Help with latencymon

[mozart999uk]

Hi. I'm getting occasional asio spikes in Cubase and trying to diagnose the problem.

I've been trying to run latencymon whilst cubase is running to see if I can catch one but running latencymon causes Cubase to spike and drop out like crazy. Soon as I stop the monitoring process, cubase returns to normal.

Running DPC latency checker is no problem at all.

Any ideas?

Cubase 9.0.20, Windows 7 x64.

RME 9632

[Pete Kaine]

Yeah, I'd expect this, we'd normally advise against running latency mon alongside your sequencer for this reason. You test the drivers on the system without it running to rule out the drivers, and then if it then happens with Cubase running then you start to examine why it's doing it in the sequencer.

Maybe try running a test project in Reaper to see if it happens in both. If not you can confirm it's a Cubase specific setting or issue.

Possibly related, what CPU are you running in the system?

Maybe try flicking ASIOGuard on/off (depending on current setting) and test too.

[mozart999uk]

Thanks Pete. That's really useful stuff,

Latency mon seems to be running fine without Cubase running. All good figures.

The spikes are a pain as they only happen every now and then.

I think it's something to do with the move from Cubase 8.5 to 9 and also connected with UHE diva. Been chatting to UHE about it too.

The machine is a 3xs audio system, i7 3930 oc'd at 4.2. It's been rock solid since I've bought it 4 years ago. First time anything like this has happened ....

[Pete Kaine]

Yeah, if latency mon is looking fine and it started after an update, I'd be looking to dig further into the software side of things. Do you still have the 8.5 build installed? If so, does it do it on there too?

The reason I was asking about cores is that there is a known multi-core issue that has workarounds, but in this instance, you don't have enough cores in the CPU for that to be a concern.

[mozart999uk]

Thanks for all your advice. I'll try the VST3 suspend option although the plugin in question isn't VST3, I think there are some in the project ....

I haven't had a chance to test it thoroughly in Cubase 8.5 as sometimes it's half an hour before there's a spike.

I'll try and do a longer test.

I'll try asio guard too.

[CS70]

Hi. I'm getting occasional asio spikes in Cubase and trying to diagnose the problem.

Identifying the reason for these spikes is not easy, but it helps being structured and systematic: it could be an hardware issue, a system issue (drivers or other OS behavior), a Cubase issue or a plugin issue. You want to eliminate these one by one.

In random order:

- enter the BIOS and run a full RAM diagnostic check, just in case

- disable any hardware component you don't need, so that your system is as barebone as possible: no network card, no bluetooth radio, nothing else than the minimal set of hardware (and hence drivers) to use the system to produce audio. Certain component's drivers are badly coded and clog the CPU with interrupts, creating real-time processing issues.

- similarly, remove any software component you don't strictly need. Stuff like "call home" updaters can wake up at the wrong moment, triggering system-wide page swapping, garbage collection and other stuff that can momentarily reduce the realtime performance of your system. This is a little harder as there's no unified way to control what gets run, but you can use stuff like Technet's Process Explorer to see what is running and especially what the servicehost processes do (they're quite opaque to the standard tools like taskmanager). All unnecessary services should be disabled, and you can see if anything is started at system start by looking in the registry.

- check if the OS is reporting some issue, using the Event Viewer. Windows has become rather good in reporting what's happening, if you know where to look.

- look at the task scheduler to see if there's any task which is scheduled to run (or did run) when you're working.

- change the DAW and verify if the problem occurs with other DAWs or not (Reaper, free/demo editions of ProTools, Mixbus etc).

- obviously, work on a project without using plugins and see if the spikes keep appearing.

- check your disk for faulty sectors and ensure you have plenty of space, and, if it's a rotating drive, that it's not too fragmented

The fact that the system used to work well is promising - but driver updates, system updates, DAW updates or the wrong plugin can create havoc.

[mozart999uk]

Thanks - that's really really helpful.

There's another issue that's cropped up - the bios has started forgetting it's settings so have replaced the battery. We'll see what happens. Not entirely sure the two are related .....

[mozart999uk]

Haven't had much time to deal with this issue as was so busy leading up to Christmas (managed to limp through trying to ignore the dropouts) but having solved my bios issues with a new battery I'm now looking again at solving the dropouts problem.

I ran latency mon for an hour and it came up with this screenshot.

https://photos.google.com/photo/AF1QipM ... rqZ8Wcp2Y2

I have the text report if that would be useful for anyone

Main driver culprit seems to be:

storport.sys highest execution of 1.094 (ms)

Any thoughts on this one? Google isn't producing any concrete results.....

Going to look at all the other advice from this thread over the next few days and see if any of those tweaks make a difference.

[Matt Houghton]

The image URL appears to have been truncated, and the link is not working...

[mozart999uk]

Sorry about that....seems to be working this end. Might be a google permissions thing?

Here's the link again

https://photos.google.com/photo/AF1QipM ... rqZ8Wcp2Y2

or perhaps this one

https:*?*?*/mD40eR

[Pete Kaine]

Storport is drive subsystem.

Thing is, windows starts to housekeep the system at the 20min mark, and that includes drive optimizing which can easily cause that to flag. That shouldn't happen whilst your inside of Cubase however, unless your jamming across the room for an hour at a time.

What's the test look like at 15mins? I don't really pay much heed to it past that point.

In general chipset driver update would apply here as best practice, although I doubt it'll have much of an impact.

[mozart999uk]

Thanks for your reply.

Here's a shot after 15 minutes

https:*?*?*/icLhtm

I've checked regarding power management. Everything in the power profile is set to 100%. I've turned off all C states in the bios and CPUZ is reporting a constant CPU speed with no throttling take place.....

I've found a bios update, chipset update and LAn update for the MOBO

[job]

Might be worth seeing if Process Explorer can shed some light https://docs.microsoft.com/en-us/sysint ... s-explorer

Run it, right click on the columns at the top and select columns, go to the Process Memory tab and check the boxes for page faults and page fault delta. Click ok. Sort by PF Delta so you know what the highest one is. This will hopefully narrow down your search.

As a troubleshooting step, pull your ethernet lead and/or disable wifi. You never know.

Page faults can be a pain

[CS70]

Thanks for your reply.

Here's a shot after 15 minutes

https:*?*?*/icLhtm

I've checked regarding power management. Everything in the power profile is set to 100%. I've turned off all C states in the bios and CPUZ is reporting a constant CPU speed with no throttling take place.....

I've found a bios update, chipset update and LAn update for the MOBO

There's a bunch of things you can do, but the challenge of that kind of issues is that it's a bit of detective work so it's very time consuming to do it online, and it may require a modicum of understanding of the software environment.

Have you checked for errors in the Event Viewers? Windows has a lot of built-in diagnostics if you know where to look.

Also, a good starting point is download the Windows Performance Toolkit and run the xperf or WPR and produce an etl or etw file to look at.

[mozart999uk]

Thanks for your reply. Appreciate these are hard ones to diagnose remotely,

I'm trying to work out what changed prior to the problem - I've got to look at emails and see if I can work out dates. I know the date of the tracks where I first noticed the problem .....

Have been looking at event viewer over the last few days. I don't get any critical errors but I do get 4 errors repeatedly - WMI, Side by side,Volmgr and service control manager.

I'll try the windows performance toolkit.

Am also checking all the drives for errors .

Is there anything I can run alongside Cubase so that when a spike happens I might be able to see what the problem is?

[CS70]

Out of curiosity, what the recurring errors say?

[mozart999uk]

Thanks again for getting back to me so quickly.

Here's some screenshots

https:*?*?*/dV6j9R
https:*?*?*/eQo9b6
https:*?*?*/i9Q0im
https:*?*?*/cnnfim
https:*?*?*/h2aY3m
https:*?*?*/b93mOm
https:*?*?*/eGmUb6
https:*?*?*/kMWvG6

I think the crash dump error might be related to the fact that I have Virtual memory off. Not sure if I turned that off at some point or if scan.co.uk never enabled it to start with. I have 64GB in the machine and don't get near to using it all up but have plenty of space on the Cdrive (ssd) so happy to enable it .....

[Pete Kaine]

Is there anything I can run alongside Cubase so that when a spike happens I might be able to see what the problem is?

Xperf, as mentioned, is the proper tool for this and Latency mon is a very basic version of it. The problem with both tests and especially Xperf is that it's polling the CPU constantly, and due to that is just as likely to cause false negatives when testing alongside real-time based software (ASIO) where they are fighting for clock cycles.

We wouldn't have disabled virtual memory at the factory as it carries some risk if your not careful, although it isn't one that I disapprove of personally and I do it at home, However, it should only really be used in the event you know you have enough physical memory for it not to fall flat on its face mid-project.

Your errors are interesting too, might be an answer in there.

Side X Side error = really common, C++ libraries have skewed due to multiple installs with other bits of software. - This fix looks correct - https://www.memory-map.co.uk/support/20 ... or-Message

I normally just update to the latest one, and it fixes it, although if it's stubborn you may need to delete a few older ones too.

"ShieldC failed to load" - Driver package not loading at startup. Without knowing what ShieldC is for, however, can't really comment much more.

Application Generation context failed = looks to refer to the old Waves Licence Center software, not starting properly. Looks to be the client that was discontinued back in 2015, worth updating to the replacement client to cure this one.

Event 10 error = https://support.microsoft.com/en-gb/hel ... ll-service although much like the side by side error, it can largely be ignored.

The interesting one here is the volume error. Given DPC says the storage subsystem is causing your spikes, then seeing "VOLmgr error 2" as one of your reports is a bit disconcerting. That would suggest that it is erroring or corrupting somehow and I'd be looking to run some disk diagnostic software on it.

Normally I go with the diagnostic tool from the firm that made the drive, although I guess you've got a section in there and may not be sure of which one it is? If you know, then go to the manufacture website.

If not I'd use something like HDtune (smart check) or Crystal Disk Info to run a quick SMART diagnostic, although there are loads of freeware ones that might be able to do a deeper scan than those two tools if needed. If you suspect it isn't the OS then use "Windows Drive Fitness Test" (WDFT) for a deep scan instead, but it just can't do the drive it resides upon.

[mozart999uk]

Thanks Pete for your reply.

I'll take all that onboard and see what I can sort out.

I've upgraded the drives to large ones since I bought the machine so that may well be it. Mainly Samsung SSDs. One crucial SSD and a WD HD.

I'll have a look at them all

[Pete Kaine]

WD and Samsung both have their own tools, otherwise you might need to use something a bit more generic if you suspect the Crucial as I don't recall them having one.

https://support.wdc.com/downloads.aspx?p=3
http://www.samsung.com/semiconductor/mi ... oad/tools/

[mozart999uk]

Brilliant. Thanks again......

[mozart999uk]

Quick update......

Crystal Disk, samsung magician and WDC data lifeguard all test ok.

Windft found a "smart self test" failure on my crucial BX100 (sample drive). I ran the crucial storage executive software on the bx100 which reported ok. It did say there is a firmware update available for the drive.

I found an app running "intel rapid storage technology". In it was an option for link power management which was enabled. I've disabled that but it's not made a difference.

Latency mon is still giving the same driver storport.sys as the culprit.

In other news:

I've tried enabling the page file - no difference

I've run memtest for 5 hours - all ok.

Updated to latest RME driver

I tried disabling my LAN card but that threw up a load of soft e-licenser errors up. I think mainly from waves.....

I was watching event viewer when a spike happened and it seems googleupdate was trying to run at the same time. I've disabled that in services now but no improvement.

around 2 weeks before I noticed the problems, I had updated from Kontakt 5.6 to 5.7........not sure if that might be anything to do with it.

I'm going now to have a look at the side by side and other errors ......

Was wondering about the updated bios and chipset drivers??? One of them is a sata update.....

[CS70]

Latency mon is still giving the same driver storport.sys as the culprit.

Shots in the dark here without having hands on the system but:

1) is it possible that for some reason Storport's idle power management is enabled? You can check looking in the registry looking at the EnableIdlePowerManagement in Hardware/StorPort. Any nonzero value enables it.

2) It could be that your miniport is experienceing a large number of errors and bus resets. Storport is better than SCSI when facing errors but still may attempt to perform a full bus reset. You can check in Microsoft/Storport in the event viewer.

[mozart999uk]

Thanks CS70.

is it possible that for some reason Storport's idle power management is enabled?

It was. I've changed the value to zero.

You can check in Microsoft/Storport in the event viewer.

I couldn't find this entry in the event viewer. Do you have any tips on where to find it?

I've looked into the volmgr errors in event viewer. "Crash dump initialisation failed" came up- I guess because I don't have a page file active?

[CS70]

I couldn't find this entry in the event viewer. Do you have any tips on where to find it?

I've looked into the volmgr errors in event viewer. "Crash dump initialisation failed" came up- I guess because I don't have a page file active?

Applications and Service Logs/Windows/Microsoft/StorPort