The No.1 Website for Pro Audio
Huge latency issues on the latest 1903 Windows release - SOLVED
Old 14th June 2019
  #91
Lives for gear
 
daskeladden's Avatar
 

Quote:
Originally Posted by Avenue Road View Post
I started timing the spikes and they're occurring every 2 minutes on the dot. Can anyone else experiencing this try it out to see if your timing is consistent?
My spikes are totally random. Sometimes seconds after I start Latencymon, sometimes after 6 minutes and so on


I was going to check something in Event Viewer and found out that it is full of bug due to the latest cumulative update. More of the issue here:
https://support.microsoft.com/en-us/...using-custom-v

This is starting to be a parody. Cannot look at the memory diagnostic log

What I have noticed after the last cumulative update is that my spikes are much more severe. I never had over 2300, usually around 1300, and now is over 6000.
(I looked at the CPU in task manager and it was no reaction when the spikes occur, so maybe Latencymon is part of the issue. Also maybe Microsoft .NET Framework has something to do with it)
Attached Thumbnails
Huge latency issues on the latest 1903 Windows release - SOLVED-annotation-2019-06-14-020618.jpg  
Old 14th June 2019
  #92
Here for the gear
 
Avenue Road's Avatar
 

Well, I found it on my machine. Acronis has a service called Managed Machine Service Mini which wakes up every two minutes and presumably makes a kernel call. I've had Acronis on this machine for a long time so it would appear that ntoskrnl on 1903 is not responding to this routine's call as efficiently as before. Fortunately, the service is not important so I've disabled it and the system is now stable, as you can see from LatencyMon.
This is a dual boot with one of my DAW machines so at least I know 1903 can be stable on this hardware. My office system has a lot more non-essential stuff running on it than my DAW.
Attached Thumbnails
Huge latency issues on the latest 1903 Windows release - SOLVED-latencymon_li.jpg  
Old 14th June 2019
  #93
Lives for gear
 
daskeladden's Avatar
 

I don't have Acronis so that not the issue for me. Can you take the test with the updated Latencymon 6.71?
By the way how can you see that something makes a Kernel call?
Old 14th June 2019
  #94
Lives for gear
 
wakestyle's Avatar
1903 finally came and I installed it and subsequent updates a few days ago. Applied all my usual tweaks + declutter. All seems fine but .ntoskrnl.dll randomly spiking for around 750us. (usually all around 100us or less (browsers open, more than one media player open, the computer on all night, etc. typical use-case for me))
Occasionally it is spiking higher to 1700+us, seems random. Even leaving the system idle will get a big spike every 20 min or so, or if using a web browser or watching MPEG it seems to spike more frequently.
As of lately, these spikes don't seem to cause audible glitches in Cubase or other audio streams.

gigabyte z87
nvidia 1050

If I had to guess I'd say it might have to do with memory compression. These spikes only seem to occur 1) when items/menu windows are accessed 2) system is idle and compresses unused objects? haha.

Last edited by wakestyle; 16th June 2019 at 03:54 AM.. Reason: updated some testing info
Old 16th June 2019
  #95
Gear Maniac
 
sosho's Avatar
FWIW: Just went online with my working partition to install some updates . The 1903 is not offered anymore on my system via windows update . 1809 running fine
Old 16th June 2019
  #96
Lives for gear
 
wakestyle's Avatar
So, I tried disabling memory compression and so far, testing has shown my huge latency spikes (1700us+) are gone. I'm still getting periodic 700us spikes but Latmon still validates after 1 hr + now.
-open powershell as admin
-type 'Disable-MMAgent -mc'
-reboot

(to renable use the command 'Enable-MMAgent -mc' instead)

anyone like to try this and see if it helps on their system?

cheers
Old 16th June 2019
  #97
Lives for gear
 
daskeladden's Avatar
 

Anybody tested wakestyle fix?

Another issue that is just for Korg keyboard users. Korg only works on midi between midi1-midi10 and Windows 1903 update ****ed up that order. I tried to reinstall the drivers but it always placed it self on midi11 and midi15. And the rest of the midi was grayed out.



Luckily there is a fix for it:
Old 16th June 2019
  #98
Hi All.

I've been traveling a ton, and completely missed that there was a separate thread on this. Same info I posted on the W10 thread:

Ok. For folks with audible glitches/pops/drops in 1903 that weren't there in 1809, here's the process:
  • Open the Feedback Hub app. Report a problem.
  • Summarize the problem as appropriate. Select "Devices and Drivers" and then "Audio and Sound" for the category.
  • Choose to "recreate my problem" and record audio with glitches present https://matthewvaneerde.wordpress.co...-feedback-hub/
  • Once filed, grab the link from Feedback Hub > Feedback > My feedback and send to me pete dot brown @ microsoft dot com with an obvious audio-related subject.

I appreciate you all helping out here. Thank you.

If you have only latencymon spikes, not audible glitches, that is less actionable (and can sometimes just be misleading). Team is interested in audible defects at this point.

Thanks!

Pete

Last edited by Psychlist1972; 16th June 2019 at 08:19 PM.. Reason: Fix URL
Old 16th June 2019
  #99
Quote:
Originally Posted by wakestyle View Post
So, I tried disabling memory compression and so far, testing has shown my huge latency spikes (1700us+) are gone. I'm still getting periodic 700us spikes but Latmon still validates after 1 hr + now.
-open powershell as admin
-type 'Disable-MMAgent -mc'
-reboot

(to renable use the command 'Enable-MMAgent -mc' instead)

anyone like to try this and see if it helps on their system?

cheers
How much memory does your PC have?

Pete
Old 16th June 2019
  #100
Gear Maniac
 
sosho's Avatar
since the audible glitches I get on my main PC are very small ( maybe I would not hear those in a blind test ) I just did some unscientific fiddeling . Some observations :
The were no new updates but the spikes in latencymon seem to be smaller . Only thing I changed was try O+O Shutup 10 with recommended settings.
And I installed Cubase 10.0.20 on my testpartition , activated the Steinberg power scheme or whatever it is called in english . There were no more spikes in latencymon . I played back 2 demosongs in a cycle while being online via ethernet , 64 samples buffer on an RME fireface UC .
Maybe those latencymon spikes have something to do with powersaving .
If there were any audible glitches it sounded like very light static . With latenymon running.



However there is one thing : The RME PCI card produced totally garbeled audio with a buffer smaller than 128 samples while being online via ethernet on 1903 . On 1809 with the ethernetcable unplugged it went down to 32 samples without any audible problems ( picture 1) maybe some IRQ thing , maybe not
The other screenshots are from the cycle play test on 1903 . I posted some other recent latencymon screenshots before .
edit when cubase is not running
Old 16th June 2019
  #101
Lives for gear
 
norbury brook's Avatar
 

I updated this afternoon and had to re install my Yamaha USB drivers and my Syncrosoft drivers but my macine is running very sweetly on the new update.

Asus P79x-3960X-32 gigs ram- steinberg UR824-Cubase 10.x


M
Old 16th June 2019
  #102
Lives for gear
 
wakestyle's Avatar
Quote:
Originally Posted by Psychlist1972 View Post
How much memory does your PC have?

Pete
16GB on my PC

I'm also seeing similar results on my Laptop (older model) with .ntoskrnl.exe
6 GB ram

and this is only for the most recent cumulative update

update -add pics - systems at idle, nothing running

studio PC -mmagent enabled


studio PC mmagent disabled


laptop -mmagent enabled

Last edited by wakestyle; 17th June 2019 at 03:54 AM.. Reason: adding info + screenshots
Old 16th June 2019
  #103
Lives for gear
 
wakestyle's Avatar
Quote:
Originally Posted by norbury brook View Post
I updated this afternoon and had to re install my Yamaha USB drivers and my Syncrosoft drivers but my macine is running very sweetly on the new update.

Asus P79x-3960X-32 gigs ram- steinberg UR824-Cubase 10.x


M
Did you check with latmon? I thought the same thing until I ran the test for 20 min or so.

cheers
Old 16th June 2019
  #104
Lives for gear
Quote:
Originally Posted by Psychlist1972 View Post
[*]Open the Feedback Hub app. Report a problem.
Is anyone actually reading these? I thought this is something like online petitions where people can vent and move on, but in reality that nothing changes.
I remember using this few years ago and then I realised this was a mockery.
Old 16th June 2019
  #105
I've been running LatencyMon here for a bit on my 1903 system.

The only thing that is causing any high DPCs is resizing Chrome, or moving Chrome windows around. Before that, highest was around 300. Resizing Chrome gives me 1300-1600. Nothing else high right now. Other than that, it was typically < 80, unless moving windows around or something that would pop it up to around 200-300. But in my case, it's definitely the nvidia driver that causes the higher DPCs. I sometimes get pops in Chrome, but haven't had any in the DAW.

I know that's not the case for everyone.

Windows 10 Pro 1903
i9 9900K
64GB RAM
Intel nvme SSD
nvidia 980ti (yeah, it's old)
MOTU 24 i/0 (x3), MOTU 2408mk3 (x1), on PCIe 424

Pete
Old 17th June 2019
  #106
Quote:
Originally Posted by elcct View Post
Is anyone actually reading these? I thought this is something like online petitions where people can vent and move on, but in reality that nothing changes.
I remember using this few years ago and then I realised this was a mockery.
The audio team will if:
  • you capture the glitch trace data as requested
  • Send me the link as requested

I've seen a *ton* of reports make their way into the internal bug tracking. It all depends on how actionable the info is, and how factual it is vs. opinion.

Pete
Old 17th June 2019
  #107
Lives for gear
 
fireberd's Avatar
I have a dual boot with Win 10 Pro and the Windows Insider versions. I've got Pro fairly well cleaned and not showing any Resplendence Latency Mon problems. I recently did a 2 hour recording session and about another hour mixing and mastering and no problems. However, the latest Insider version, which is very minimal and basically just the Insider OS is showing Latency Mon spikes. I reported that on the feedback hub.

As I didn't have any problems with the 2 hour recording session, I have to wonder if Latency Mon is the "problem".
Old 17th June 2019
  #108
Lives for gear
 
norbury brook's Avatar
 

Quote:
Originally Posted by wakestyle View Post
Did you check with latmon? I thought the same thing until I ran the test for 20 min or so.

cheers
I didn't let it run for 20 mins, just 5.

I'll let it run today and see.


To be honest if I'm NOT getting any audio dropouts etc then I don't actually care what Latency mon says :D


I shall report back later.

M
Old 17th June 2019
  #109
Lives for gear
 
daskeladden's Avatar
 

Quote:
Originally Posted by norbury brook View Post
I didn't let it run for 20 mins, just 5.

I'll let it run today and see.


To be honest if I'm NOT getting any audio dropouts etc then I don't actually care what Latency mon says :D


I shall report back later.

M
Also restart pc before running a test. Cause my spikes are tamed to around 400 when pc have been on for about 24 hours without a restart. Could it be that "something" in the OS is sending something to the Kernel for about 24 hours then stops...I don't know. That being said even if the system seems to tame the spikes to around 400 it is way over what I had on the 1809 build.

I have not tested PC mmagent disabled yet cause I'm unsure what it would do with the system. Is this safe, and would it affect other things? I use my pc for audio, video editing and games. I have 32GB of memory (look in signature)
Old 17th June 2019
  #110
Lives for gear
 
norbury brook's Avatar
 

Ok, I fired up this morning and within a few minutes I'd had the NTkernel spike.


I then loaded up a heavy cubase mix I'd be working on lots of tracks/plugins @ 96k and looped it for 20 minutes. Buffer size for this test was 512 samples @ 96k

Smooth as silk no dropouts nada even though you can see Cubase is maxed out.

i then opened a VI project and dropped the buffer down to 32 samples, I never go this low normally but again the project worked fine with regards to no audio glitches etc. i messed with ASIO guard on and off and record armed tracks etc, the cpu was quite high for a small project but @32 samples it ran perfectly.


So, from a personal point of view as long as I'm working trouble free i'm not sure what relevance the latency monitor spikes have.

I shall keep an eye on this thread though.
Attached Thumbnails
Huge latency issues on the latest 1903 Windows release - SOLVED-cubase-latency-test.jpg  
Old 17th June 2019
  #111
Lives for gear
 
daskeladden's Avatar
 

Quote:
Originally Posted by norbury brook View Post

i then opened a VI project and dropped the buffer down to 32 samples, I never go this low normally but again the project worked fine with regards to no audio glitches etc. i messed with ASIO guard on and off and record armed tracks etc, the cpu was quite high for a small project but @32 samples it ran perfectly.
.
Did you still have 96KHz on 32 buffer size? What was your overall latency (input + output) with that settings? Did you use it without Asio guard enabled on that setting without glitches and pops?
Old 17th June 2019
  #112
Lives for gear
 
norbury brook's Avatar
 

Quote:
Originally Posted by daskeladden View Post
Did you still have 96KHz on 32 buffer size? What was your overall latency (input + output) with that settings? Did you use it without Asio guard enabled on that setting without glitches and pops?
no 44.1 @32 samples.


I used ASIO guard on and off it just made the ASIO meter in cubase higher with it off but still no dropouts.



M
Old 17th June 2019
  #113
Lives for gear
 
daskeladden's Avatar
 

Quote:
Originally Posted by norbury brook View Post
no 44.1 @32 samples.

Okay, then I understand. Cause When I try 96KHz with 32 buffer size it is unusable cause it click and pops like hell
Old 17th June 2019
  #114
Lives for gear
 
norbury brook's Avatar
 

Quote:
Originally Posted by daskeladden View Post
Okay, then I understand. Cause When I try 96KHz with 32 buffer size it is unusable cause it click and pops like hell
yes well it would :D that's the equivalent of 16 samples at 48k :D


It's a strange issue for sure. If I hadn't seen this thread I'd be none the wiser as my machine is working great after the update.

Wasn't there an issue with latency monitor when windows 10 first came out? I remember something, it was giving false spikes. Maybe it's the same problem again or are people actually getting audio dropouts?


M
Old 17th June 2019
  #115
Lives for gear
 
fireberd's Avatar
I don't remember if Resplendence Latency Mon was an issue when Win 10 first came out, but the "DPC Latency Checker" program was and it gave false hits. I haven't checked lately but the last time I checked it was not updated for Win 10, although the web site said they were working on it.
Old 17th June 2019
  #116
Quote:
Originally Posted by fireberd View Post
I don't remember if Resplendence Latency Mon was an issue when Win 10 first came out, but the "DPC Latency Checker" program was and it gave false hits. I haven't checked lately but the last time I checked it was not updated for Win 10, although the web site said they were working on it.
It is still not for 8/8.1/10...
Old 17th June 2019
  #117
Lives for gear
 
Pictus's Avatar
 

Quote:
Originally Posted by fireberd View Post
As I didn't have any problems with the 2 hour recording session, I have to wonder if Latency Mon is the "problem".
You can try the Microsoft way, no need for LatencyMon
https://forums.guru3d.com/threads/si...d-isrs.423884/
Old 17th June 2019
  #118
FWIW, I have yet to get any emailed links to Feedback Hub glitch traces (I've posted here, the main Windows 10 thread, and to individuals who have asked me about the "huge latency issues" with 1903).

A reminder that latencymon info is not useful to the audio team (they can often be misleading), but glitch traces are.

Huge latency issues on the latest 1903 Windows release

Pete
Old 17th June 2019
  #119
Gear Maniac
 
sosho's Avatar
Does the problem I mentioned before with the rme audio pci card producing garbeled audio below 128 samples buffer on 1903 ( usb interface does not have that problem ) count as glitch ? I will have to do the same test on 1809 while being online before . Yesterday ( offline) it played with a 32 samples buffer without any problem on 1809. Thank you
Old 18th June 2019
  #120
Quote:
Originally Posted by sosho View Post
Does the problem I mentioned before with the rme audio pci card producing garbeled audio below 128 samples buffer on 1903 ( usb interface does not have that problem ) count as glitch ? I will have to do the same test on 1809 while being online before . Yesterday ( offline) it played with a 32 samples buffer without any problem on 1809. Thank you
Yes, if you can hear it, it counts as a glitch. Please provide that info in the feedback description (what worked on 1809 vs 1903).

Pete
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