding the processing, there can be a lot of noise reduction and sometimes de-reverb, de-mouth click... Today it's all AI, deep learning and all the buzz words plugins which are all very CPU hungry. Committing is not the greatest option because there can be a lot of changes during the production, and it's very convenient to be able to tweak and also to be able to go back to edit during the mixing phase too. And the committing itself takes time too.

Anyway thanks for your input, I loved it, it kind of brought me back to reality. :)

[its2loud]

I’m not entirely sure what Izotope ARA is going to buy here with relation to offline bouncing. If anything, it’s going to increase the amount of time unless you first render the Izotope processes. Also, the ARA integration is only for Spectral Repair.

From my experience the factors that matter most with offline bounce impact are plugins that do a significant amount of signal analyzing as well as processing. Examples of these are Spiff and Soothe or similar. Any type of real time noise reduction, particularly Clarity VX or any Neural Engine plugin. Neutron, or any multi processing plugin.

What Darryl mentions about committing tracks first before offline bouncing will help, but remember that committing itself is really just an offline bounce so if you’re committing a track with any of these types of plugins your results will probably not be much different.

[Ivan Bergamo]

Originally Posted by JFreak

Software upgrade will not give you performance boost, hardware upgrade will. But with hardware upgrade you will also need software upgrade.

Makes sense in my book too. But I've seen some exceptions - that prove the rule, in the past.

[Ivan Bergamo]

Originally Posted by its2loud

I’m not entirely sure what Izotope ARA is going to buy here with relation to offline bouncing. If anything, it’s going to increase the amount of time unless you first render the Izotope processes. Also, the ARA integration is only for Spectral Repair.

From my experience the factors that matter most with offline bounce impact are plugins that do a significant amount of signal analyzing as well as processing. Examples of these are Spiff and Soothe or similar. Any type of real time noise reduction, particularly Clarity VX or any Neural Engine plugin. Neutron, or any multi processing plugin.

What Darryl mentions about committing tracks first before offline bouncing will help, but remember that committing itself is really just an offline bounce so if you’re committing a track with any of these types of plugins your results will probably not be much different.

Thanks. I didn't mean to relate ARA with Bouncing in any way, it was just my thought to wait for the feature and then to upgrade. I was hoping to get the RX transcript feature with ARA too. :)

[Darryl Ramm]

Originally Posted by its2loud

From my experience the factors that matter most with offline bounce impact are plugins that do a significant amount of signal analyzing as well as processing. Examples of these are Spiff and Soothe or similar. Any type of real time noise reduction, particularly Clarity VX or any Neural Engine plugin. Neutron, or any multi processing plugin.

What Darryl mentions about committing tracks first before offline bouncing will help, but remember that committing itself is really just an offline bounce so if you’re committing a track with any of these types of plugins your results will probably not be much different.

Committing is only going to makes sense if there is a workflow reason to do it. e.g. are there multiple finished earlier chapters in the session that you can literally commit while working on the later chapters... and benefit from that across multiple bounces. I had no idea of the practicality, sounds like not.

I'm not a great fan of all these heavy duty plugins (and get off my lawn ) , but they are maybe a reminder that software can always outpace what hardware can deliver.