Intermittent late buffers and glitches

[David Inglesfield]

I have an intermittent problem with late buffers and resulting audible glitches.

I’m a bit mystified as to what’s going on exactly. I’ll list a few things that I’ve noticed. These are all relating to one particular project.

NB: The driver setting is ASIO (Audiophile 24/96 PCI card). The track consists only of MIDI tracks (no audio tracks), plus a number of VST effects plug-ins.  I should mention that I am using two instances of Shortcircuit (32 bit sampler VST). I know it's old but I find that it runs pretty well, usually with no problems of this kind. All the samples loaded are located in the project folder, and all are 32 bit float, 44.1 khz (same as the project).

1. Yesterday, with buffer size set to 512, the engine load peak was starting at about 50%, but then as I played through the track, it would go over 100% up to ten times by the end of the track, with a glitch audible with each peak above 100%.

2. I altered the buffer size to 256, and the engine load peak now starts at about 72% but increases only to 83% or 86% and doesn’t generally go over 100%. (The one exception to this was the first time I played the track this morning, where at one point it jumped to 466%, but this didn’t happen on subsequent playbacks.)

Regarding the buffer size setting, I was under the impression that it would be the other way around, i.e. a smaller buffer size (going more towards ‘fast’ on the adjustment slider) would be more likely to produce glitches. I did originally have it on 256, but a long time ago I altered it to 512 and that fixed the glitches I was experiencing at that time. Now it's back on 256, as described above. But it's going in the opposite direction that appears to have fixed the issue.

3. The amount of RAM used by the project varies at different times. Using Task Manager, I could see last night that Cakewalk, with this project loaded, was reported as using as much as 700 to 900 MB. I even saw it increase to over 1 GB at one point. BUT the same project is at other times reported as using only 300 or 400 MB. This morning, the RAM usage was initially reported as around 700 or 800 MB, but some time after it was running at around 300 MB (it further decreased and at one point I saw it as low as 132 MB). Why is there such a huge range, and is it significant in increasing/reducing glitches?

4. Regarding freezing tracks, I have used this in the past to reduce the engine load and eliminate glitches. But yesterday, when I tried this on two different VSTs (Thump One and Odin), the effect was the opposite, i.e. it dramatically increased the amount of glitching.

NB: The one thing I’ve just thought of with regard to freezing is that when I used it in the past, the audio tracks created by the freeze were running off an internal mechanical drive. Not long ago I replaced that internal HD with an SSD, so yesterday the frozen audio tracks were running off the internal SSD. Would that be likely to cause a problem, versus a mechanical? (The SSD, a Samsung 840 EVO, has 800 GB free out of 1 TB, by the way.)

Anyway, as freezing tracks was problematic, I unfroze them and am now back to all the VSTs running live.

As mentioned above, the track is currently playing OK, but I don’t understand why, and I would like to understand, so as to avoid these problems in future.

[John Vere]

Your SDD drive can absolutely cause this issue. Just last winter I started having problems and a new drive completely solved it. 
I use Samsung SSD drives and they have a app called the Magician. You can run that to analyze.
But best of all it will clone any drive including you OS drive and you just swap it out and carry on. Took 20 minutes. I bought two 500 GB drives at $50 each and it was like getting a new computer.  

The drives start to slow down after so many read/ write cycles and recording will do that. 
 

The other thing comes to mind is there’s always a big chance of a plug in causing you grief and looking at meters might not clue you in. I can’t use Isotope stuff at all as example.  
All you can do is remove them one by one until you find the culprit. 
And one last item is to make sure your system is optimized by running this https://www.resplendence.com/latencymon

[David Inglesfield]

22 minutes ago, John Vere said:

Your SDD drive can absolutely cause this issue. Just last winter I started having problems and a new drive completely solved it. 
I use Samsung SSD drives and they have a app called the Magician. You can run that to analyze.
But best of all it will clone any drive including you OS drive and you just swap it out and carry on. Took 20 minutes. I bought two 500 GB drives at $50 each and it was like getting a new computer.  

The drives start to slow down after so many read/ write cycles and recording will do that. 
 

The other thing comes to mind is there’s always a big chance of a plug in causing you grief and looking at meters might not clue you in. I can’t use Isotope stuff at all as example.  
All you can do is remove them one by one until you find the culprit. 
And one last item is to make sure your system is optimized by running this https://www.resplendence.com/latencymon

Thank you for that. I had been thinking about downloading Magician but hadn't got round to it. I think I'll do it now. The drive is fairly new but it might be something to do with that, certainly with regard to the worse performance when VSTs were frozen.

I've also concluded that the VST Thump One was causing excessive engine load. I was using it to generate a kick drum, but exporting audio of the kick sound and running it as a sample instead through Shortcircuit,  while disabling Thump One, has resulted in a much lighter engine load, peaking generally at about 60%. I'll probably leave the buffer setting on 256. That did cause me problems in the past, but maybe I can get away with it on some tracks, and switch to 512 only when there are issues.

Edited October 26, 2023 by David Inglesfield
Corrected typo

[Billy86]

Are you using any Soundtoys plugins?  The last update has caused known intermittent glitch issues in CW. There is a thread on it on this forum.  Like others, I’m waiting for the fix. 

[David Inglesfield]

48 minutes ago, Billy86 said:

Are you using any Soundtoys plugins?  The last update has caused known intermittent glitch issues in CW. There is a thread on it on this forum.  Like others, I’m waiting for the fix. 

No, no Soundtoys plugins.

Updating what I wrote yesterday, I downloaded Samsung's Magician and it reported that my SSD is in 'Good' condition with no bad sectors.

The Toybox Audio plugin Thump One was definitely causing problems, but I'm still confused as to why,  at the stage that I was still running Thump One, altering buffer settings from 512 to 256 improved performance. And also why freezing VSTs led to worse performance, when there was a huge amount of space free on a fairly new SSD reported as being in 'Good' condition.

[Bill Phillips]

On 10/26/2023 at 10:42 AM, John Vere said:

Your SDD drive can absolutely cause this issue. Just last winter I started having problems and a new drive completely solved it. 
I use Samsung SSD drives and they have a app called the Magician. You can run that to analyze.
But best of all it will clone any drive including you OS drive and you just swap it out and carry on. Took 20 minutes. I bought two 500 GB drives at $50 each and it was like getting a new computer.  

The drives start to slow down after so many read/ write cycles and recording will do that. 
 

The other thing comes to mind is there’s always a big chance of a plug in causing you grief and looking at meters might not clue you in. I can’t use Isotope stuff at all as example.  
All you can do is remove them one by one until you find the culprit. 
And one last item is to make sure your system is optimized by running this https://www.resplendence.com/latencymon

Thank you! All drives in my new DAW are Samsung. I've been using 3rd party cloning software when I need it which isn't often but now I can use Magician which also has a lot of testing and troubleshooting features. I opened it up and learned that my 3 new M.2 PCIe drives needed firmware updates which I did all at once and very quickly. Thanks again.