I need some SERIOUS help

[Delta280]

I am at my wit's end.

 

I suppose its best I start from the beginning.  On June 29th, I ordered a new EVGA RTX 2060 XC Gaming, an EVGA Supernova G3 550W power supply, and a Western Digital 3TB 5400 RPM HDD.  I had an HP prebuilt (HP Envy 750-427c http://support.hp.com/ca-en/document/c05257472), but I wanted to upgrade it as it still had some good parts in there that I thought I didn't need to upgrade.  I upgraded from a GTX 750 Ti and a 300 watt power supply.  Upgraded the prebuilt, and all seemed to be going well (except for my wifi card, had to scrap that because new GPU covered up m.2 slot, so I now use a TP-LINK AV1000 powerline adapter).  All seemed to be going well until I noticed that after going away from my computer, and when I would come back after my monitors have turned off (PC is still on) my computer would be strangely slow.  I open up task manager, only to see one of my logical processors (CPU0) was maxed out completely, and this was due to the 'System Interrupts'  process in task manager. 

 

I do have a video of me playing Battlefield V with this CPU issue going on to show how much it impacts my game performance: https://www.youtube.com/watch?v=u-WUU8BDKWM. 

After learning about DPC Latency, I installed LatencyMon to see what driver was causing the issue.  It would be 'dxgkrnl.sys' and 'nvlddmkm.sys' at the top of the list when sorted by total execution (ms).  I've tried absolutely everything to fix this, and it seems that when I roll back to game ready driver version 419.67, the issue seems to stop (I do wish to use the fully up-to-date drivers). All then seemed to be going well, but then I started to notice my CPU was acting up again.  In the 3DMark Firestrike benchmark, I recorded two different results that show that my CPU wasn't reported as fast as it was the first test (first test CPU turbo clock max was 3,995 MHz, second test was 3,794 MHz).  I also started to notice that performance was worse in games like Assassin's Creed Odyssey after this started happening.

First firestrike benchmark (no problems): https://www.3dmark.com/fs/19892126

Second firestrike benchmark (problems): https://www.3dmark.com/fs/19892126

 

After this, I contacted EVGA finally to determine what was the issue, and they said that it sounded like a driver issue, so that I should contact Nvidia instead.  Nvidia told me to install the hotfix driver 431.68, which did not solve the issue.  I also started getting the cpu0 core maxed out issue again. When the cpu0 core was maxed out due to the system interrupts process, I went through device manager to try and disable devices to see which would help, and disabling the RTX 2060 stopped the CPU0 100% usage issue. 

 

Another CPU issue I was experiencing was where my CPU frequency would get strangely locked, and it would stay at my base speed (3.41 GHz).  It didn't change when at idle, or in-game either (Battlefield 4 in this picture).  This happens not

so often, but still another issue I'm experiencing.

 

 

I also, just recently, started experiencing another issue, not sure if this is an issue, but it involves the 'Client Server Runtime Process' and the 'Desktop Window Manager' process.  These processes in task manager were using about 20% of the GPU, and would also increase when I move my mouse cursor around fast, drag a window, or even scrolling on a page.

Update: I also discovered that with google chrome open on my second monitor, there is more GPU usage on desktop window manager than on my main monitor, perhaps its because I am using an adapter to connect the monitor to my GPU?

 

I have no clue why all of these issues are happening, and it's been over a month since they started happening. If ANYONE can help I would be extremely thankful.

 

Things I've tried in an attempt to fix the issues:

-Updating BIOS

-Updating drivers (chipset, graphics, ethernet, sound card)

-Using DDU to reinstall graphics card drivers

-Disabling devices in device manager (disabling the GPU fixed the cpu core maxed out issue)

-Undervolting CPU

-Increasing power limit on GPU to 113% (MSI Afterburner)

-Setting GPU to prefer Maximum performance in Nvidia Control Panel

-Power plans High performance and Ultimate Performance

-Disabled Fast Startup

-

 

My specs:

Processor: Intel Core i7-6700 @ 3.4 GHz
Graphics Card: EVGA RTX 2060 6GB XC Gaming
Motherboard: Odense2-S (HP 828A)
Ram: 8x2GB SK Hynix DDR4 @ 2,128 MHz
Storage: Seagate - 1 TB 7200 RPM & Western Digital - 3 TB 5400 RPM
Power Supply: EVGA Supernova G3 550W 
Case: HP ENVY 750-427c Chassis

Mouse: Logitech G502
Keyboard: Redragon Kumara K552

 

Attached are pictures and text files that show some of the issues I'm experiencing (task manager, latencymon results, DPC Latency Checker, etc.)

latencymon results.txt

LatencyMon results without CPU0 maxed out.txt

 

If any other information is needed just let me know!

 

 

---------------------------------------------------------------------------

 

---------------------------------------------------------------------------

 

---------------------------------------------------------------------------

[homeap5]

At first try to check if any 3rd party software is running in background. Use Autoruns for that.

 

You have some crappy prebuilt mobo I guess, so it may be also problem with pcie power. Try to underclock your gpu using MSI Afterburner (both - clock and memory - to the left using sliders) and decrease maximum power usage to 80% for tests.

[Delta280]

23 minutes ago, homeap5 said:

At first try to check if any 3rd party software is running in background. Use Autoruns for that.

 

You have some crappy prebuilt mobo I guess, so it may be also problem with pcie power. Try to underclock your gpu using MSI Afterburner (both - clock and memory - to the left using sliders) and increase maximum power usage to 80% for tests.

For Autoruns, how would I check for third party software?  Are those the selections that are highlighted in yellow?

 

Also I reduced the power usage for my GPU to 80% and underclocked memory and clock, but still no fix.  I just discovered that when I close out of google chrome (with lots of tabs open) and just sit at the desktop, my GPU usage will actually go up higher than with google chrome open.

Edited August 3, 2019 by Delta280

[homeap5]

34 minutes ago, Delta280 said:

For Autoruns, how would I check for third party software?  Are those the selections that are highlighted in yellow?

 

Also I reduced the power usage for my GPU to 80% and underclocked memory and clock, but still no fix.  I just discovered that when I close out of google chrome (with lots of tabs open) and just sit at the desktop, my GPU usage will actually go up higher than with google chrome open.

You must analyze list in Autoruns by yourself. If you have no experience - made screenshots.

 

Your computer starts to do something when idle - it's a little suspicious, so maybe run AdwCleaner.

[phoenixflower]

What I would do is get a spare USB Drive or better yet a spare SSD of some sort.  Install Pop OS on it and run games through Lutris/Steam.  If you don't experience any studdering then it's a Windows 10 scheduling issue.  Sometimes Drivers can get mismatched in the registry and do silly things.

 

If you want to go on the advise of my experience instead of testing, it smells like a Windows 10 scheduling problem and you should just skip to re-installing Windows and go through the process described below

 

Moving your mouse really fast should have a CPU load delta of less than 1%.

 

 

If it is a Windows 10 scheduling problem

backup your data and reinstall Windows 10

wait for Windows 10 to i install the graphics driver for you

download Geforce experience and let it update the drivers for you

 

If it is not a Windows 10 scheduling problem scroll to the bottom

 

As pointed up above it could be a malware related problem which would impact Windows 10 scheduling

 

 

Do not access the internet in your web browser or use web services until Windows 10 has gone through it's full update cycle after re-installing it.

 

After all of this had been done you can download EVGA-Precision X1 or MSI-Afterburner to work with your graphics card.

After re-installing Windows 10 you should buy a copy of ESET and install it (it has a video game mode for online video games)

be sure to install https everywhere, your favorite ad-blocker, privacy badger and a web-rtc blocker in your favorite browsers.

 

If it is not a Windows 10 scheduling problem

Depending on how old your PC is, you may need to replace the thermal paste on your CPU

It may be a broken/dying USB port.  Sometimes the interrupt manager on Windows will overcompensate for bad hardware

You can test this by simply not allowing USB devices to go to sleep -> you can accomplish this by going into your power plan (which is hopefully performance) and turn off all the power saving features in advanced settings.  You can either enter the number 0 or enter the word "never" to indicate sleep mode should be de-activated for your peripherals.

If you have another PCI-E interface, it may be a dying interface.  Try swapping it around.

 

It hope some of these tips help

[Delta280]

 

 

 

 

 

 

 

 

 

 

8

25 minutes ago, phoenixflower said:

 

 

 

 

 

 

 

 

 

8 minutes ago, phoenixflower said:

What I would do is get a spare USB Drive or better yet a spare SSD of some sort.  Install Pop OS on it and run games through Lutris/Steam.  If you don't experience any studdering then it's a Windows 10 scheduling issue.  Sometimes Drivers can get mismatched in the registry and do silly things.

 

If you want to go on the advise of my experience instead of testing, it smells like a Windows 10 scheduling problem and you should just skip to re-installing Windows and go through the process described below

  

Moving your mouse really fast should have a CPU load delta of less than 1%.

 

 

If it is a Windows 10 scheduling problem

backup your data and reinstall Windows 10

wait for Windows 10 to i install the graphics driver for you

download Geforce experience and let it update the drivers for you

 

If it is not a Windows 10 scheduling problem scroll to the bottom

 

As pointed up above it could be a malware related problem which would impact Windows 10 scheduling

 

 

Do not access the internet in your web browser or use web services until Windows 10 has gone through it's full update cycle after re-installing it.

 

After all of this had been done you can download EVGA-Precision X1 or MSI-Afterburner to work with your graphics card.

After re-installing Windows 10 you should buy a copy of ESET and install it (it has a video game mode for online video games)

be sure to install https everywhere, your favorite ad-blocker, privacy badger and a web-rtc blocker in your favorite browsers.

 

If it is not a Windows 10 scheduling problem

Depending on how old your PC is, you may need to replace the thermal paste on your CPU

It may be a broken/dying USB port.  Sometimes the interrupt manager on Windows will overcompensate for bad hardware

You can test this by simply not allowing USB devices to go to sleep -> you can accomplish this by going into your power plan (which is hopefully performance) and turn off all the power saving features in advanced settings.  You can either enter the number 0 or enter the word "never" to indicate sleep mode should be de-activated for your peripherals.

If you have another PCI-E interface, it may be a dying interface.  Try swapping it around.

 

It hope some of these tips help

Thanks for all of the tips!  This forum thread so far has been more help than any other platform I've used.  For the mouse load for CPU, it's actually not really noticable, only for the GPU, which usually makes my CPU go up to 5% overall utilization.  Not sure if this is normal, but I tested it on an identical prebuilt (two of the same HP systems in my household, mine is the upgraded one), and it showed similar results that GPU usage would go up.  This issue doesn't concern me as much, but I'm more concerned about the CPU0 core going to 100% due to system interrupts.  I upgraded Windows 10 to version 1903, and I have yet to see if that problem will resurface.  I've also done scanning on many different antiviruses (don't worry I keep only one enabled at a time), which show my system isn't infected.  Also, I forgot to mention in my original post that I've done sfc /scannow in an elevated command prompt, and the results showed nothing was corrupt.  And another thing I noticed is that sometimes when I hear Steam notifications (chat msgs, going online, etc.) there is a bit of static.  But if it comes back yeah I might have to reinstall windows.  It's strange though given that this issue doesn't happen on older Nvidia drivers, but on newer ones it does (in windows 10 v1803 anyways).  My 3 year old board only has 1 PCI-E interface for a GPU, so I wouldn't be able to change PCI-E slots.

1 hour ago, homeap5 said:

 

 

 

 

1 hour ago, homeap5 said:

You must analyze list in Autoruns by yourself. If you have no experience - made screenshots.

 

Your computer starts to do something when idle - it's a little suspicious, so maybe run AdwCleaner.

Nothing looked out of the ordinary in Autoruns, and I ran Malwarebytes Free and AdwCleaner, both came up with no results (except for some HP bloatware that came with my PC, deleted some of them but they've been with my PC since I got it three years ago and I'm guessing that isn't the problem).

[phoenixflower]

Most video cards pre-pascal are very well understood driver wise.  Pascal cards and RTX cards are still attempting to be understood at the driver level in the various installation environments.  One of the things that NVIDIA really has dropped the ball on is what type of interrupt they chose for their Pascal onwards graphics cards.  Requiring legacy style interrupts instead of favoring more modern interrupt style techniques.  This directly impacts the interrupt tables for the various operating systems and in many regards can be a point of contention as to why Windows 10 simply runs better behind a Linux Kernel Virtual Machine (KVM) then on bare metal Ryzen platforms.  This engineering choice probably caused by a bottom line problem can also impact how your Intel based CPU will interact with your GPU.

 

If you experience high CPU load/GPU load 5-10 minutes after entering what you would consider being in an idle state.  Then you have malware, or you have a driver that is acting like malware.

 

If you want to look into the interrupt thing (this is a very techy topic).  Sun Microsystem has graced us with a powerpoint presentation from 2006 regarding this very relevant topic in 2019 - http://community.qnx.com/sf/sfmain/do/downloadAttachment/projects.community/discussion.community.topc21944/post93964;jsessionid=C7505E29A4E991B7F42ED649A13F90FC?id=atch11922

 

One thing that you definitely keep in mind is that a lot of those scanners are fingerprint based.  Which means that if you were unlucky enough to get malware that doesn't conform to a registered fingerprint the anti-virus and malware scanners (they are different) will simply not detect it.  I usually recommend ESET because they have HIPS (Heuristic Intrusion Prevention System).  Instead of looking at the fingerprint of malware it looks at the actions of your executables and decides if it's malicious or not on top of having all of the fingerprints.  Windows 10 is a pretty attractive ecosystem for malware and viruses due to the number of systems that depend on Windows 10, Windows Server 2016, and Windows Server 2019.

 

 

I'm glad your problem resolved itself with a Windows update, most Windows based problems are resolved with updating after a recent hardware change/re-installing.  Feel free to come back anytime

 

 

[Delta280]

2 hours ago, phoenixflower said:

Most video cards pre-pascal are very well understood driver wise.  Pascal cards and RTX cards are still attempting to be understood at the driver level in the various installation environments.  One of the things that NVIDIA really has dropped the ball on is what type of interrupt they chose for their Pascal onwards graphics cards.  Requiring legacy style interrupts instead of favoring more modern interrupt style techniques.  This directly impacts the interrupt tables for the various operating systems and in many regards can be a point of contention as to why Windows 10 simply runs better behind a Linux Kernel Virtual Machine (KVM) then on bare metal Ryzen platforms.  This engineering choice probably caused by a bottom line problem can also impact how your Intel based CPU will interact with your GPU.

 

If you experience high CPU load/GPU load 5-10 minutes after entering what you would consider being in an idle state.  Then you have malware, or you have a driver that is acting like malware.

  

If you want to look into the interrupt thing (this is a very techy topic).  Sun Microsystem has graced us with a powerpoint presentation from 2006 regarding this very relevant topic in 2019 - http://community.qnx.com/sf/sfmain/do/downloadAttachment/projects.community/discussion.community.topc21944/post93964;jsessionid=C7505E29A4E991B7F42ED649A13F90FC?id=atch11922

 

One thing that you definitely keep in mind is that a lot of those scanners are fingerprint based.  Which means that if you were unlucky enough to get malware that doesn't conform to a registered fingerprint the anti-virus and malware scanners (they are different) will simply not detect it.  I usually recommend ESET because they have HIPS (Heuristic Intrusion Prevention System).  Instead of looking at the fingerprint of malware it looks at the actions of your executables and decides if it's malicious or not on top of having all of the fingerprints.  Windows 10 is a pretty attractive ecosystem for malware and viruses due to the number of systems that depend on Windows 10, Windows Server 2016, and Windows Server 2019.

 

 

I'm glad your problem resolved itself with a Windows update, most Windows based problems are resolved with updating after a recent hardware change/re-installing.  Feel free to come back anytime

 

 

Alright, after some further testing there still appears to be some stutters/hitches every now and then (Battlefield 3, 4, and V) where the game freezes for about a second or less.  Not as bad as it was but it still is impacting gameplay.  I think I'm going to go ahead with reinstalling Windows 10, but would I want to factory reset or install a fresh image of Windows completely?  This was originally an HP system and I'm not sure with what route to go with.

[phoenixflower]

I would first backup my data either locally or to something like backblaze.  Then I would call HP regarding re-installing Windows 10.  You may need their help activating Windows as they used an OEM key from Microsoft.  Sometimes the key appears as a sticker in the manual for your computer or on a disc that comes with your computer.  If you lost the sticker or didn't not receive one then HP can help you re-activate Windows 10.  After you've successfully activated after a fresh install I would recommend the above course of action in the previous post in regards to re-installing Windows 10.

 

Re-installing is your best bet as reset and refresh are known to simply not work.  Windows will do Windows things and sometimes changing hardware will confuse it.

 

For actually re-installing it, you can use the windows creation tool from Microsoft as that's exactly what HP used to provision your computer.  Again, you may need HP's help in getting chipset drivers or drivers related to your motherboard and non-commodity hardware.

 

If there's anything else that's confusing I'd be happy to help

[Delta280]

5 minutes ago, phoenixflower said:

I would first backup my data either locally or to something like backblaze.  Then I would call HP regarding re-installing Windows 10.  You may need their help activating Windows as they used an OEM key from Microsoft.  Sometimes the key appears as a sticker in the manual for your computer or on a disc that comes with your computer.  If you lost the sticker or didn't not receive one then HP can help you re-activate Windows 10.  After you've successfully activated after a fresh install I would recommend the above course of action in the previous post in regards to re-installing Windows 10.

 

Re-installing is your best bet as reset and refresh are known to simply not work.  Windows will do Windows things and sometimes changing hardware will confuse it.

 

For actually re-installing it, you can use the windows creation tool from Microsoft as that's exactly what HP used to provision your computer.  Again, you may need HP's help in getting chipset drivers or drivers related to your motherboard and non-commodity hardware.

 

If there's anything else that's confusing I'd be happy to help

I linked my Windows 10 license to my Microsoft account so I shouldn't have any problems with that.  I do have two hard drives, should I clear both of them and then install windows?  Or just clear the one with windows on it? Or don't clear them at all?  I'm not sure how the whole process works for it, and even when I do reinstall windows would my BIOS software get cleared if I wipe my hard drive?  Thanks again for all of your help!

[phoenixflower]

Okay, licensing Windows isn't a big deal because either way your machine is licensed because your computer came with an OEM license.  If the Windows "needs activation" watermark appears then just be sure to call HP in regards to your Windows 10 licensing and mention you need to do a fresh install because of hardware problems.  The staff will be very helpful.

 

When replacing components on your PC/opening your PC to remove or add parts, be sure to fleece the power supply.  This means completely shutting down the PC (wait for the fans to stop spinning), and then put your power supply in the off position.  After you have the power supply in the off posiition.  Hold the power on button for 30 seconds.  That means 1 one-thousand, 2-one-thousand, 3 one-thousand,. . .,30 one-thousand.  Be sure to keep the power cord attached to the wall during the process and ensure your power supply is in the off position.

 

Now you can start replacing parts.  fleecing the power supply helps drain your system of all the power and reduces the overall chance of damage due to static electricity.  Also be sure to keep yourself ground by keeping one hand on the case, or get an anti-electrostatic wristband.

 

Well, in this case because malware is suspected and you haven't had it inspected by an IT team to verify it's clear of any virus/malware.  I would wipe both drives.  When you install Windows be sure to only have one drive in the system as Windows will do some really convoluted things in a multi-drive scenario.  After you have Windows 10 installed.  Shutdown the PC, and then re-introduce the second drive to the system.

 

After you have re-introduced the second drive be sure to follow the above steps in the previous post in regards to installation procedure.

 

 

Fortunately your BIOS is installed on independent Flash on your motherboard.  Wiping your drives will have no impact on the health of your motherboard or in particular, your BIOS.

 

and of course!  Computers can be tricky sometimes, just be sure to pay it forward   

[Delta280]

Alright sounds good! I will have to try these out later when I have more time, do you have a discord or something I could contact you by in case I need more help?

[phoenixflower]

https://discord.gg/PnK7gq