i7 Macbook Pro - 11.0.2 / 11.0.3 constant -9173 errors

[A Chapman]

Hi all,

Just wondering if anyone has a similar setup to mine, and is experiencing similar problems :

Macbook Pro 17" 2.2ghz i7 (2011 - 8,1) 8GB RAM / 10.8.5

PT11.0.3 (but happened also on previous version).

Playback engine set to ignore errors and dynamic plugin processing on.

Got my audio on a LaCie D2 FW800 drive (but also happens if I put sessions on other drives).

The symptoms :

Repeated -9173 CPU Overload errors. Usually, after one of these errors, all playback is severely distorted, almost like everything is being ring modulated. If I go back to the Playback Engine and change the buffer size, the playback returns to normal until the next error.

Actual CPU usage is being measured by the System Usage window at around 40-50%, and there's no visible spike before the errors.

Using a lot of third party plugins, all of which are updated to the latest versions. I've also removed all unnecessary plugins from the plugins folder. Plugins I'm typically using :

Various Waves 9.2 (Rbass, Supertap, Mondomod, IR-L)
Izotope (Alloy 2 / Trash 2)
Soundtoys (Devil-Loc Deluxe, Little Microshift, Radiator)
IK Multimedia T-Racks compressors and EQ (LA2A/1176/Pultec) and Amplitube
Blue Cat Patchwork / MB-7 (hosting 64bit vsts, usually Valhalla Room)
PSP NobleQ
NI Guitar Rig
Abbey Road Brilliance Pack

However, these problems predate much of these plugins, as I had them with PT10 also.

Deleted preferences, repaired permissions, zapped PRAM, forced the use of the discrete graphics card only, followed all of the Avid optimisations.

I'm increasingly convinced that this chipset and/or this revision of the MBP has a problem of some kind when it comes to this kind of processing. This version of the MBP is the one which can draw more power than the PSU can deliver - which can sometimes result in severely drained battery after a long session. I know people who use them for video, and they can't do long renders on it, as usually the battery runs out before it's finished.

Would be really glad to hear from anyone else with a similar setup, particularly if you think you found a solution to this problem. Unfortunately I think the only solution in the long term for me is to ditch it and get a different machine.

[JFreak]

Have you tried with zero 3rd party plugins installed?

[A Chapman]

Hi JFreak - not on an actual session, no, as of course without the third party plugs my session would have drastically lower CPU usage. I have simulated high CPU usage with no 3rd party plugs installed, and get the same problem eventually - again, after about 40% CPU usage.

[YYR123]

And you said you had the same issues with PT10 too right?

[propower]

Does apple activity monitor match the system usage percentage?

What are the rest of the system details? Sample rate, buffer size, interface etc..

[A Chapman]

CPU overload errors under PT10, yup. Lots (and lots) of EXC SIGBUS crashes (RAM related - PT trying to address RAM that doesn't exist). Situation is improved in 11 as it's all 64bit, so no more RAM-related crashes, but still problematic.

Have had the logic board changed twice, and done full wipe/re-install. Situation was the same under 10.6 and 10.7.

[Darryl Ramm]

It sounds like you've done a lot of good stuff but as JFreak mentioned always suspect your plugins with CPU related glitches. Temporarily remove all the plugin files and try again. If that works then put them back in batches until you find the culprit(s).

Don't fixate on the CPU meters, the CPU glitches that can cause these 9173 errors are so transient that the meters just often don't help.

What sample size, sample rate are your sessions running at? What interface and driver version are you using? If the interface is Firewire how is it chained with the disk drive? And what exactly is the playback engine set to (esp. I want to make sure you are not using aggregate IO).

I would suspect/ try to exclude VIs as the problem first. Then any up-sampling/high latency plugins. But any old buggy plugin can cause problems.

I don't notice any VI plugs explicitly called out in your list, but if you are running any where are the VI samples installed (I'm curious if some of your VST plugs are VIs)? What is the make/model/rpm spec of the system drive and any other drive you have the samples on?

I have an older 17" MBP that has always been rock solid for me, and my impression is that people don't run into problems on more recent 17" MBP like yours. And that battery issue you noted is far from limited to you MBP.

[A Chapman]

Originally Posted by propower

Does apple activity monitor match the system usage percentage?

What are the rest of the system details? Sample rate, buffer size, interface etc..

Neither Activity Monitor nor System Usage indicate actual CPU overloads. Nothing in Activity monitor looks out of the ordinary, or particularly CPU hungry.

Only ever work at 44.1khz. Problems happen on every buffer size and interface. At the moment I'm using an MBox 2, but it happens on my Saffire Pro 40, built in output, MOTU 828, NI Audio2DJ etc etc.

[Darryl Ramm]

Originally Posted by A Chapman

CPU overload errors under PT10, yup. Lots (and lots) of EXC SIGBUS crashes (RAM related - PT trying to address RAM that doesn't exist). Situation is improved in 11 as it's all 64bit, so no more RAM-related crashes, but still problematic.

Have had the logic board changed twice, and done full wipe/re-install. Situation was the same under 10.6 and 10.7.

Hang on you are having SIGBUS application crashes as well? The leading cause of those is also bad plugins... and usually just coding errors in the plugins, not necessarily that you are running out of memory/address space. But old crashes on PT 10 may not be at all relevant to recoded AAX-64 plugins.

Do you had any recent SIGBUS crash on Pro Tools 11? Look in the System Console (Console.app) under User Diagnostic Reports for Pro Tools crashes and see how many are there and if they look similar (the stack trace of the thread that caused the crash looks similar, not just the fact they might be all "SIGBUS" crashes) and post some here. In the report header it will say which thread caused the crash, like "Crashed Thread: 23 blah blah" whatever that "23" number is your past here needs to include the stack trace for that thread and all the information above that. Just paste all the report if you can. If they look different post several.

[A Chapman]

Originally Posted by Darryl Ramm

Hang on you are having SIGBUS application crashes as well?

Only under Pro Tools 10, and to be honest I haven't seen one in a while as I've been using CPTK, with all the audio cached to RAM.

No SIGBUS crashes under 11 as yet.