The hardest microstutter to troubleshoot

[mpw90]

I'm essentially encountering stuttering in games and some applications, despite my temperatures being fine and low usage of the components.

 

Prior to all of this, I have had difficulties with the system since I purchased the parts, and built it. I have had 3 motherboard replacements due to hard freezing, which appears to have been a PCIe lane problem from PCIe Gen 3 on the board. This is now resolved.

 

I have also had a problem with my GPU throttling due to heat problems, this was replaced and considered a faulty part.

 

Description of Problem:

On every setup I have tried, games will encounter some form of short micro stutter. This lasts a short amount of time, and occurs at seemingly unknown* intervals. It can happen in a series of stutters, or not happen for over one hour. Once it begins, it is very noticable and prominent. The stutter will last approximately between 250ms to 1 second. I observe that the frametime will often spike, but not always.

 

This was tested on both Linux and Windows 10, on separate drives, with fresh installs, over multiple disks. This has been tested under 'diagnostic' conditions, whereby I have installed and updated my drivers, and disabled anything that is not required, nor unrelated to the task at hand.

 

There are no overclocks, though I have tested with stock speeds at higher voltage. And tested with overclocks. If anything, an overclock will help. Though, I am keeping it removed. This began before any overclocking took place.

 

I have attempted the following:

 

* Replaced components

* Reinstalling the OS (and trying Linux, on different HDD's/SSD's, and removing them when other drive is in use)

* Flashing BIOS

* AMD chipset drivers

* Updating Nvidia drivers (and removing old ones with DDU)

* Reseating components

* Removing peripherals and components until the bare essentials

* Resetting CMOS

* Reducing GPU power target

* Checking VRM temps

* Disabling drivers, and keeping the bare esential drivers required to run

* Checking LatencyMon -- sometimes ntoskrnl will spike, sometimes it wont. Sometimes I will have a high ISR latency, sometimes I wont.

* Monitoring voltage

* 3 different monitors, cables and display interfaces (DVI, HDMI and displayport)

* Disabled and uninstalled monitoring software

* Reduced graphics settings in both Nvidia Control Panel and games themselves (remember, it happens in some applications, too) -- VSync off and on

* Using Prime95 for stability at stock settings -- passes all torture tests

* Tried different power plans, with different CPU % settings

* Purchased a PSU tester, and it passes all the tests.

 

I also disable unnecessary services and processes on both operating systems.

 

I've replaced the board, the GPU, and almost every other component. I've removed the network card and disabled the wifi and bluetooth drivers. My temperatures are good. To the point where Prime95 will max out around 78C on the test that produces the most heat. The GPU rarely goes above 70C unless in stress testing, and averages at around 61C during load. CPU averages around 58C during load.

 

There appears to be no throttling, or no obvious sign in monitoring software when this takes place.

 

The voltages appear to be fine, and hit the target voltages.

 

I do not know what to monitor anymore. I am kind of at a loss for where to look.

 

Clearing 'Standby Memory' on Windows appears to help for a short period of time, so I would have assumed disk/ram. Though, yesterday when I 'undervolted' (reduced power target) my GPU, it also helped for a long period of time. This does not always work, though. Sometimes I can remove the stanby memory and it will still happen, just less so. I cannot find the cause.

 

Component List

 

CPU: AMD - Ryzen 5 1600 3.2GHz 6-Core Processor

 

Motherboard: ASRock - AB350M Pro4 Micro ATX AM4 Motherboard

 

Memory: Corsair - Vengeance LPX 8GB DDR4-2400 Memory (2 x 8GB) (16GB)

 

Storage: Crucial - MX300 275GB 2.5" Solid State Drive

 

Storage: Seagate - BarraCuda 1TB 3.5" 7200RPM Internal Hard Drive

 

Video Card: EVGA - GeForce GTX 1060 6GB 6GB SC GAMING Video Card

 

Case: Thermaltake - Core V21 MicroATX Mini Tower Case

 

Power Supply: Corsair - TXM Gold 550W 80+ Gold Certified Semi-Modular ATX Power Supply

 

Wireless Network Adapter: Gigabyte - GC-WB867D-I PCI-Express x1 802.11a/b/g/n/ac Wi-Fi Adapter

 

Monitors: Acer - XF240H 24.0" 1920x1080 144Hz Monitor x 2

 

https://uk.pcpartpicker.com/list/p9mKsZ

Edited August 28, 2018 by mpw90

[Corsair Art]

Conduct a memtestx86 and let me know if you run into issues. Might just be faulty even though it passed stress testing.

 

-Art

[mpw90]

Conduct a memtestx86 and let me know if you run into issues. Might just be faulty even though it passed stress testing.

 

-Art

 

UPDATE

I secured the GPU in PCIe Gen 2 slot x16. Not the main PCIe Gen 3 slot. I loaded CS:GO and at splash screen, I noticed an increase of 120 fps (yes, I am not joking, what the hell?) I was hitting around 550-580 fps, when usually it is the low 400s. Even when I loaded the game, it reduced, but significantly higher in FPS.

 

 

 

YET, the stutters occured, and then I had a huge sequence of them and the system hard locked. This is exactly the same symptom I had with this board previously but with the PCIe Gen 3 slot. Previously, the Gen 2 slot was the only slot that would work.

 

 

 

Now, I am pretty convinced it is still the board. I have seen so many other horror stories about this board, and this is my 4th board after 3 RMA's in 6 months. Even after them asking me to change CPU (which I did), changing my GPU 3 times (1 time due to EVGA confirming a fault, and another testing an AMD GPU), and trying single sticks of RAM, etc.

 

 

 

Is it best to use Occam's Razor here and return this board for a final time, and change manufacturer? There is a chance it is the CPU. There is a chance it is the memory.

[Corsair Art]

UPDATE

I secured the GPU in PCIe Gen 2 slot x16. Not the main PCIe Gen 3 slot. I loaded CS:GO and at splash screen, I noticed an increase of 120 fps (yes, I am not joking, what the hell?) I was hitting around 550-580 fps, when usually it is the low 400s. Even when I loaded the game, it reduced, but significantly higher in FPS.

 

 

 

YET, the stutters occured, and then I had a huge sequence of them and the system hard locked. This is exactly the same symptom I had with this board previously but with the PCIe Gen 3 slot. Previously, the Gen 2 slot was the only slot that would work.

 

 

 

Now, I am pretty convinced it is still the board. I have seen so many other horror stories about this board, and this is my 4th board after 3 RMA's in 6 months. Even after them asking me to change CPU (which I did), changing my GPU 3 times (1 time due to EVGA confirming a fault, and another testing an AMD GPU), and trying single sticks of RAM, etc.

 

 

 

Is it best to use Occam's Razor here and return this board for a final time, and change manufacturer? There is a chance it is the CPU. There is a chance it is the memory.

 

With how many times it's failed you, I personally would RMA and sell or put away and get a new one.

 

-Art

[mpw90]

With how many times it's failed you, I personally would RMA and sell or put away and get a new one.

 

-Art

 

Indeed, I have just ordered a new motherboard. As I am only 99.999% sure that this is indeed the cause, I would like to keep this thread open and potentially update it should the stutter continue, and this freezing be a separate issue (though, I don't think that is the case)

[mpw90]

Changed RAM, changed SSD. Same motherboard. Far more stable, less stuttering, but it still exists. I cloned the SSD (so, it's not entirely foolproof plan).

 

I plan to next test the motherboard. If it is not the motherboard, then it is either the PSU or the CPU.

[mpw90]

As can be seen here: https://i.imgur.com/G3QbAsq.png -- the spikes are still happening, but to a lesser degree. This is after a new SSD and new RAM.

[mpw90]

As per this thread: https://forums.evga.com/The-hardest-stutter-to-troubleshoot-m2850576-p2.aspx#2852897 -- I have attempted to run the kill ryzen script to detect hardware faults. I do not know if there are false positives with this, but I do get the error.

 

 

 

C:\Users\mpw90\Desktop\Stutter Solving\kill-ryzen-win-master>bin\x64\Release\kill-ryzen-win || exit /b

 

bzip2.c

 

bzip2.c

 

bzip2.c

 

bzip2.c

 

bzip2.c

 

bzip2.c

 

bzip2.c

 

bzip2.c

 

bzip2.c

 

bzip2.c

 

bzip2.c

 

bzip2.c

 

bzip2.c

 

bzip2.c

 

bzip2.c

 

bzip2.c

 

bzip2.c

 

bzip2.c

 

bzip2.c

 

bzip2.c

 

bzip2.c

 

bzip2.c

 

bzip2.c

 

bzip2.c

 

Unhandled Exception: System.AggregateException: One or more errors occurred. ---> System.Exception: FAIL

 

at kill_ryzen_win.Program.<>c__DisplayClass0_0.<Main>b__0(Int32 x) in C:\Users\mpw90\Desktop\Stutter Solving\kill-ryzen-win-master\Program.cs:line 43

 

at System.Threading.Tasks.Parallel.<>c__DisplayClass17_0`1.<ForWorker>b__1()

 

at System.Threading.Tasks.Task.InnerInvokeWithArg(Task childTask)

 

at System.Threading.Tasks.Task.<>c__DisplayClass176_0.<ExecuteSelfReplicating>b__0(Object )

 

--- End of inner exception stack trace ---

 

at System.Threading.Tasks.Task.ThrowIfExceptional(Boolean includeTaskCanceledExceptions)

 

at System.Threading.Tasks.Task.Wait(Int32 millisecondsTimeout, CancellationToken cancellationToken)

 

at System.Threading.Tasks.Parallel.ForWorker[TLocal](Int32 fromInclusive, Int32 toExclusive, ParallelOptions parallelOptions, Action`1 body, Action`2 bodyWithState, Func`4 bodyWithLocal, Func`1 localInit, Action`1 localFinally)

 

at System.Threading.Tasks.Parallel.For(Int32 fromInclusive, Int32 toExclusive, Action`1 body)

 

at kill_ryzen_win.Program.Main(String[] args) in C:\Users\mpw90\Desktop\Stutter Solving\kill-ryzen-win-master\Program.cs:line 51

[mpw90]

Okay, as much as it pains me to say, I don't think it's the ASRock AB350m Pro4. Though, thankfully it's not. After 4 RMA's!

 

I tried the Gigabyte AB350M-DS3H and the stutters were still occurring.

 

New motherboard frametime spikes 1 - https://i.imgur.com/WLzCW2z.png

New motherboard frametime spieks 2 - https://i.imgur.com/1zkMAHI.png

 

So, I am beginning to think this *is* the CPU.

 

So, I am at a loss still.. This CPU was manufactured in Malysia week 42 of 2017.

 

If indeed this is an AMD issue, I would actually be kind of furious. This has been 6 months of struggling to find the cause. Whilst I would be relieved to discover it is indeed the cause (what else can it be at this point), it would be my second CPU, 4th motherboard, 3rd lot of RAM, 2nd GPU, 2nd SSD, and 4 or 5 installation of Windows, and maybe 3rd installation of Linux.

 

Not to mention all the time consumed updating drivers, BIOS, disabling services, plugging and unplugging cables, and using the screw driver, and thermal paste, etc.

[Corsair Art]

Okay, as much as it pains me to say, I don't think it's the ASRock AB350m Pro4. Though, thankfully it's not. After 4 RMA's!

 

I tried the Gigabyte AB350M-DS3H and the stutters were still occurring.

 

New motherboard frametime spikes 1 - https://i.imgur.com/WLzCW2z.png

New motherboard frametime spieks 2 - https://i.imgur.com/1zkMAHI.png

 

So, I am beginning to think this *is* the CPU.

 

So, I am at a loss still.. This CPU was manufactured in Malysia week 42 of 2017.

 

If indeed this is an AMD issue, I would actually be kind of furious. This has been 6 months of struggling to find the cause. Whilst I would be relieved to discover it is indeed the cause (what else can it be at this point), it would be my second CPU, 4th motherboard, 3rd lot of RAM, 2nd GPU, 2nd SSD, and 4 or 5 installation of Windows, and maybe 3rd installation of Linux.

 

Not to mention all the time consumed updating drivers, BIOS, disabling services, plugging and unplugging cables, and using the screw driver, and thermal paste, etc.

 

Please keep in mind we don't mind replacing this memory kit if within warranty but I'm not so sure that would help with the problem.

 

-Art

[mpw90]

Please keep in mind we don't mind replacing this memory kit if within warranty but I'm not so sure that would help with the problem.

 

-Art

 

Appreciate that, Art. Thanks.

 

I am convinced now that this is the CPU -- AMD have approved an RMA, so I suppose I will be without a processor for a while! This is the final test, before having to try another PSU (I cannot see how this would be a PSU issue).

[mpw90]

Please keep in mind we don't mind replacing this memory kit if within warranty but I'm not so sure that would help with the problem.

 

-Art

 

I have replaced my CPU with AMD. I will be testing this further this evening to determine whether or not the stutter still occurs.

 

Also worth noting that I replaced my RAM (Amazon were fine with this), and I am now running CL14, 3200Mhz DDR4 G Skill.

 

The remaining component that has not been changed is the Corsair TXM 550W+

 

I find it hard to believe a PSU could cause stutters, though.

[mpw90]

It is still happening.

 

[ame]

[/ame] - This is a clip I recorded in game of it happening, but for a very short amount of time. I noticed that, actually, the FPS *do* drop. They drop down to 89 from around 250-300fps. This is when the issues happens.

[Corsair Art]

It is still happening.

 

- This is a clip I recorded in game of it happening, but for a very short amount of time. I noticed that, actually, the FPS *do* drop. They drop down to 89 from around 250-300fps. This is when the issues happens.

 

If now the only common item that you have between the two builds is the PSU then please submit a ticket for that.

 

-Art

[mpw90]

If now the only common item that you have between the two builds is the PSU then please submit a ticket for that.

 

-Art

 

Corsair replaced the part, and I even went as far to buy an RX580 as well to test. It still persists.

 

I am so genuinely confused.

[Sarkoth]

Have you tried switching hard drives / installing on another drive? One of them could be failing / be heavily fragmented if HDD. That's the only thing not yet tackled as far as I haven't missed anything.

[mpw90]

Have you tried switching hard drives / installing on another drive? One of them could be failing / be heavily fragmented if HDD. That's the only thing not yet tackled as far as I haven't missed anything.

 

 

 

Yes, I have tried 4 different disks, all tested separately, with others disconnected. All new.

 

SSD 1 - Windows 10, format after issue not solved and tried Linux - didn't work

SSD 2 - Linux, format after issue not solved and tried Windows 10 - didn't work

 

HDD 1 - Windows 10, format after issue not solved and tried Linux - didn't work

HDD 2 - Linux, format after issue not solved and tried Windows 10 - didn't work

[Szychmistrz]

Mate do you have replaced PSU ?? i have ordered g.skill trend z rgb ram to give chance to amd but from all components (mobo gpu and g.skill gonna come in friday) i didnt change only cpu and PSU so iam corsious if its still microstutter presist after changin PSU on your case . I d like to say iam with you mate and i still having this problem since i bought all components, maybe Ram gonna solve the problem if not i am gonna swich to intel

[xin0710]

Yes, I have tried 4 different disks, all tested separately, with others disconnected. All new.

 

SSD 1 - Windows 10, format after issue not solved and tried Linux - didn't work

SSD 2 - Linux, format after issue not solved and tried Windows 10 - didn't work

 

HDD 1 - Windows 10, format after issue not solved and tried Linux - didn't work

HDD 2 - Linux, format after issue not solved and tried Windows 10 - didn't work

 

So you can confirm your issue is not a Windows 10 issue?

 

Because I've been experiencing a similar issue with CSGO starting late 2017 after I updated to the Windows 10 1703 Creators Update. This was back when CSGO was still on the Scaleform UI. I have not experienced any form of microstuttering in other games. It only happens on CSGO.

 

However, my issue isn't exactly just a "microstutter". You may classify it as a microstutter, but the experience on my end is more of a 5 second GPU driver freeze with a 100% GPU usage spike then dropping to 0% usage before recovering again.

 

Based on that experience, I've gone through most of the GeForce drivers starting with 378.78 manually DDU installing all the way up to my current driver 418.81 without any success -- granted, I wasn't trying to methodically fix the issue like you were. And yes, I am aware of the power monitoring tools causing microstutter and have stopped using MSI Afterburner entirely.

 

I recently reformatted my boot drive with a fresh install of 1803 three days ago. Even though I have yet to experience the "microstuttering" in the middle of a competitive match, (Edit: Jinxed myself just now.) I did however, experience the same GPU freezing while hosting an offline server which ended up kicking my friend out.

 

If you find my experience similar to yours, then I doubt it has anything to do with your AMD processor since I am on a first generation Core i7-950 on the x58 platform. Our common denominator would probably be Corsair components (RAM and PSU). But the only use I have for the iCUE software is for my Corsair K95 RGB keyboard since my RAM and PSU are pretty dated and do not include Corsair Link functionality.

 

If you can 100% confirm for me that Windows 10 is not the culprit, then I have reason to believe that the issue lies within Valve Software, Steam, and CSGO.

Edited February 15, 2019 by xin0710
I played 1 match.

[xin0710]

Welp, I guess the thread is dead.

 

I'll just leave these "microstutter" clips here for future Googlers:

 

NaVi s1mple

https://www.twitch.tv/s1mple/clip/AffluentMildKangarooDerp

 

Shroud (RTX 2080Ti)

https://www.twitch.tv/shroud/clip/AdventurousFancyCheesePhilosoraptor

 

Seangares (GTX 980Ti)

https://www.twitch.tv/sgares/clip/AttractiveGentleCheeseKappaClaus

 

mouz ropz (GTX 1080)

https://www.twitch.tv/ropz/clip/ChillyTardyPuppyPMSTwin

[A Computer Guy]

By any chance is your computer plugged into the same circuit as a power-line network adapter or another piece of heaver equipment like a furnace, home exhaust fan, or refrigerator, or some other kind of noisy electrical equipment?

 

If you have different circuit runs in your home try plugging your computer into a different circuit or even take your computer to a different location and power on and test there.

 

Just something to try since you seem to have exhausted so many options.

[xin0710]

Me, nope... and my computer is on a UPS. But I'll consider testing without the UPS and on a different circuit after testing this 358.91 driver with WDDM 2.0...

 

I just need Windows 10 to quit trying to force install version 388.13 even after opting not to in DDU safe-mode.

 

As for OP, it seems he either fixed his issue or just gave up.

[xin0710]

Update:

- Testing without UPS on a different 30 Amp circuit didn't solve the issue.

- 358.91 with WDDM 2.0 drivers provided longer session stability but still experienced one freeze after days of testing.

 

I am currently on the 419.17 driver which included random application crash fixes for Windows 7. Hopefully, it also has an effect on Windows 10.

 

Also, CSGO just got a major update:

[GRAPHICS]

– Added an optional -d3d9ex command line switch to reduce CPU memory use by about 40%. If you encounter graphics problems with this feature, please send an email to CSGOTeamFeedback@valvesoftware.com with the subject “D3D9EX”.

 

This significantly improved DWM alt+tab recovery performance for me.

I suggest anyone experiencing microstutter in CSGO to try out this new command-line argument as it will probably be implemented for all users in the future.

 

:!: This is not to be confused with the old -nod3d9ex command used to disable DWM Aero effects on Windows Vista. It is most likely the opposite of that.

[cplifj]

Have you tried switching hard drives / installing on another drive? One of them could be failing / be heavily fragmented if HDD. That's the only thing not yet tackled as far as I haven't missed anything.

 

This, and even the Sata/Raid firmware could be the culprit.

 

A minor incompatibility between sata firmware and SSD or HDD firmware.

 

Could thus also try to update SSD firmware.

[xin0710]

Progress on my part!

 

Using the newly added -d3d9ex command-line switch basically turned the 5 second CSGO "microstutter" into a full single frame indefinite freeze.

This single frame gets stuck on screen indefinitely while the game audio and command input still function in the background.

Alt+tabbing back into the game did not help. In fact, the csgo.exe process stops using the GPU Engine completely at this point.

This was a horrible experience for competitive matches since the game never actually recovers from this frozen frame state.

So I am forced to close the game entirely every time it happened.

 

However, having this single frame stuck on my screen indefinitely basically made it easier to test and troubleshoot the microstutters.

I can just leave the game running offline in the background without having to actively wait for a microstutter to occur.

And yes, it happens offline as well, which mostly eliminates any networking related cause.

 

After hours of Googling, I came across a Borderlands 2 freezing issue that has existed since 2014 up to this day.

Borderlands players have solved this issue by turning off the in-game PhysX.

...

 

It never crossed my mind that CSGO, a non-PhysX game, could possibly be affected by the PhysX system software driver.

Either way, I open up NVCP and configure PhysX to run on the CPU instead of the GPU.

 

After a full week of testing this and idling my game offline, I haven't experienced a single freeze or microstutter. :thinking:

 

A few more days pass and I decide to set PhysX to run back on the GPU.

I then go out for lunch and return to a frozen CSGO screen... :stickpoke

 

Welp, my fingers are crossed.

Hopefully, this will be the last post you guys will ever see from me on this thread.

I wish OP the best of luck with his case. -- jinxed again.

 

Edit:

Nope. It still froze. Although, it happens less often with PhysX set to CPU.

 

JasonR:

[ame=

]

[/ame]

 

Valve, please fix!

Edited March 22, 2019 by xin0710
Still froze.