System usage jumps from 10% to 100%

[ddjembe mutombo]

I am having a recent issue where my PT 11 will be operating at 10% system usage upon opening a moderately busy session. The first time I play back everything operates smoothly, but as soon as I hit stop the system usage jumps to 100%. Does anyone know what causes this?

PT 11 is on a 2011 Mac Pro. Quad core. 8 gig ram. OSX Yosemite.

Thanks!

[ddjembe mutombo]

Anybody have an idea? I've used this pro tools set up for more than four years now and have recently started experiencing this issue. I'm not paying for avid support to have someone tell me it was a OS X update...

I've attached photos of what it looks like during initial playback and one of what it looks like as soon as I stop initial playback. I have to shut down the session for it to return to normal...

[mesaone]

What all plug-ins do you have in this session? If Master Meter, get rid of it. If SansAmp, put a dither plug-in above it.

What version of PT11? Make sure it's one that is qualified for Yosemite (PT 11.3.0 or newer)

And why is your quad core showing 16 threads?

[audiobob]

Originally Posted by mesaone

And why is your quad core showing 16 threads?

He must have 2 quad cores, which are 8 cores that show up at 16 cores with hyper threading. I have a 12 core that show's 24.

I have spikes at low buffer settings. The go away with a higher buffer. It still is a mystery to me. All I run are Avid compressor/limiter and DMG Audio EQuick and EQuality, and I still get random spikes at low buffers for low latency recoding. I get spikes with no plugs.

[mesaone]

Originally Posted by audiobob

He must have 2 quad cores, which are 8 cores that show up at 16 cores with hyper threading. I have a 12 core that show's 24.

That could be it, thanks. Dual quad-core Xeon processors (which is what I assume you're referring to) is still an 8-core machine (8 physical cores, 16 threads) so I was thrown off by "quad" in his specs.

Let's see if the answer is as simple as a plug-in not playing nice.

[audiobob]

Originally Posted by mesaone

Let's see if the answer is as simple as a plug-in not playing nice.

I wish it were that for me....I'd dump that offender.

[Darryl Ramm]

With these symptoms (CPU burn when you hit stop) be especially careful of misbehaved plugins especially those with denormalization problems, that Sansamp issue as already mentioned, but it is not just limited to Sansamp. You need to do some careful debugging, including disabling plugins in batches, and if needed remove third party .aaxplugin files from the plugin folder (but denomralization issues are not limited to only third party plugins, with Sansamp as a clear example). If you suspect denormalization problems with a plugin then inserting a dither plugin in front of the offending plugin should cure the problem, but it may be quicker to identify suspect plugins by just disabling them.

[ddjembe mutombo]

Sorry, forgot to say it was a dual quad-core. PT 11.2.0

Ok, I found the plug-in. Sound Toys Decapitator. Is there a way to fix it to where I don't have to use dither every time?

Thanks, guys

[Darryl Ramm]

Originally Posted by ddjembe mutombo

Sorry, forgot to say it was a dual quad-core. PT 11.2.0

Ok, I found the plug-in. Sound Toys Decapitator. Is there a way to fix it to where I don't have to use dither every time?

Thanks, guys

Contact the plug-in vendor. It is an likely an inherent problem in how their code handles very small floating point values (i.e. it actually does nothing special to avoid what should be to a plugin developer the well understood "denormalization" problem). Nothing you as a user can do, dither or find a different plugin.

Are you on the latest version of their plugins? They may have fixed this, who knows. Quite often vendors, including Avid do a poor job of addressing these issues and even seeming to care at all about them, seem to just think if it's a problem users know how to deal with it.

Until then stick a dither plugin there and if the problem goes away you have a work-around and can be pretty confident that it was a denormalization problem...

[ddjembe mutombo]

Thanks, man. It's odd that this started this month. I don't remember updating anything in September, but that is when the problem happened. I'll use the dither until I hear anything from Sound Toys.