Discussion in 'Operating Systems' started by Extraordinary, Mar 28, 2019.
Guess I'll still stick to 1803 for the time being until the latency spike issue is fixed.
yeah. maybe we'll find something in future KB patchnotes for 1903.
furthermore: 1903 Intel LAN drivers (24.1+) aren't available yet for my I211
might be one of my culprits, since I only ran firefox + music @youtube.
new chipset drivers will hopefully launch around zen2 release
Yep that's the one
so does that mean the latency spikes were fixed by a microcode? atlest for intel??
Has been for me. Can't speak for Ryzen, sorry.
seeing as my pc is intel 6700k and my uncles are i5 3470 and i5 8400 I hope this true, as I really do want to put his pc the current build cause it might be other year or more before they see updates again seeing that stuff is always ignored by him. And I think I might try my hand at how a inplace upgrade works for me.
the inplace update on one my uncles pc worked right for change
I have even newer then in this update and still get it.
Here it's 0×25, I have 0×27
So the problem with the microcode fixed it for me I see going around is that most people only run lat mon for 5-20 minutes and see no spikes and call it good and I get it. The weird issue with this kernel spike issue is it comes in waves randomly. I could go 2 hours and not see it then it happens bam like a punch to your pcs chest. To get a real good idea to see if its finally fixed you have to run it for an hour to two just using ur pc. I have also when I was on 1903 some reboots the notoskrnl spikes would only bee 600us or so which isnt good but not bad. Other reboots I would see the 1500us ones. Its a very odd issue that still I can confirm is not fixed yet. But on 18922 it is. So they just have to backport the fix which hopefully is in the next cumulative update. If not ill be pissed off.
Oh, great, so it doesn't apply to my 2 machines that are affected? 2500k and 4770k.
This is interesting, but I don't want to move to an insider ring just yet.
second gen core i isn't affected. (in my testing)
if you suffer from ntoskrnl spikes... please upvote
New update out https://www.tenforums.com/windows-1...ndows-10-v1903-build-18362-207-june-24-a.html
Can anyone with the ntoskrnl spike issues see if this fixes it?
That's a big one, seems to be insider channel from what I can find and made some changes although I am not sure if there's any specific info and these updates don't usually have a lot on them if it's insider only.
Lots of file changes and took a long time to install so it's doing something, will be interesting to see what the update notes for next month will say although I assume it'll primarily focus on various security aspects. Hmm.
Since you installed kind sir can you run latencymon for about an hour as u do your usual business and see if u get any spikes?
I could see sure, system is kinda bottlenecked but if there is anything out of the usual then yeah why not it still needs to do a bunch of tasks after rebooting so I can leave it doing it's thing for an hour.
The NT OS Kernel hitting a higher value I assume? Yeah it only took a few seconds for that to hit to about half of the total bar length and overtake the network card as the highest reported. (Steam queued up another update but even that doesn't seem to be a problem and the kernel overtook that easily.) will leave it running for a while but I assume this is the issue.
EDIT: Guessing a future build of Windows 10 might take care of it assuming it gets the traction going for reported issues in 20H1 if it's still happening there, would be nice if a cumulative update could solve it for the current affected build(s) but honestly from Threshold to Redstone to 19H1 and 20H1 now I'm not expecting much though I would like to see the issue rectified of course but it doesn't seem to be how Microsoft does things unless it's incredibly critical which granted this seems like a serious problem but not a complete stopper or a massive security hole though this latency and delay probably carries consequences too which if it's still a thing since the OS release (19H1) then that just shows Microsoft isn't prioritizing it as a major concern and might see it resolved in a future build.
Though I might be a bit pessimistic (Realistic as I see it unfortunately.) but we will see though I assume from the results I am currently seeing it is still a concern as of this update even if this is a older system with various faults and flaws and slower areas compared to a modern machine.
EDIT: Or to be clear I find it to be a poor show that it is still a ongoing issue months after 19H1 got finalized and then it's full launch and several cumulative updates in but it doesn't seem like it's a big priority for Microsoft and what we've seen so far unless it's almost a critical problem it's just going to end up as a eventual fix in a later build though there is a chance some future update could also resolve the latency problem and it's consequences but I am not expecting much even if it should be hotfixed sooner rather than made some later priority or fixed as part of the ongoing Windows 10 service model and gradual build releases either for 19H2 when or if that comes out or 20H1 or possibly even later as I recall hearing about a latency issue in 20H1 and it's builds too but if this catches traction and gets upvotes and such on the feedback hub it could get pushed forward. (Seriously though? A popularity contest for OS bugs?)
yeah so ntoskrnl.exe still spiking to over 1300+? i cri
1300 as the norm? I was up to around 1900.xxx - 2000.xxx but then the ongoing Steam game update could have hit things a bit and it's going to be a while before it gets done due to it's size.
(Patching existing files after all, delta similar to these cumulatives in a way so with a 17 GB patch for Wolcen replacing many of the files that means it's HDD hammer time.)
EDIT: But that's I/O and it overtook the NIC and it leading for the first few seconds so even if it causes some general slowndown as the HDD is having full up dealing with this it might not affect the DPC latency value but yeah it's a older machine too.
In any case there is a noticeable amount of DPC latency and a big spike from the NT OS Kernel here.
EDIT: Yep, Steam is settling down but the latency is still sticking so it's a thing alright.
EDIT: Well a measured value of 1300 somewhere as the normal for the latency spike still being in effect that is just to clarify, bit higher on my end possibly just due to hardware but well it's definitively still there.
Yeah so they didnt fix the issue ty sir for checking very sad microsoft can fix these spikes causes audio pops and microstutter when it happens
Got same results as Jonas mentioned.
I got it fixed by running ISLC with Wanted timer resolution set to 1.00
After that I played Division 2 for 90 minutes and was highest values was around 600-800.
Can anyone explain how this is possible? is it timer resolution that's doing it?