Dropout consistently always at 4:17 - SOLVED

[chuckebaby]

32 minutes ago, gmp said:

  In my previous post was was incorrect - the problem was not solved and cropped up again. I finally did fix it and it has to do with having an Aux Track input echo turned off. If it's "on" there is no dropout  Here's my new post explaining this and warning others that may have this problem. 

 

 

I'll try the repro steps in a bit

Good job on listing out the steps. I hadn't figured that part out yet. Like @David Baay I couldn't reproduce it going by what you commented before.

[gmp]

Thank you Chuck E Baby for the link below to the similar problem involving Aux Tracks. This enabled me to fix my dropout problem

 

 

[Noel Borthwick]

@gmpcan you send me a simple project that has the issue? We'll look into it.

Delete all the unnecessary tracks and synths and just keep the tracks required to repro the problem.

[Matthew Sorrels]

I couldn't get it to actually stop the audio engine but I did notice a giant CPU spike on one core at around the 4:20 mark (starting from an empty project, adding an aux track and turning off echo and then pressing play, on the empty project).  That CPU spike was large enough if this project had any audio or real content it may have caused a total drop out, or if I had less cores/CPU power.  Some sort of garbage collection/clean up maybe?

[gmp]

6 hours ago, Noel Borthwick said:

@gmpcan you send me a simple project that has the issue? We'll look into it.

Delete all the unnecessary tracks and synths and just keep the tracks required to repro the problem.

Here's the bun, let me know how it does

https://onedrive.live.com/?id=C256E2DAE26B22A5!105&cid=C256E2DAE26B22A5

[David Baay]

I'm not understanding the routing at a glance. It wanted a bunch of hardware outputs assigned when I loaded it, and there was no way to know what was 'Master' bus. By default, the only thing going to my Main Out 1-2 ended up being the El Guitar track which outputs to a bus named A 1-2. But the project does drop out at 00:04:16:06 SMPTE, so it's not just your hardware setup.

I can't spend any more time right now, but will go over the routing, and see if I can replicate it with a simpler project.

EDIT: Cakewalk actually hung at the dropout in my case, and never came back. That's going to make stripping it down more difficult if it persists.

Edited September 29, 2019 by David Baay

[gmp]

Thanks David. Noel tried it and found out it's a buffer problem as he expected. He said he hopes to fix it. Here's his post

 

 

[Noel Borthwick]

20 hours ago, Matthew Sorrels said:

I couldn't get it to actually stop the audio engine but I did notice a giant CPU spike on one core at around the 4:20 mark (starting from an empty project, adding an aux track and turning off echo and then pressing play, on the empty project).  That CPU spike was large enough if this project had any audio or real content it may have caused a total drop out, or if I had less cores/CPU power.  Some sort of garbage collection/clean up maybe?

The drop out occurs since its a runaway buffer allocation loop. The engine allocates buffers typically from a preallocated pool but when there are special requirements it will allocate memory when needed (within limits). In this case it never stopped allocating so the engine forces a dropout and releases all memory buffers as a failsafe recovery. The spike you see is likely the deallocation.

This is mainly a failsafe situation and doesn't happen normally. Somehow in this circuit its not properly recycling buffers which may be the root issue here. These issues can be tricky to fix so it might take awhile to diagnose.

[chuckebaby]

I downloaded the bundle file this morning (all 554MB)  but haven't had a chance to load it up yet. If Noel is good with it and knows whats going on I see no reason for me to trouble shoot this. Im just glad it was able to be verified. Problems like this are sometimes very difficult to nail down and only come up in 0.1% of the time.

[gmp]

7 hours ago, Chuck E Baby said:

I downloaded the bundle file this morning (all 554MB)  but haven't had a chance to load it up yet. If Noel is good with it and knows whats going on I see no reason for me to trouble shoot this. Im just glad it was able to be verified. Problems like this are sometimes very difficult to nail down and only come up in 0.1% of the time.

Don't waste any time troubleshooting since Noel is already on top of it, but thank fro being wiling. And I must say I have a great debt of gratitude to you for pointing me to that obscure post over a year ago that fortunately you remembered. This gave me the clue to try turning on and off input echo.

 You've helped solve a year or so of frustration and troubleshooting, trying to get to the bottom on all 3 of these issues - the dropouts at 2:08, dropouts at 4:17 and the strange dropouts I'd get even when I left the computer for 5 min or more. So this is  a big relief to get this behind me. Thanks for all the excellent help you give in this forum 

[John]

Chuck has one heck of a good memory and is downright helpful. One of the best mods on the this forum!

[gmp]

53 minutes ago, John said:

Chuck has one heck of a good memory and is downright helpful. One of the best mods on the this forum!

I sure agree and agree also that we have some excellent mods and helpful people on this forum

[Noel Borthwick]

This issue has now been fixed for the 09 release. Thanks for your help sending a repro project file.
BTW you can locally fix the issue by temporarily deleting and undoing the deletion of the Aux track.

[gmp]

6 hours ago, Noel Borthwick said:

This issue has now been fixed for the 09 release. Thanks for your help sending a repro project file.
BTW you can locally fix the issue by temporarily deleting and undoing the deletion of the Aux track.

That's pretty interesting. So since I'm using 2017.07 that deleting and undoing the deletion of the Aux track, does that in effect bring the Aux Track up to date. In other words was the problem with the Aux Track because I originally created it using an older version of CbB and now the newest version 2017.07 doesn't have this problem when creating a new Aux Track?

The reason I ask is someone tried to test this by creating a new Aux Track in one of his projects instead of downloading my BUN and he wasn't able to replicate the bug.

Still this delete and undo is a pretty nice workaround. Do you think this will work on my template that has an Aux Track? And then new songs I create using the template will no longer have this bug?

Thanks again for your super attentive laser focus on making CbB the best. I knew when Bandlab got you they'd have a gem with your long years of experience with Cakewalk ..

[Noel Borthwick]

No it was actually a bug in the code setting up the playback state for the aux track (the patchpoint actually) that was leading to this bug. It would continue consuming memory until it dropped out. 
Your project had the patch point outputs in a different order than the default which shouldn't have been a problem, but it was confusing the code and preventing the patchpoint output from ever streaming. Deleting and undoing would restore it which is started working after that. 

In 09 (which is live now) the streaming code properly handles this case so no workaround is necessary. You should be able to load and play your project normally without dropouts.

PS: the new dropout diags in 09 flags this with error code 12, "Internal audio buffer memory limit overflow" should such a problem reoccur, making it much easier to highlight such problems.

[Matthew Simon Fletcher]

On 10/2/2019 at 5:04 AM, Noel Borthwick said:

No it was actually a bug in the code setting up the playback state for the aux track (the patchpoint actually) that was leading to this bug. It would continue consuming memory until it dropped out. 
Your project had the patch point outputs in a different order than the default which shouldn't have been a problem, but it was confusing the code and preventing the patchpoint output from ever streaming. Deleting and undoing would restore it which is started working after that. 

In 09 (which is live now) the streaming code properly handles this case so no workaround is necessary. You should be able to load and play your project normally without dropouts.

PS: the new dropout diags in 09 flags this with error code 12, "Internal audio buffer memory limit overflow" should such a problem reoccur, making it much easier to highlight such problems.

Hi Noel,

Apologies if this isn't related but I'm currently experiencing issues with error code 12 within a project and also freezes leading to crashes.
Whilst this project predates the 09 version, I'm now running the latest version and have saved the project since - i'm wondering though whether there is something that is baked into the save.

I do have aux tracks in this which I have since removed and reinstated which I thought had fixed it but it then returned to doing this. The problem occurs on two different machines but seems to be more frequent on one, which I assume is related to the fact that this relates to increasing levels of memory consumption. This was the only thread I found relating to this particular code, so I assume it's quite an unusual and specific bug.

I've got a ticket open but we've not progressed past standard troubleshooting processes and i'm fairly sure there's something more significant at work.
As this is preventing me from working I would be most grateful if you would let me send you the link to the project and some dump files.

Best wishes,
Matthew

[Noel Borthwick]

@Matthew Simon Fletcher please PM me the link as well as the case number you have.