• Welcome to TechPowerUp Forums, Guest! Please check out our forum guidelines for info related to our community.

My games stutter w/ mouse lag (13900k/4090/3080) - Need Help

Benjji

New Member
Joined
Jan 19, 2024
Messages
21 (0.07/day)
I find it interesting that this issue 'suddenly' started happening for you. Yeah, could be an update or perhaps some change you made or 3rd party app you installed, perhaps even a faulty driver installed in the background, who knows. Your memory could have have been on the edge of stability this whole time and the issue is now presenting itself.
yeah to be fair it probably was on the verge of being unstable the entire time but something pushed it over, either a driver (nvidia drivers are having problems recently) or a windows update (there was reports of cpu degredation in update 23h2 ). they are on different die versions so im leaning more towards it being that the entire time (i bought 2x8 ddr4 sets seperately) and me either not noticing it fully until now or it was somehow exacerbated with a recent update, as when i was trying to retrain the ram to make sure they were all detected i saw that the dram voltage seemed to spike, either from a singular ram stick or from the other set. with only 2 dimms it was at 1.232v but with 4 dimms it was at 1.376v. although this could just be because of it being 4 dimms and it needing to output more voltage to compensate.
 
Joined
Apr 24, 2022
Messages
168 (0.18/day)
System Name lawooder
Processor 5800x3D
Motherboard B550M Aorus Elite
Cooling Arctic Freezer 34 DUO
Memory 32GB 4000Mhz CL16 (2x16GB)
Video Card(s) RTX 3070 Ti
Storage SN770 250GB/SN750 500GB/SA510 1TB
Display(s) VG279QM 1080P 280hz + VG27AQ 1440P 165HZ
Case MSI MAG Something
Audio Device(s) fiiO EK10 + HD 599
Power Supply Focus GX 750 Watts
Mouse XM1r
Keyboard Xtrfy K4
Joined
May 8, 2006
Messages
25 (0.00/day)
why for sake you need 2005, 2008, 2010 etc. in 2k24?
There's a reason most recommend leaving existing C++ redist runtime versions installed. Some apps you use may still require these version to run properly. I believe even Epic Games Launcher still installs c++ 2010 for example. It all depends on what apps/games are being used. I still play a game that was originally released in in 2010. This is why that c++ pack was created, to install all versions to raise compatibilitiy across the board.

Okay, so I ran into another 'stutter, lag, desync' issue and wanted to share to possibly help others experiencing the same problem.

This issue lies w/ a game originally released in 2010 and is a 2 thread application, Quake Live. I gave up on the other game I was trying to play because I concluded it's an issue only the DEVS can fix so it's not worth my time anymore, so here I am trying to jump back into Quake Live. In theory, the game should run flawlessly without a hitch, yet the game is clearly lagged, there's stutter, and everything feels like it's 500ms behind. It's just an awful experience. These Hybrid CPU's with how threads are scheduled just don't work properly with older games, it's infuriating. Through my troubleshooting, I discovered a couple of things.

1) Disabling E-Cores hurts performance and sometimes causes performance to absolutely tank. I ran latency and benchmark tests and found my memory and L3 cache latency would spike randomly if I disabled e-cores from the bios. My benchmark scores in say cinebench were also slightly lower. When E-Cores were re-enabled, memory and l3 cache latency dropped significantly along w/ my ring increasing a good amount. Here's a good article on this phenomenon. It's a bit long, but a good read. It's also good to note that my performance also dropped when even PARKING E-CORES. Absolutely ridiculous. It appears that e-cores are somehow directly tied to the performance of the chip's overall performance of the L3 cache, memory latency and even ringbus.


2) Thread scheduling is ultimately broken and will not work correclty w/ legacy/older apps. Windows 11 is also a no go if you play older games because the thread director makes the problem even worse. You'll never get proper performance and responsiveness due to the broken thread scheduling.

So now on to the first problem. I have to make an older game/app work w/ E-Cores enabled. Unfortunately, I couldn't get this to work properly on Windows 11 due to the thread director. Setting core affinity unfortunately did not work to 'resolve' the problem, only alleviate the symptoms. Parking the e-cores made performance worse while spiking latency. The only option was to leave e-cores on and tweak the thread scheduling policies. So due to the thread director, I dropped Windows 11 and went back to Windows 10 so the thread director wasn't interfering.

Now on Windows 10, here's how I finally got this app/game to run properly without having to set core affinities or park/disable e-cores. First I disabled core parking, ensured 'Standard Core Parking (0)', was the enabled core parking policy, not 'Hetero Core Parking.' Hetero Core Parking wants to park/unpark p/e cores as it sees fit for the task at hand. I then changed the thread scheduling policy for both short and long threads to: 'Prefer Performant Processors.' After doing this, the game was no longer lagged, out of sync, and just worked. I went ahead and tested this configuration in other games and this seems to work in every game I've thrown at it, thus far. It's also good to mention that I gained 50 fps in the finals after using this configuration. This tells me that Windows 10 default thread scheduling doesn't work properly for all apps/games which I guess we already knew.

I guess the big takeaway of this post is that disabling E-Cores, as I was doing, can negatively impact performance. Latency will be higher and performance will suffer. This to me feels more like a bug, but who knows. After coming to this conclusion I researched and found that some were already aware of this discrepancy. Some users claim this is a bug and others claim this 'bug' doesn't affect every CPU, only some. Some claim that if you enable a single e-core, you won't get hit with the performance penalty, but that single e-core can negatively impact older/legacy apps. I can't say for sure, but it's definitely been a nightmare ever since switching over to Intel's Hybrid arch....

If anyone is interested in trying this, I recommend using QuickCPU. It's a great app to customize your core parking policies, thread scheduling policies, etc. It's a lot like ParkControl. It's far easier than unhiding all of the power plan policies in the registry.

yeah to be fair it probably was on the verge of being unstable the entire time but something pushed it over, either a driver (nvidia drivers are having problems recently) or a windows update (there was reports of cpu degredation in update 23h2 ). they are on different die versions so im leaning more towards it being that the entire time (i bought 2x8 ddr4 sets seperately) and me either not noticing it fully until now or it was somehow exacerbated with a recent update, as when i was trying to retrain the ram to make sure they were all detected i saw that the dram voltage seemed to spike, either from a singular ram stick or from the other set. with only 2 dimms it was at 1.232v but with 4 dimms it was at 1.376v. although this could just be because of it being 4 dimms and it needing to output more voltage to compensate.
Are you using a single 4 dimm kit or 2 separate 2 dimm kits? If using 2 separate 2 dimm kits together, that could be your issue. Kits are binned together, so you can't mix and match even if you buy 2 identical 2 dimm kits thinking they're compatible with each other. You'll run into all kinds of weird memory stability issues otherwise. Also, it's notoriously difficult to stabilize 4 dimm kits on DDR5 above 6000. What kit or kits are you using?

Recommend everybody with nvidia card to update.

It's a very good driver and fixes serious issues.
Cool, thanks for the post. I'll definitely install it.

Looks like it fixes stuttering when v-sync is enabled (Which I don't use) and scrolling in websites? Either way, good update looks like.
 
Last edited:

Benjji

New Member
Joined
Jan 19, 2024
Messages
21 (0.07/day)
Are you using a single 4 dimm kit or 2 separate 2 dimm kits? If using 2 separate 2 dimm kits together, that could be your issue. Kits are binned together, so you can't mix and match even if you buy 2 identical 2 dimm kits thinking they're compatible with each other. You'll run into all kinds of weird memory stability issues otherwise. Also, it's notoriously difficult to stabilize 4 dimm kits on DDR5 above 6000. What kit or kits are you using?
yeah they are two seperate ddr4 kits, although it has only seemingly started to act up recently, i couldve maybe encountered a bsod once every few months but now its nearly once a week. both error codes seem to point to faulty hardware (0xc0000221 and 0xc0000428) specifically RAM related hardware problems, every time ive encountered a bsod it will bsod loop until i remove a ram stick and then it will post to windows no problem. one kit is v3.31 micron b and the other is 3.34 micron e.
ive added the bsod screens, i do have more which have different reasons of why the pc couldnt boot but they all either had 0xc0000221 and 0xc0000428 error codes also added a screen that i saw before the BIOS appears, should be the third image. that appeared like that until i restarted the computer. after restarting it went back to bsod looping, also have another image of when i was able to get into the BIOS but some UI elements didnt load

ive also ran an DISM and SFC scan each time these bsods happened and they never detected a corrupted file, also im pretty sure the most recent bsod that happened (the images are from the most recent bsod) it did it after me changing the RAM mhz from 3400 to auto (both kits are rated 3600mhz CL18) and after booting it was only set to 2133. restarted again and changed the xmp from 2 to 1 to try and get the rated xmp mhz and then the bsod loop happened.
also sometimes after reseating ram to fix the bsod loop not all ram sticks are added to the total available RAM. the DIMMS are detected in BIOS but dont contribute to the total RAM size. also NZXT CAM doesnt recognise all 32gb when both windows and the BIOS do, it only shows 24gb. although that could just be CAM being faulty. it also seems only recently that my SOC voltage have spiked. before the SOC would stay at around 1v. now it goes to 1.08v and is seemingly causing a sort of coil whine near the IO ports because of this, the vcore has done something similar (goes between 1.142 and 1.250) and the DRR voltage goes from 1.232 to 1.376 when XMP is enabled
 

Attachments

  • 1708615535715.png
    1708615535715.png
    7 MB · Views: 57
  • 1708615579542.png
    1708615579542.png
    172.5 KB · Views: 45
  • 1708615586437.png
    1708615586437.png
    241.7 KB · Views: 46
  • 1708615595518.png
    1708615595518.png
    142.6 KB · Views: 53
  • 1708616158983.png
    1708616158983.png
    94.7 KB · Views: 47
Last edited:
Joined
Oct 2, 2020
Messages
938 (0.63/day)
System Name ASUS TUF F15
Processor Intel Core i7-11800H
Motherboard ASUS FX506HC
Cooling Laptop built-in cooling lol
Memory 24 GB @ 3200
Video Card(s) Intel UHD & Nvidia RTX 3050 Mobile
Storage Adata XPG SX8200 Pro 512 GB
Display(s) Laptop built-in 144 Hz FHD screen
Audio Device(s) LOGITECH 2.1-channel
Power Supply ASUS 180W PSU
Mouse Logitech G604
Keyboard SteelSeries Apex 7 TKL
Software Windows 10 Enterprise 21H2 LTSC
There's a reason most recommend leaving existing C++ redist runtime versions installed. Some apps you use may still require these version to run properly. I believe even Epic Games Launcher still installs c++ 2010 for example. It all depends on what apps/games are being used. I still play a game that was originally released in in 2010. This is why that c++ pack was created, to install all versions to raise compatibilitiy across the board.

Okay, so I ran into another 'stutter, lag, desync' issue and wanted to share to possibly help others experiencing the same problem.

This issue lies w/ a game originally released in 2010 and is a 2 thread application, Quake Live. I gave up on the other game I was trying to play because I concluded it's an issue only the DEVS can fix so it's not worth my time anymore, so here I am trying to jump back into Quake Live. In theory, the game should run flawlessly without a hitch, yet the game is clearly lagged, there's stutter, and everything feels like it's 500ms behind. It's just an awful experience. These Hybrid CPU's with how threads are scheduled just don't work properly with older games, it's infuriating. Through my troubleshooting, I discovered a couple of things.

1) Disabling E-Cores hurts performance and sometimes causes performance to absolutely tank. I ran latency and benchmark tests and found my memory and L3 cache latency would spike randomly if I disabled e-cores from the bios. My benchmark scores in say cinebench were also slightly lower. When E-Cores were re-enabled, memory and l3 cache latency dropped significantly along w/ my ring increasing a good amount. Here's a good article on this phenomenon. It's a bit long, but a good read. It's also good to note that my performance also dropped when even PARKING E-CORES. Absolutely ridiculous. It appears that e-cores are somehow directly tied to the performance of the chip's overall performance of the L3 cache, memory latency and even ringbus.


2) Thread scheduling is ultimately broken and will not work correclty w/ legacy/older apps. Windows 11 is also a no go if you play older games because the thread director makes the problem even worse. You'll never get proper performance and responsiveness due to the broken thread scheduling.

So now on to the first problem. I have to make an older game/app work w/ E-Cores enabled. Unfortunately, I couldn't get this to work properly on Windows 11 due to the thread director. Setting core affinity unfortunately did not work to 'resolve' the problem, only alleviate the symptoms. Parking the e-cores made performance worse while spiking latency. The only option was to leave e-cores on and tweak the thread scheduling policies. So due to the thread director, I dropped Windows 11 and went back to Windows 10 so the thread director wasn't interfering.

Now on Windows 10, here's how I finally got this app/game to run properly without having to set core affinities or park/disable e-cores. First I disabled core parking, ensured 'Standard Core Parking (0)', was the enabled core parking policy, not 'Hetero Core Parking.' Hetero Core Parking wants to park/unpark p/e cores as it sees fit for the task at hand. I then changed the thread scheduling policy for both short and long threads to: 'Prefer Performant Processors.' After doing this, the game was no longer lagged, out of sync, and just worked. I went ahead and tested this configuration in other games and this seems to work in every game I've thrown at it, thus far. It's also good to mention that I gained 50 fps in the finals after using this configuration. This tells me that Windows 10 default thread scheduling doesn't work properly for all apps/games which I guess we already knew.

I guess the big takeaway of this post is that disabling E-Cores, as I was doing, can negatively impact performance. Latency will be higher and performance will suffer. This to me feels more like a bug, but who knows. After coming to this conclusion I researched and found that some were already aware of this discrepancy. Some users claim this is a bug and others claim this 'bug' doesn't affect every CPU, only some. Some claim that if you enable a single e-core, you won't get hit with the performance penalty, but that single e-core can negatively impact older/legacy apps. I can't say for sure, but it's definitely been a nightmare ever since switching over to Intel's Hybrid arch....

If anyone is interested in trying this, I recommend using QuickCPU. It's a great app to customize your core parking policies, thread scheduling policies, etc. It's a lot like ParkControl. It's far easier than unhiding all of the power plan policies in the registry.


Are you using a single 4 dimm kit or 2 separate 2 dimm kits? If using 2 separate 2 dimm kits together, that could be your issue. Kits are binned together, so you can't mix and match even if you buy 2 identical 2 dimm kits thinking they're compatible with each other. You'll run into all kinds of weird memory stability issues otherwise. Also, it's notoriously difficult to stabilize 4 dimm kits on DDR5 above 6000. What kit or kits are you using?


Cool, thanks for the post. I'll definitely install it.

Looks like it fixes stuttering when v-sync is enabled (Which I don't use) and scrolling in websites? Either way, good update looks like.
well, I play NFS MW 2005, use Windows Live Gallery 2012 lol. But, this doesn't mean everyone do. So, my thoughts are install something relatively modern for everyone, and for those who need retro knows better what is required ;)
 

MLadie

New Member
Joined
Mar 21, 2024
Messages
1 (0.00/day)
After dealing with this problem for a month, I finally found this post and first of all I must say I'm impressed by Kokkatc. He managed to explore every possible solution in a way I believe very few people out there would bother.

Interestingly enough, I came up with this issue on the 17th of February and I'm only seeing this in Overwatch 2. I'm going mad because I thought, at first, that the game would be somehow the culprit, especially because a very big patch came out a few days earlier. Turns out that NO ONE noticed this issue. I am very sensitive to input lag, stutters and mouse lag, but there are many pro players out there that have way better reflexes than mine, so it's impossible that no one noticed the same issue, if it's just the game.

Pretty similarly to what Kokkatc described, the game never loses frames. Temps are fine and I've tried EVERYTHING (including flashing bios multiple times, trying every possible version of XMP, disabling XMP, reinstalling windows multiple times (both 10 and 11, just to be sure)). In the shooting range, the game looks somehow playable, but when I get into a match everything feels floaty, desync'd, mouse responsiveness changes from one frame to another and it's very hard to aim. I have no issues in other games and I am 1000% sure the game was ok before the update. Another thing that seems interestiny is that the issue is way more visible when capping framerate at my monitor refresh rate, 240. When the issue occurs, every model starts moving erratically and there's major screen tearing. Turning on Vsync seems to make everything better looking, but the added input lag makes it just not worth.

Now I know what you're thinking: it's only Overwatch and there was a big update. But why only me? It made the game completely unplayable, it's impossible that no one else complained about it. I'm going mad and honestly I want to quit gaming for good, because I'm tired of these stupid issues.

Now Kokkatc pointed out that most of these things bring the attention towards memory, which is interesting. Also important is that I haven't made any changes to my pc that could explain why was it ok 1 month ago and now it's totally unplayable. I might buy another set of RAM just to be sure, because I'm running DDR5 as well and I want to be sure about my ram sticks. I wonder if there were any updates on Kokkatc situation, and I want to know if the noticed the same behaviour in all games, including single player games. For example, Doom Eternal, Cyberpunk and Forza Horizon are insanely demanding, but they run perfectly on my system. When you had this issue, did you notice it in EVERY game?

My PC:
i7 13700k
MSI RTX 4090
32 GB Corsair Vengeance DDR5 6200 MHz
Razer Deathadder V3 Wireless

I tried many other things, just like Kokkatc did. No one worked.
One thing that seems interesting, is that in my case reinstalling OS seems to mitigate the issue for a short time, then it comes back. Some people pointed out this could be related on network as well.
 
Last edited:
Joined
Nov 6, 2018
Messages
3 (0.00/day)
Alright, so I have an update on this, and boy has it been a nightmare. I'm not saying my problem is resolved until I go for 2-3 weeks without issues.

So, I have found 4 or so things directly impacting performance in the games I play. The first being memory stability. What really annoys me about this one is that I have tried 3 different kits, all on mobo's QVL, all kits passing every single memory test I throw at it (memtest, tm5 w/ anta extreme, ycruncher, etc etc etc, running for HOURS, as much as 8-12hrs) showing zero errors. I decided to take the time and manually overclock my memory, fine tune primaries, secondary, tertiary timings and voltage. I went high on voltage and performance is already significantly better. Stutters are less, mouse lag is better, basically night and day difference. We're only talking about a difference of -10ns so it really shouldn't yield the kind of performance gains I'm seeing but here we are.. regardless, my memory clearly wasn't stable even though every test I take suggest it is. DDR5 on z690 and z790 has been a complete nightmare to deal with, at least in my experience. I never had any such issues on DDR4.

Okay, second thing I found. HAGS (Hardware accelerated gpu scheduling) ON for the game I play increases stutter, SIGNIFICANTLY. So this is now disabled. Even REBAR ON adds stutter for the games I play, so I disabled this from the bios as well.

Third, and probably the most frustrating of all. My 13900k is clearly FAULTY. I noticed something was off when I decided to add a +.02-.03 voltage offset to see if perhaps my CPU wasn't receiving enough voltage at default levels. This was a night and day difference. Adding a slight positive voltage offset increased my in game responsiveness, significantly. Also, stuttering was far less pronounced. The problem wasn't fixed entirely, but there was a notable improvement. This led me to swapping out my 13900k w/ a 13700k I still have on hand because that kind of behavior suggests a bad cpu or possibly bad cpu imc.. After I swapped it out and jumped into a game... holy mother of God, I honestly forgot what smooth gameplay is supposed to feel like at this point. All stuttering, lag, etc, was miraculously gone in every game that I play. Everything felt snappy and responsive again, all at stock settings w/ the exception of my overclocked memory settings. That I applied because my stock XMP/XMP2/XMP Tweaked settings do not work well at all for some reason. So, in one game all my problems seemed to have vanished... in the other, desync and input lag was rampant when in an actual game/match/online of course. I find it interesting that now my 13700k seems to be fine when it wasn't initially, probably due to my memory being unstable at the time.

Now the last problem I found w/ a game I play is a game that uses EAC (Easyanticheat). The moment my game becomes desynced while in game, I checked event viewer and immediately found this error:

<Event 2, Kernel-EventTracing - session "dc3a3596-71e1-45a3-b2ea-39ad5322fe51" failed to start with the following error: 0xC0000022>

I quickly found out using ProcExplorer that <dc3a3596-71e1-45a3-b2ea-39ad5322fe51> is Easyanticheat. Something on my system is blocking the session from starting, perhaps preventing EAC to connect with its servers. This error pops up every time I play this game and when this happens, desync and input lag are hilariously bad, we're talking anywhere from +200-500ms. It's an absolutely horrific experience. Keep in mind, I just reinstalled my OS and there are no 3rd party apps. Only drivers (chipset, lan, gpu, etc) are installed along w/ fully updating Windows. It's 100% clean from garbage or 3rd party nonsense. Something inherent to Windows is blocking EAC from starting which makes this problem incredibly obscure and frustrating. What's more likely is the game devs fumbled on the EAC implementation. I have found several potential window's services that could be responsible for blocking EAC from starting a session. Also, running EAC as admin makes the issue worse and making sure EAC is allowed in my firewall and an exclusion set for Defender makes no discernable difference in my case. I'll post back more once I diagnose this issue a little more today but I think I'm getting very close here. I also informed the DEVS of the issue and kindly asked they just disable EAC until it's fixed but I'm not holding my breathe. I've also learned that many others have faced the same issue across other games utilizing EAC such as Apex, Elden Ring, etc.

<Event 2, Kernel-EventTracing - session "dc3a3596-71e1-45a3-b2ea-39ad5322fe51" failed to start with the following error: 0xC0000022>

This happens when whatever game you're playing uses EAC and that game is not run as Admin. Depending on the game the event can happen every 5 minutes (to the second) starting 5 minutes after the game is started or every minute (to the second) 5 minutes after starting the game.

This has been happening for 3-4 years that I know of which is when I first noticed it and figured out how to resolve it.

I tested it just now to see if anything had changed and it is still exactly the same as it was years ago. With the game run as Admin, the error never happens even after hours of gameplay. With the game not being run as Admin the error happens 5 minutes after starting the game and continues to happen every minute thereafter.
 

Attachments

  • error.png
    error.png
    55.4 KB · Views: 91
Top