The No.1 Website for Pro Audio
 All  This Thread  Reviews  Gear Database  Gear for sale     Latest  Trending
Huge latency issues on the latest 1903 Windows release - SOLVED
Old 2 weeks ago
  #361
Gear Addict
 
daskeladden's Avatar
 

Quote:
Originally Posted by EvilDragon View Post
This doesn't happen with a particularly busy CPU, though. Sure Kontakt is streaming 300-400 voices in these cases, but my CPU meter is ~10%...


And again, this did not happen before 1903 - with those same instruments, and same audio interface and other hardware.
Have you tested Windows Performance Monitor? Test one thing at the time CPU, Memory, Network adapter and so on. I found out that I have some problems with my memory which I have expected for a long time. I get hard page faults also when running nothing. Memory is cheap now so gonna buy some soon. By the way this do not show in Task Manager or Latencymon.
Here is a screenshot when testing only memory:


Here is how you use Windows Performance Monitor:
https://www.windowscentral.com/how-u...tor-windows-10
Old 2 weeks ago
  #362
Lives for gear
 
EvilDragon's Avatar
Thanks, will check that out. 64 gigs of pretty good quality HyperX (Kingston) DDR4 here... of course, a faulty RAM stick can always happen.
Old 2 weeks ago
  #363
Gear Maniac
 
sosho's Avatar
by accident I have the same Kingston , 32 GB . I checked it several times overnight booting of a cd or stick . memory can also go bad , happened to me years ago on an old powermac . But 1809 would probably have had the same symptoms
Old 2 weeks ago
  #364
Quote:
Originally Posted by EvilDragon View Post
Thanks, will check that out. 64 gigs of pretty good quality HyperX (Kingston) DDR4 here... of course, a faulty RAM stick can always happen.
Can you get a glitch trace recorded demonstrating the artifacts?

If so, please send me a link to the feedback hub entry.

https://matthewvaneerde.wordpress.co...-feedback-hub/

Pete
Old 2 weeks ago
  #365
Lives for gear
 
EvilDragon's Avatar
Quote:
Originally Posted by Psychlist1972 View Post
Can you get a glitch trace recorded demonstrating the artifacts?
I can try, how do I get a glitch trace?
Old 2 weeks ago
  #366
Quote:
Originally Posted by EvilDragon View Post
I can try, how do I get a glitch trace?
Step 6 in that blog post. Glitch Trace = Log while audio is playing and the glitch happens.

Pete
Old 2 weeks ago
  #367
Lives for gear
 
EvilDragon's Avatar
Ah! Gotcha. Here's the link to the feedback hub entry: https://aka.ms/AA5sexb
Old 1 week ago
  #368
Lives for gear
I hope it's OK if I ask here. I'm having an issue that is a bit different to the one mentioned in this thread but somewhat related. This issue only started in the last few days on my system (and I haven't made any significant changes at all since then).

ASIO audio works fine, but when playing audio through any Windows application like foobar2000, PotPlayer, Google Chrome .etc, I get an audible stutter every 3 - 4 seconds. I'm using a Focusrite Clarett 2Pre as my audio interface.

I'm on Windows 10 1903 and have ensured the patch is installed above.

I've gone through Windows Sound troubleshooter and also ensured that any Audio Enhancements in Windows are off and that my sample rate matches that of the playing audio file.

Has anyone experienced this and are there any suggestions I can try?

I have had this issue once in the past too, and again it just seemed to come out of nowhere. I can't recall how I got rid of it last time, likely a re-install.

The issue does not occur when using HDMI audio output to my computer monitor.

Edit: Just re-installed and upgraded the Focusrite Clarett drivers and I think maybe that has resolved it. Still very strange as I had been using those same 2.1.9 drivers for months without an issue and all of a sudden, it started.
Old 1 week ago
  #369
Quote:
Originally Posted by fgimian View Post
I hope it's OK if I ask here. I'm having an issue that is a bit different to the one mentioned in this thread but somewhat related. This issue only started in the last few days on my system (and I haven't made any significant changes at all since then).

ASIO audio works fine, but when playing audio through any Windows application like foobar2000, PotPlayer, Google Chrome .etc, I get an audible stutter every 3 - 4 seconds. I'm using a Focusrite Clarett 2Pre as my audio interface.

I'm on Windows 10 1903 and have ensured the patch is installed above.

I've gone through Windows Sound troubleshooter and also ensured that any Audio Enhancements in Windows are off and that my sample rate matches that of the playing audio file.

Has anyone experienced this and are there any suggestions I can try?

I have had this issue once in the past too, and again it just seemed to come out of nowhere. I can't recall how I got rid of it last time, likely a re-install.

The issue does not occur when using HDMI audio output to my computer monitor.

Edit: Just re-installed and upgraded the Focusrite Clarett drivers and I think maybe that has resolved it. Still very strange as I had been using those same 2.1.9 drivers for months without an issue and all of a sudden, it started.
Same bit depth and rate on Windows audio as you use on ASIO?

Has it been working fine since the reinstall?

Pete
Old 1 week ago
  #370
Lives for gear
Quote:
Originally Posted by Psychlist1972 View Post
Same bit depth and rate on Windows audio as you use on ASIO?

Has it been working fine since the reinstall?

Pete
Thanks for the response Pete, definitely the same sample rate, I did try both 24-bit and 16-bit settings in the Advanced panel (the audio file is 16-bit) and both caused the same glitches.

And yeah, it had been working fine for several months which is when I had re-installed. I listen to music constantly when I'm not producing so was quick to notice.

Strangely, installing the latest beta Clarett driver (3.4.0) seems to have sorted it out. I'm not sure if it was the update or the re-install of the driver which helped, very strange.

If it pops up again, I'll let you know and we can troubleshoot further before I re-install drivers

Thanks again!!!
Fotis
Old 1 week ago
  #371
Gear Maniac
 

Quote:
Originally Posted by fgimian View Post
Thanks for the response Pete, definitely the same sample rate, I did try both 24-bit and 16-bit settings in the Advanced panel (the audio file is 16-bit) and both caused the same glitches.

And yeah, it had been working fine for several months which is when I had re-installed. I listen to music constantly when I'm not producing so was quick to notice.

Strangely, installing the latest beta Clarett driver (3.4.0) seems to have sorted it out. I'm not sure if it was the update or the re-install of the driver which helped, very strange.

If it pops up again, I'll let you know and we can troubleshoot further before I re-install drivers

Thanks again!!!
Fotis
Hi, can you provide the link to the 3.4.0 Clarett driver. I'm running the 2.1.9 beta
Old 1 week ago
  #372
Old 1 week ago
  #373
Lives for gear
Quote:
Originally Posted by doughnuts View Post
Beat me to it, that's indeed where I downloaded it from

I just realised that it says:

Quote:
Windows Clarett Thunderbolt users connecting to computers with Thunderbolt 3 ports should continue to use Focusrite Control 2.1.9 below
I'm using a Thunderbolt 3 based PC and am running 3.4.0 without issue thus far, but obviously YMMV.
Old 1 week ago
  #374
Gear Head
 
jjnssn's Avatar
 

Quote:
Originally Posted by jjnssn View Post
Hey daskeladden,
Thank you for starting this thread and following it through. I have been able to record two consecutive 20 min session in protools with no failures with the Latest KB in 1903.

Pete thank you also. I am grateful for your and the folks at MS persistence on this issue.

Cheers!
J
I updated with KB4506991 today. Out of curiosity I ran Latency Monitor while I was working on other things.

Had a spike with NTOSKRNL at 500 µs somewhere in between 38 minutes. I don't recall having run latency monitor last time after the initial fix, rather only what really matters and that in uninterrupted recording. I'll test the recording again as well as removing this latest KB to verify.
Old 1 week ago
  #375
Lives for gear
 
b0se's Avatar
So is this actually fixed then? It seems sporadic in its efficacy.
Old 1 week ago
  #376
Gear Addict
 
daskeladden's Avatar
 

So I did the Cumulative update KB4512508 today and the system is still pretty good. Maybe the ntoskrnl.exe spikes are a little bit higher again? I did get one ntoskrnl.exe spike to 300 which I never did after the KB4505903 fix. I have not tested Latencymon more than 10 minutes ... Can somebody else test this...
Old 1 week ago
  #377
Quote:
Originally Posted by b0se View Post
So is this actually fixed then? It seems sporadic in its efficacy.
The issue we introduced in 1903 has been fixed.

I suspect many of the other issues brought up are not related to that fix/problem, but folks are paying more attention to reported numbers, or there are other unrelated issues.

Just a hunch, as I do not have any real way to prove that at the moment.

Pete
Old 1 week ago
  #378
Gear Head
 
jjnssn's Avatar
 

Quote:
Originally Posted by jjnssn View Post
I updated with KB4506991 today. Out of curiosity I ran Latency Monitor while I was working on other things.

Had a spike with NTOSKRNL at 500 µs somewhere in between 38 minutes. I don't recall having run latency monitor last time after the initial fix, rather only what really matters and that in uninterrupted recording. I'll test the recording again as well as removing this latest KB to verify.


Although I measured the spike yesterday, I setup and have completed 2 separate 40 min recording sessions 12 Audio/1 MIDI with Aux Sends, 64 buffers on my Avid HDX system without any stop in recording.

I would say the latest KB hasn't changed the performance since the fix was put in place, meaning the spike I measured is not significant for what I am doing.
Old 1 week ago
  #379
Lives for gear
 
EvilDragon's Avatar
@ Psychlist1972 - are there any useful bits that could be extracted from my glitch trace?
Old 1 week ago
  #380
Quote:
Originally Posted by EvilDragon View Post
@ Psychlist1972 - are there any useful bits that could be extracted from my glitch trace?
I haven't yet heard back from the team on that. I'll ping.

Pete
Old 1 week ago
  #381
Gear Addict
 
daskeladden's Avatar
 

Just for fun I tried to disable my network card and turn off antivirus (Microsoft defender) to see how low latency I can get out of my system. Pretty pleased with the result. One of these days I gonna get a dedicated pc for music so I can turn off "everything". By the way here are the result:
Old 6 days ago
  #382
Quote:
Originally Posted by daskeladden View Post
Just for fun I tried to disable my network card and turn off antivirus (Microsoft defender) to see how low latency I can get out of my system. Pretty pleased with the result. One of these days I gonna get a dedicated pc for music so I can turn off "everything". By the way here are the result:
Can you try with disabling network but *not* Defender?

Pete
Old 6 days ago
  #383
Gear Nut
 
SF_Green's Avatar
Quote:
Originally Posted by daskeladden View Post
Just for fun I tried to disable my network card and turn off antivirus (Microsoft defender) to see how low latency I can get out of my system. Pretty pleased with the result. One of these days I gonna get a dedicated pc for music so I can turn off "everything". By the way here are the result:
Impressive interupt, ISR, and DPC numbers. But your pagefaults are high, especially for 32 Gb of RAM.

Any comment on that, Pete?
Old 6 days ago
  #384
Quote:
Originally Posted by SF_Green View Post
Impressive interupt, ISR, and DPC numbers. But your pagefaults are high, especially for 32 Gb of RAM.

Any comment on that, Pete?
Network is always going to have higher DPC timings, especially WiFi and Bluetooth. Wired can as well, of course.

Page faults: Not sure, there are a number of different types. It really depends on how the latency checker is looking for page faults, and if it's doing anything to distinguish between the different types. It's not a number I'd micro-manage.

We don't have a good explanation of it, but this is helpful (but old):

https://stackoverflow.com/questions/...es-page-faults

Edit: I know latencymon says "hard pagefault", but I don't know how accurate that is.

Edit2: JFYI: Hitting "edit" for a post here is enough to get me a 506us DPC time for ndis.sys (intel wired). Networking is so essential, but if you're in a deep recording session and want to keep things to a minimum, the old advice to temporarily disable network still holds up. Of course, if you don't hear a glitch, then it's really just managing to a number.

Pete
Old 6 days ago
  #385
Gear Maniac
I didn't realize I left LatencyMon while recording at 32 asio buffer using Vsti and an amp sim. I initially turned it on after updating windows 10 1809. Take a look at the time.
Attached Thumbnails
Huge latency issues on the latest 1903 Windows release - SOLVED-20190816_224601.jpg  
Old 6 days ago
  #386
Gear Addict
 
daskeladden's Avatar
 

Quote:
Originally Posted by SF_Green View Post
Impressive interupt, ISR, and DPC numbers. But your pagefaults are high, especially for 32 Gb of RAM.

Any comment on that, Pete?
Hardpage fault seems pretty random and could be related to a process in the background or my memory I think. Latencymon itself also cause some hardpage fault. I have my bluetooth enable so that can cause some. But most of all I think it is related to my memory. My layout is not ideal cause I have two 8GB and one 16GB making a total of 3 sticks. I'm in the process of buying two 16GB sticks with higher MHz. Here are the screenshot of the pagefaults on my last Latencymon test with defender enabled. I think msmpeng.exe is related to defender:


Quote:
Originally Posted by Psychlist1972 View Post
Can you try with disabling network but *not* Defender?

Pete
Here it is with just the network card disabled:
Old 6 days ago
  #387
Quote:
Originally Posted by daskeladden View Post
Hardpage fault seems pretty random and could be related to a process in the background or my memory I think. Latencymon itself also cause some hardpage fault. I have my bluetooth enable so that can cause some. But most of all I think it is related to my memory. My layout is not ideal cause I have two 8GB and one 16GB making a total of 3 sticks. I'm in the process of buying two 16GB sticks with higher MHz. Here are the screenshot of the pagefaults on my last Latencymon test with defender enabled. I think msmpeng.exe is related to defender:



Here it is with just the network card disabled:
Great, thanks. Even though that is slightly higher, it's still perfectly acceptable. I don't generally recommend that people run without AV, even when (temporarily) disconnected from the network. There's USB, old drives, etc. that are also vectors.

Page faults with defender: when you re-enabled it, it may have started a scan. Or it may have been writing a bunch to the event log. Not sure. But, if it's not materially impacting your perf here, I would look at it a little cockeyed, but would otherwise ignore.

Pete
Old 5 days ago
  #388
Gear Maniac
 
sosho's Avatar
Some programms need the network card for the copy protection (MAC ?) .
latency is also slightly lower on my machine with enabled intel ethernet but with the cable unplugged . Defender is always active here
Old 4 days ago
  #389
Quote:
Originally Posted by Psychlist1972 View Post

Edit2: JFYI: Hitting "edit" for a post here is enough to get me a 506us DPC time for ndis.sys (intel wired). Networking is so essential, but if you're in a deep recording session and want to keep things to a minimum, the old advice to temporarily disable network still holds up. Of course, if you don't hear a glitch, then it's really just managing to a number.

Pete
While on some systems network drivers are sub par, by itself it should not be a dealbreaker.

I test our X2 line with 128 channels 96k recording overdubs at 64 samples buffer for hours and we do not accept a single dropout in multiple passes. We have 2 onboard ethernet adapters on these systems, one connected to internet (Defender active) and the other typically in a closed network like a dedicated Dante network.
If these systems pass this scenario for multiple generations (they do), I am pretty confident it is fair to say that internet/ethernet as a realtime audio dealbreaker parameter is panic based on very old configurations of Windows, drivers and adapters, combined with slower CPUs and system bandwidth.

That doesn't mean it does not make sense to verify it on your own system, it means that most likely there is no reason to be offline.

In short: Test whether your drivers behave well over at least an hour, but don't panic about ethernet.
Old 4 days ago
  #390
Gear Maniac
 
sosho's Avatar
Except for not getting distracted ; -) I was just checking out of curiosity if there is a difference between disconecting the cable or disabling internet access for that machine on the router . I was even playing back small projects while being online via a cheap Ralink Wifi stick , no audible glitches .
I decided to check latencymon only in case I get audible glitches .
Post Reply

Welcome to the Gearslutz Pro Audio Community!

Registration benefits include:
  • The ability to reply to and create new discussions
  • Access to members-only giveaways & competitions
  • Interact with VIP industry experts in our guest Q&As
  • Access to members-only sub forum discussions
  • Access to members-only Chat Room
  • Get INSTANT ACCESS to the world's best private pro audio Classifieds for only USD $20/year
  • Promote your eBay auctions and Reverb.com listings for free
  • Remove this message!
You need an account to post a reply. Create a username and password below and an account will be created and your post entered.


 
 
Slide to join now Processing…
Thread Tools
Search this Thread
Search this Thread:

Advanced Search
Forum Jump
Forum Jump