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.