Voicemeeter Banana - Audio pops/clicks, system stutter after Win 10 v2004 update

The Virtual Audio Mixer discussions and support...
HOWIEC
Posts: 12
Joined: Mon Apr 30, 2018 5:44 am

Voicemeeter Banana - Audio pops/clicks, system stutter after Win 10 v2004 update

Post by HOWIEC »

Hi,

I recently installed the Win10 May 2020 update and noticed audio pops/clicks and stutter during gaming or some video/audio playback.

Not sure if this is normal but LatencyMon shows a very high frequency of DPCs for HDAudBus.sys when Banana is running.
Upon exiting Banana, the DPCs slow down a lot, average interrupt to process latency drops too, and then I don't seem to get the audio pops/clicks and stutter during gaming.

I uninstalled and reinstalled both VB Cable and Banana v2.0.5.3 as instructed to do after an update and checked all settings including audio formats (e.g. 24bit 48kHz). I'm using the onboard Realtek audio device on an Asus Maximus X Hero WiFi mb and didn't have this issue prior to the v2004 update.

Attached are some screenshot examples of DPC counts with and without Banana running.

Are you guys having similar issues?

Thanks,
Howie

DPC count with Banana running:
Image

DPC count without Banana running:
Image
HOWIEC
Posts: 12
Joined: Mon Apr 30, 2018 5:44 am

Re: Voicemeeter Banana - Audio pops/clicks, system stutter after Win 10 v2004 update

Post by HOWIEC »

Maybe we need a new Realtek audio driver for Win10 v2004?

UPDATE:
I figured it's going to take a while before Microsoft, Realtek, or someone else fixes this so I went back to v1909 and everything's back to normal...
I tried various Realtek drivers and even used MS's basic driver on v2004 but the audio issues persisted. This leads me to believe that there's something inherently wrong within the OS itself vs drivers.
xcasxcursex
Posts: 173
Joined: Tue Feb 18, 2020 12:04 am

Re: Voicemeeter Banana - Audio pops/clicks, system stutter after Win 10 v2004 update

Post by xcasxcursex »

The DPC counts increasing significantly are normal. You should expect them since the driver is in use. It's not running on voodoo magic ;) Remember, LatencyMon is mostly intended for testing an IDLE system. Likewise, you should expect to see an increase in average int to process latency. The system is doing stuff. There are use-cases for testing during execution like this, but....

As per the author:
Should I run my audio software while LatencyMon is testing my system ?
In general, no. In particular, the interrupt to process latency that the software measures already simulates the workings of an entire audio process. You should run most tests without running any other software in the background. Running your audio software together with LatencyMon only makes sense if you wish to measure hard pagefaults of the audio process.

For example I'm here on 1909 with VM not running vs running over a 20s period I see 200 vs 16742. In your case you're actually seeing a lot more without VM running, 4 times more than the no-VM-to-VM ratio I have, but that's likely because I have super small buffers set and am beating the crap out of it ;) Still, I don't see anything out of the ordinary there, and it might even be better than what I'm seeing. But there are so many variables between your setup and mine, it's impossible to make anything useful of these observations really, other than what we already know - increasing DPC counts is normal.

Also, your execution times when VM is shutdown are quite high compared to mine. I get 0.014779 with it not running and 0.056697 running - a ratio of 3.83:1. In your case it's practically 1:1. Now, these kind of numbers are not to be taken as something standalone, but I did notice that stand out. That's far more interesting than the increasing DPC count. Then again, it's likely again just because of my config being more demanding.

Run that test again on 1909 - if you want to compare the two OS's, then compare them....
Related - don't transfer the settings across manually, humans are error prone, and what this results in, is not "comparing 1909 vs 2004" but "comparing 1909 and a human vs 2004 and a human". Backup your settings and import them.

Speaking of error-prone - Windows version upgrades done in place over the top of an existing installation should be set on fire at best, and at worst trusted about as far as you could throw Redmond.


I've just a little doubt at this point that something is awry with 2004's audio (there are a few posts about it here) but I don't think it's time to jump ship back to 1909 just yet. I can only assume you've posted about this elsewhere (lots of views on imgur) what's the chatter like over there? Got a link?
HOWIEC
Posts: 12
Joined: Mon Apr 30, 2018 5:44 am

Re: Voicemeeter Banana - Audio pops/clicks, system stutter after Win 10 v2004 update

Post by HOWIEC »

xcasxcursex wrote:The DPC counts increasing significantly are normal. You should expect them since the driver is in use. It's not running on voodoo magic ;) Remember, LatencyMon is mostly intended for testing an IDLE system. Likewise, you should expect to see an increase in average int to process latency. The system is doing stuff. There are use-cases for testing during execution like this, but....
The DPC count seems to be increasing too rapidly on v2004. It's almost 14x the count after 1min than on v1909!
Vincent Burel also commented saying there seems to be something else going on independent of Banana.
Image

Running LatencyMon while some CPU or other resources are being consumed is not an issue if you know what you're doing and are comparing similar metrics under specific and repeatable scenarios. All of those screenshots are from when my PC was "idling" at a ~1 or 2% CPU usage.
xcasxcursex wrote:Also, your execution times when VM is shutdown are quite high compared to mine. I get 0.014779 with it not running and 0.056697 running - a ratio of 3.83:1. In your case it's practically 1:1. Now, these kind of numbers are not to be taken as something standalone, but I did notice that stand out. That's far more interesting than the increasing DPC count. Then again, it's likely again just because of my config being more demanding.
Which driver execution time(s) are you specifically referring to?
Of course these times are going to be different between mine and yours as it depends on many factors including HW specs/performance and resource loading...
xcasxcursex wrote:Run that test again on 1909 - if you want to compare the two OS's, then compare them....
Related - don't transfer the settings across manually, humans are error prone, and what this results in, is not "comparing 1909 vs 2004" but "comparing 1909 and a human vs 2004 and a human". Backup your settings and import them.
The symptoms I described was the comparison, nevertheless, see pic above now that I've downgraded to 1909.
How exactly do you suggest I "backup/restore settings" going from 1909 to 2004? There are only a handful of ones that I need to check which is quicker than unnecessarily digging up and create registry keys/files. Also, downgrading to 1909 automatically restored all settings. Of course, I check settings after reinstalling certain drivers just in case (audio, graphics).
xcasxcursex wrote:I've just a little doubt at this point that something is awry with 2004's audio (there are a few posts about it here) but I don't think it's time to jump ship back to 1909 just yet. I can only assume you've posted about this elsewhere (lots of views on imgur) what's the chatter like over there? Got a link?
I posted on Reddit: https://www.reddit.com/r/Windows10/comm ... opsclicks/

If you don't have problems, then no need to go back. I did have problems on v2004.
xcasxcursex
Posts: 173
Joined: Tue Feb 18, 2020 12:04 am

Re: Voicemeeter Banana - Audio pops/clicks, system stutter after Win 10 v2004 update

Post by xcasxcursex »

HOWIEC wrote:The DPC count seems to be increasing too rapidly on v2004. It's almost 14x the count after 1min than on v1909!
As I said, at least on your PC, they are decreasing LESS compared to 1909, than mine. Now this is only relevant in a 1909 vs 2004 discussion, if we were to test 2004 on my PC, which I WILL be doing ASAP, and if you are to test 1909 on yours.... but it's just an indication that the DPC counts you're seeing are nothing to be concerned about. On that note:
HOWIEC wrote:Running LatencyMon while some CPU or other resources are being consumed is not an issue if you know what you're doing and are comparing similar metrics under specific and repeatable scenarios. All of those screenshots are from when my PC was "idling" at a ~1 or 2% CPU usage
Your PC isn't idling, certainly with respect to audio drivers, when VM is running. I get what you mean by 'idling' in the sense that is it barely utilised and not being operated, and that's reasonable use of the term, EXCEPT in this context, where we are specifically talking about utilisation of the audio driver, in which case, if VM is running, it's the opposite of idle, it's constantly capturing and playing back audio (even when you're not playing any audio!).

The only specific repeatable scenario you used LatencyMon to test, was that the driver was in fact generating DPCs as it was actively processing audio, when it should have been, when VM is running. There's NO indication of a problem there, at all.

Which driver execution time(s) are you specifically referring to?
Same as you, the DPC.
Of course these times are going to be different between mine and yours as it depends on many factors including HW specs/performance and resource loading...
Which I said:
Now, these kind of numbers are not to be taken as something standalone, but I did notice that stand out. That's far more interesting than the increasing DPC count. Then again, it's likely again just because of my config being more demanding.
And don't forget, the point here isn't the counts, it's the ratio between them.
How exactly do you suggest I "backup/restore settings" going from 1909 to 2004?
Using the menus in VM.
There are only a handful of ones that I need to check which is quicker than unnecessarily digging up and create registry keys/files.
That's nice but speed is not the object here, consistency is. A handful of settings is a handful you can get wrong. It's not about the likelihood of you getting them wrong, it's about completely excluding human error as a possible cause of problems.
xcasxcursex wrote:If you don't have problems, then no need to go back. I did have problems on v2004.
Well it's either broken or it's not. Neither would surprise me at the moment, since we're seeing scattered reports of issues, along with scattered replies that it's fine. If we look at your thread, there's a guy who modified his MMCSS and that was his problem. Now let's be clear - I DO THAT TOO. But I also know that it's something that might help or might break stuff, and I wouldn't expect an installation of a new OS version to simply work with my old settings.

Looking around, I can find a report that another guy had his CPU configured incorrectly, one guy who said his ASIO buffers had to be bigger than normal (reminds me of one of the threads here), and about double that number explicitly saying they're fine and countless who aren't reporting any issue. We're a long, long way from saying there's a problem with 2004 right now... but at the very least, there are a few teething problems.

I understand you had problems on 2004, but that doesn't mean the problem is 2004... but like I said, it wouldn't surprise me ;)
Either way, making a "Public Service Announcement" that "There is something seriously wrong with Win10 v2004 (drivers, audio stack, etc.)!" Is a little rash. I mean by all means, post about it, please, I'm glad you did! Just... Perhaps a little less '2004 broken' and more of a 'My PC broken on 2004, anyone else? Pls halp'. I'm sure it wasn't your intention to make a 'Public Disservice Announcement' ;)


Now just to be clear, I'm not saying "There's definitely nothing wrong with 2004". What I am saying, is that proper testing and analysis and evidence is required before coming to the conclusion that "There's definitely something wrong with 2004".... And we don't have that yet. It's a really good thing to talk with people if you have problems; it might solve yours, it might save them from trap, it's good.... But one shouldn't go making waves about something being broken until they can actually say so with certainty, otherwise you might be saving people from a nice upgrade for no reason, and you're trying to get help with a problem that isn't actually your problem so you get no solution....That's bad.
xcasxcursex
Posts: 173
Joined: Tue Feb 18, 2020 12:04 am

Re: Voicemeeter Banana - Audio pops/clicks, system stutter after Win 10 v2004 update

Post by xcasxcursex »

Geez I just saw this:
I have many other optimizations like disable C-states, disable EIST, disable dynamic tick, disable core parking, etc. and checked everything with v2004 but figured it's likely to be an actual Windows bug vs settings.
Yeh no. You can't go complaining that an operating system isn't working properly when you specifically program it not to work properly, LOL. Try a clean install without all the "tweaks" that mostly break things. The timer tweak stands out here, because that DEFINITELY changed in 2004 and wasn't a good idea in the first place.... But then again, none of those tweaks, are. I get a funny feeling you've been hanging around guru3d.
HOWIEC
Posts: 12
Joined: Mon Apr 30, 2018 5:44 am

Re: Voicemeeter Banana - Audio pops/clicks, system stutter after Win 10 v2004 update

Post by HOWIEC »

Jesus, there are so many flaws with your replies and logic... I'd rather not waste my time but here it goes:
xcasxcursex wrote:
HOWIEC wrote:The DPC count seems to be increasing too rapidly on v2004. It's almost 14x the count after 1min than on v1909!
As I said, at least on your PC, they are decreasing LESS compared to 1909, than mine. Now this is only relevant in a 1909 vs 2004 discussion, if we were to test 2004 on my PC, which I WILL be doing ASAP, and if you are to test 1909 on yours.... but it's just an indication that the DPC counts you're seeing are nothing to be concerned about.
Explain why you think a 14x increase in DPC counts for HDAudBus.sys between v1909 and v2004 is NOT likely to be indicative of some problem or not likely to correlate with the symptoms I described (stutter, audio pops/clicks).
xcasxcursex wrote:On that note:
HOWIEC wrote:Running LatencyMon while some CPU or other resources are being consumed is not an issue if you know what you're doing and are comparing similar metrics under specific and repeatable scenarios. All of those screenshots are from when my PC was "idling" at a ~1 or 2% CPU usage
Your PC isn't idling, certainly with respect to audio drivers, when VM is running. I get what you mean by 'idling' in the sense that is it barely utilised and not being operated, and that's reasonable use of the term, EXCEPT in this context, where we are specifically talking about utilisation of the audio driver, in which case, if VM is running, it's the opposite of idle, it's constantly capturing and playing back audio (even when you're not playing any audio!).The only specific repeatable scenario you used LatencyMon to test, was that the driver was in fact generating DPCs as it was actively processing audio, when it should have been, when VM is running. There's NO indication of a problem there, at all.
There are cases where you could use LatencyMon while other applications, even games, are running to diagnose certain drivers or see generally if your PC is capable of handling the various loads. I'm not going to argue this any further because this is mostly a separate topic.

You state that we shouldn't run LatencyMon with Voicemeeter running (even when my CPU is basically idle at ~1%) and yet that's the only way to see the rate/count of DPCs HDAudBus.sys generates with that configuration...
xcasxcursex wrote:
Which driver execution time(s) are you specifically referring to?
Same as you, the DPC.
?? DPCs are not drivers or in units of time... An example of a driver is nvlddmkm.sys.
xcasxcursex wrote:Using the menus in VM.
None of Voicemeeter's settings changed during the upgrade from v1909 to v2004... This is moot anyway because all my settings were correct and I reinstalled VB Cable and Voicemeeter.
xcasxcursex wrote:Well it's either broken or it's not. Neither would surprise me at the moment, since we're seeing scattered reports of issues, along with scattered replies that it's fine. If we look at your thread, there's a guy who modified his MMCSS and that was his problem. Now let's be clear - I DO THAT TOO. But I also know that it's something that might help or might break stuff, and I wouldn't expect an installation of a new OS version to simply work with my old settings.

Looking around, I can find a report that another guy had his CPU configured incorrectly, one guy who said his ASIO buffers had to be bigger than normal (reminds me of one of the threads here), and about double that number explicitly saying they're fine and countless who aren't reporting any issue. We're a long, long way from saying there's a problem with 2004 right now... but at the very least, there are a few teething problems.

I understand you had problems on 2004, but that doesn't mean the problem is 2004... but like I said, it wouldn't surprise me ;)
Either way, making a "Public Service Announcement" that "There is something seriously wrong with Win10 v2004 (drivers, audio stack, etc.)!" Is a little rash. I mean by all means, post about it, please, I'm glad you did! Just... Perhaps a little less '2004 broken' and more of a 'My PC broken on 2004, anyone else? Pls halp'. I'm sure it wasn't your intention to make a 'Public Disservice Announcement' ;)

Now just to be clear, I'm not saying "There's definitely nothing wrong with 2004". What I am saying, is that proper testing and analysis and evidence is required before coming to the conclusion that "There's definitely something wrong with 2004".... And we don't have that yet. It's a really good thing to talk with people if you have problems; it might solve yours, it might save them from trap, it's good.... But one shouldn't go making waves about something being broken until they can actually say so with certainty, otherwise you might be saving people from a nice upgrade for no reason, and you're trying to get help with a problem that isn't actually your problem so you get no solution....That's bad.
The guy who modified his MMCSS priorities ended up alleviating the issues he experienced after upgrading. He didn't have said issues with v1909...

The only difference between my 2 configurations was v1909 and v2004... How can you say it's not likely to be a bug/issue or driver incompatibility with or related to v2004???

That's your opinion. To me posting about my experience is not rash or "bad" because I wasted hours upgrading, reinstalling drivers and apps, checking various settings, trying to diagnose the issue, only to then downgrade, and then reinstall drivers/apps, and check settings again!!
There are plenty of other serious issues with v2004 warranting at the very least a stern warning.

You just seem skeptical that others know what they're doing and assume that the issue is with their "settings" or that they are wrong...
Last edited by HOWIEC on Sat Jun 13, 2020 5:47 am, edited 4 times in total.
HOWIEC
Posts: 12
Joined: Mon Apr 30, 2018 5:44 am

Re: Voicemeeter Banana - Audio pops/clicks, system stutter after Win 10 v2004 update

Post by HOWIEC »

xcasxcursex wrote:Geez I just saw this:
I have many other optimizations like disable C-states, disable EIST, disable dynamic tick, disable core parking, etc. and checked everything with v2004 but figured it's likely to be an actual Windows bug vs settings.
Yeh no. You can't go complaining that an operating system isn't working properly when you specifically program it not to work properly, LOL. Try a clean install without all the "tweaks" that mostly break things. The timer tweak stands out here, because that DEFINITELY changed in 2004 and wasn't a good idea in the first place.... But then again, none of those tweaks, are. I get a funny feeling you've been hanging around guru3d.
Just by stating the above, it's clear you don't know what you're talking about.
None of those settings "break" things. Some result in higher and consistent performance (or lower average latencies) at the expense of increased power consumption.
xcasxcursex
Posts: 173
Joined: Tue Feb 18, 2020 12:04 am

Re: Voicemeeter Banana - Audio pops/clicks, system stutter after Win 10 v2004 update

Post by xcasxcursex »

You say there are flaws with my replies and logic while you repeatedly misquote me.... You clearly have no ability to grasp my logic or read my replies.

Simple example and the only one I'll bother with:

Me: Also, your execution times...
You: Which driver execution time(s)
Me: Same as you, the DPC.
You: ?? DPCs are not drivers or in units of time
Me: YOU said "driver", not me. Also;
LatencyMon: "Highest execution (ms)"

LOL
We're done.
HOWIEC
Posts: 12
Joined: Mon Apr 30, 2018 5:44 am

Re: Voicemeeter Banana - Audio pops/clicks, system stutter after Win 10 v2004 update

Post by HOWIEC »

xcasxcursex wrote:You say there are flaws with my replies and logic while you repeatedly misquote me.... You clearly have no ability to grasp my logic or read my replies.

Simple example and the only one I'll bother with:

Me: Also, your execution times...
You: Which driver execution time(s)
Me: Same as you, the DPC.
You: ?? DPCs are not drivers or in units of time
Me: YOU said "driver", not me. Also;
LatencyMon: "Highest execution (ms)"

LOL
We're done.
WTF?
I asked which specific execution time(s) for which specific driver(s) and you reply with "DPC"... that's like someone asking you which is your favorite color and you reply "painting".
You still don't get it nor does it seem like you can respond to the other issues because you probably don't know how to or don't want to admit it.

On a positive note, I recommend reading really good resources like Windows Internals (Pts 1 & 2) that can help us all understand things better for Windows and OSes in general, and hence troubleshoot better. Of course, one including myself might not understand all of it.
https://repo.zenk-security.com/Linux%20 ... dition.pdf
Post Reply