As part of Exponential Audio testing programmes for their new Surround versions of PhoenixVerb and R2 that we announced exclusively on Podcast 86, they did a direct comparison between Pro Tools 10 and Pro Tools 11. Michael Carnes takes up the story….
Here’s what I measured: First, my machine is a 2009 Mac Pro, 8-core, 2.26 GHz. I’m running OSX 10.8.5. I tested with PT 10.3.7 and PT 11.0.2. I created sessions in which each track had a 1->5.1 plug-in or a 2->5.1 plug-in. I used a roughly equal mix of PhoenixVerb Surround and R2 Surround.
Pro Tools 10
I began with a buffer size of 512. I gave 85% of the CPU to PT in the Playback Engine setting. I believe this is fairly typical. I was able to create only 24 reverbs before I began to run out of CPU. When I increased the buffer size to 1024, I was able to create 42 reverbs. Nice improvement
Pro Tools 11
I got 72 reverbs with a buffer size of 512. There was very little change in performance when I raised the buffer size.
I brought up the Apple Activity Monitor so I could see the core activity. On this generation of Mac Pro (and anything later) there’s a capability called hyperthreading. This turns each physical core into a pair of virtual cores. So it will appear as if you have 16 cores instead of 8. Pro Tools 10 makes very little use of this capability. Out of each pair of virtual cores, only one of them is busy whilst the other one is nearly idle. In Pro Tools 11, both virtual cores appear equally active. In other words, PT 11 gives you more of the computer you paid for. This isn’t just a Mac Pro thing, any I7 processor (like the top of the line Mac Mini) also has hyperthreading too.
The buffer size doesn’t really affect my plug-ins. Whatever the size is, I break that down into my own fixed buffer size. My own processor needs will be roughly the same, whatever you set the buffer size to be. In PT10, we see that the buffer size makes a big difference in this test. In PT11, less so. I think this means that other parts of PT10 are affected by the buffer size.
Pro Tools 11 offers a feature in which PT can shut down plug-ins that aren’t doing anything. That’s not what’s going on in this test. I implemented my own form of dynamic processing in this version. It works perfectly well in PT10 or Logic or any other DAW.
For the most part, I think we’re seeing the difference between Pro Tools 10 and Pro Tools 11. I think this will mainly show up on computers with hyperthreading. But even on other machines (like an i5 Macbook), there will still be a difference.
As Michael says, a lot of the differences are down to the differences in Pro Tools 10 and 11. It is interesting to see that PT10 isn’t able to make good use of the hyperthreading and adds ammunition to the advice that with Pro Tools 10 you should set the number of Processors to one less than the actual number of processors, so in Michael’s example, where he has 8 cores, it should be set to 7 and not 15.
On Pro Tools 11, there is no setting for this, Avid have taken this option along with a number of others away and these are all set in the background by Pro Tools to make best use of the computer resources.