Input delay, Hitreg, Desync - Where to fix them

[Kipperii]

Actually I found the way to solve and reproduce the desync problem, but im still testing how to maximize the fix of this problem.

[Kipperii]

I’ve been around since 2009. I actually started playing Counter-Strike on Steam back in 2006. People online might know me as haze, cloudhaze, or x58haze on GitHub and other platforms.I vividly remember when this whole issue became a hot topic. It was all thanks to a legendary Steam forum thread that has since been deleted. It was titled "CS isn't smooth in 2009." It was so iconic that it sparked the entire community conversation around desync, input lag, hit registration issues (bullets doing zero damage), "Ferrari peeking," and that awful heavy mouse feeling. It made you feel like you were trapped in a 4:3 stretched jail—your heart racing just trying to aim, while other players looked completely relaxed, jumping around and enjoying a flawless, crisp connection. That thread was amazing; it had thousands of comments analyzing different PC builds, power conditioners, and tweaks.Yet, here we are today, still facing the exact same weird issues. We all know that rare feeling when the game suddenly behaves perfectly: footsteps are crystal clear, utility trajectories look incredibly smooth, your reaction time peaks, and enemies seem to move in slow motion, making them easy to aim at.I’ve been down the ultimate rabbit hole trying to fix this. I used to mod BIOS files, revive bricked motherboards using CH341A programmers, desolder chips, mess with USB latencies, disable Windows background services, and tweak OS kernels and timers like crazy. I’ve formatted Windows thousands of times, but nothing ever provided a permanent fix. I spent years studying netcode, frame rates, frame pacing, 1% and 0.01% lows, and tools like Perfmoon. Honestly, I'm grateful that Counter-Strike 1.6 turned me into a tech geek.At this point, I believe it’s a combination of multiple factors:Network Jitter and UDP Packets: This is the hardest variable to measure. It is the hidden enemy of online gaming. Many ISPs actively prioritize UDP packets for streaming services like Netflix over the tiny, time-sensitive data packets used by competitive games like Counter-Strike.Offline Judder and Visual Lag: This issue exists even offline. If you search YouTube for the "Bruce Willis Juddering" clip, you’ll see a perfect example of what this visual stuttering looks like. It completely destroys competitive gameplay.Years ago, I used to have more "good gaming days." Nowadays, it feels like it only happens 1 out of 30 days, or even 1 out of 300. It’s incredibly rare to get a flawless session. Sometimes you can temporarily fix online issues by reducing Total Harmonic Distortion (THD), but the fix never lasts.For instance, you can temporarily clear static and residual energy from your PC by unplugging the PSU, removing the 24-pin motherboard cable, and holding down the case power button so the PSU capacitors drain the motherboard. You can then completely wipe the NVRAM by removing the CMOS battery and shorting the CLRTC jumpers. I only resort to this when my local power grid is so dirty that it corrupts the GPU or monitor OPROM, causing it to fail UEFI recognition. While this helps in both offline and online environments, it still doesn't permanently solve desync.If you look at professional gaming houses, they have a "perfect" environment. They use a high-end modem in bridge mode (zero NAT) paired with an enterprise-grade router running isolated VLANs and a single DHCP setup to completely avoid double NAT or CGNAT. They implement strict QoS/VLAN rules to prioritize game executables (hl.exe, hl2.exe, etc.) over any Windows background data.Furthermore, they operate on pristine electrical grids. Developed countries heavily regulate Total Harmonic Distortion (keeping it around 3-5%), whereas in Venezuela and other parts of LATAM, THD can easily spike to 15%, 20%, or even 30%. In environments like mine, you absolutely need a high-end, double-conversion online UPS with a strict 2% THD output, proper grounding that avoids daisy-chaining, and deep knowledge of your hardware limitations.To summarize, for a game to feel truly smooth today, we need:A strict 2-5% Total Harmonic Distortion (THD).Stable electrical voltage (110-120V or 220-240V).A rock-solid grid frequency that doesn't fluctuate (50Hz in Europe, 60Hz in LATAM). If you disassemble a monitor, the internal labels show their sweet spot for power frequency, usually allowing only a tiny 4% margin of error. If your power grid constantly fluctuates outside of that margin, it directly degrades the clarity of the image on your screen.It’s an incredibly complex mix of variables. Unfortunately, living here in Venezuela makes it very difficult and expensive to acquire the high-end equipment needed to test and isolate all of these electrical and network issues.It’s incredibly frustrating because it completely drains the fun out of gaming. I can’t even enjoy offline games anymore because I constantly notice the visual judder, micro-stutters, and desync.God bless you all. I'm not a religious person, but I just wanted to share my story and send some love to this community! <3"

PS: had to help myself with the AI fixing my Grammar, thanks again.

First, it is impressive that you have covered so many topics. I suffered for such a long time myself that I quit competitive gaming back in 2021. I know that feeling when everything is in sync and no one can touch you unless they outplay you. But while I agree with all your points on the network side (VLANs, QoS at the proper interface for outbound and inbound, and making sure your connection isn't CGNAT), I am not really convinced of this "dirty electricity" concept causing desync. Now, I am not saying it won't cause *any* issues, but if it did, you would know. Your clocks, voltage, BCLK, and in extreme cases, your monitor would fluctuate, and your frametimes would show symptoms of it at the very least.

Game netcodes have come a long way, so even subtle network jitter will only put you behind by one tick (which is still not great, but so much better than the desync described by most people here). While not every pro starts from a gaming house, I have known and seen a lot of players with incredible gaming experience. There are many layers where testing can go wrong, and a lot of popular guides already cover them.

Since you are already too deep into it:

Hardware Side:

CPU & RAM: Make sure your clocks are stable and that the system is not thermal or power throttling. Ensure your RAM is running in 1:1:1 (though technically, a 1:1 UCLK to MEMCLK ratio is more important, along with finding a stable FCLK for your system on Ryzen). Also, remember that finding stability depends heavily on your motherboard's VRM quality.
Network Adapter: A NIC with built-in hardware queues works best.

Measuring on the Windows Side:

ISR / DPC: LatencyMon. While a lot of people concentrate on the highest DPC values, the equal important thing is your average DPC values aswell.
RTTS & CapX: Use these for frametimes, 1% and 0.1% lows, and your game thread clock speed.

I will share one observation I made while trying to help a friend: his P-state in the BIOS was set to P-state 1, which played a huge role in poor 0.1% lows that weren't really visible on the monitor except for a few hitches. But it's not just the main game logic thread that spoils the experience; the network thread can also be affected, adding another layer to the poor experience.

WPT (Windows Performance Toolkit): This is the most interesting and revealing part if everything else looks fine. Here, you can see all your game threads and what cores are being used for them (checking whether some other process or app is using the same threads as the game). The tricky part is that while processes are named, game threads themselves are not explicitly named (they could be the game network thread, main game logic thread, or render thread), so you have to use logic to isolate them.

For example, to isolate and crosscheck the network thread, you have two options:

1. Under your game process threads, you can check for threads that have had interrupts matching your game server Hz (you would have interrupts every 8.3ms if its 120h server , so you check in a variance close to your network jitter, and you will find your network thread easily).
2. The easiest way is to look under your network driver process (usually NDIS), where you can see interrupts with connection IPs. Note the time and check which game thread matches that timing. This gives you a more realistic way to measure if a packet that hit your PC is being delayed or instantly delivered to your game process.

While WPT is really for advanced users, it will help you isolate problems specific to drivers, OS scheduling of your game process, network thread delays, ISR times, and saturation.

Network Side:

If you want a simple way without a lot of hassle, record a Wireshark capture while playing a game, then isolate the downstream of your game traffic and give it to Claude along with your game server tick rate so the AI can check for packet pacing jitter. It will tell you if feeling behind the server is due to network jitter, which causes the game client to set a high interpolation buffer. Of course, this will only tell you the downstream quality, not the upstream. If you want to measure both, iperf3 measures it somewhat accurately.

If you want to fully isolate it, we have to split it into:

1. Internet - Your WAN Device (Your First Router)
To test this, I personally use a 60Hz UDP test script I coded to test pacing from and to servers via my first WAN device. It helps me measure pacing, gaps, and packet batching. I run it on a server that mimics similar routing to and from game server locations (some say routing is different so it doesn't tell the whole story, but ignore them—most saturation occurs at the OLT collection point anyway, which is before peering; otherwise, gpns like ExitLag would have already fixed it). If this gives you perfect values (generally a good jitter below 2ms), then your internet connection is fine.

2. Your Local Setup (Router to PC)
If you have a router running a Linux OS like OpenWrt, it is easy to measure using tcpdump. The best run is when playing a game. Run tcpdump on the WAN interface and run Wireshark on your PC at the same time. Make sure you isolate your game connection either by IP or port before running it, otherwise it will be a mess.

Then, compare them both, find the packet ID, and crosscheck the timings. While you can't perfectly match timings (since there will be a slight difference the moment you start the test on both sides), the proper way to check is by analyzing the pacing through interpacket gaps, which will give you a clear picture of your local network quality and bufferbloat. Though remember, bufferbloat can also create queuing on the WAN interface itself for downstream traffic, so this isn't a completely reliable way to measure bufferbloat entirely.

But properfix for local bufferbloat would be Limiting your bandwidth with an qos and dscp values for traffic priority.

This post assumes familiarity with these technical terms, it's not aimed at basic users. I'd like to turn it into a full guide at some point, but that's a much bigger undertaking it needs a lot of background explanation for everyone who don't already have this knowledge, and that takes real time to put together properly.

Hello , i would like to know if changing my ethernet card "speed & duplex" from as example " 10gb/s to 100mb/s is the same as "limiting my bandwidth with a qos please"?
If it's not the same , what do you recommend as a bandwidth limit with the qos?

Buy a router with cake/sq_fodel and set it to like 80-90% bandwidth

*fq_codel

[sadboytunaswag31]

Ive had this feeling off being behind the server people shooting too fast to react to and my hit reg feels off IN MULTIPLE Games : Apex legends, Valorant , R6 (I was t3 player in val, champ r6 , master in apex ) so i dont think its a skill issue personally and Ive changed my pc THREE times with all components and still have this problem , sometimes i do some random "tweaks" that remove the issue for lets say a day and it all comes crashing down again.

My pc specs :

Msi b850 Gaming pro
32 gb ddr5 ram g skill ripjaws M5 cl 36
ryzen 7 7800x3d
Msi 5060 gaming
m.2 Kingston SNV321000g

the only thing I think could be the problem is electricity i dont believe the my house has earthing (Not knowlegeble in the "engineering aspect") so can that be the issue Idk at this point

DId you check your udp jitter with iperf3 or any other tests? How does your frametime graph and 1% low look like?
You can dm me on discord - xvenom_03

Private messaged this guy. Got no answer after some time and he wont add my discord either.
Any updates?

Ive already told you multiple times to post it in this topic instead of private forum messages as i dont look in it. Also if you sending a req in discord do send with a message. I wont accept req without proper context.

I sent u my discord multiple times and i also did send a message with my request. U ignored it and instead offered another guy under this thread to add you.

I’m desperate to try to fix this bro. In your “guide” you mainly go over analyzing if desync is happening and not much on how to fix it.
My discord is ree.n.

I already have a Linux based router with openwrt running sqm, and my jitter shows below 0.5ms on avg on most games. Good frame times, low dpc/ isr latency. I’m helpless.

I hope you can help me

[jassine]

Actually I found the way to solve and reproduce the desync problem, but im still testing how to maximize the fix of this problem.

what did you do ?

[xVenom]

The blurriness and desync is coming back at my new house after 6 months since moving in. The old thing of moving random cables effecting it and turning light switches on and off has returned. When I first moved in this wasn't happening. The creator doesn't want us playing games or something, unlucky guys maybe next life will be better

You know, man, your message made me cry (I'm not kidding). My life is pretty shitty, and online PC games were the only thing that brought me joy and fun. I used to play a lot with my friends, etc. But this input lag issue ruined everything; I became nervous, irritable, and even aggressive. For about a year now, I haven’t been playing online, but my life (IRL) has gotten even shittier, with even more pain and suffering. It feels like I'm just meant to suffer. And now, the suffering I had from the input lag doesn't seem that bad anymore. I don't know, man. Try to do whatever brings you joy and causes less suffering.I hope we can all find happiness and joy in something, at least.Or that someone will manage to fix this input lag issue and be happy too.

Hey Man, There is lot of things outside gaming that gives you joy. You just have to find something that is passionate to you if not gaming. Nevertheless DM me i will try to help you. <3

[xVenom]

I’ve been around since 2009. I actually started playing Counter-Strike on Steam back in 2006. People online might know me as haze, cloudhaze, or x58haze on GitHub and other platforms.I vividly remember when this whole issue became a hot topic. It was all thanks to a legendary Steam forum thread that has since been deleted. It was titled "CS isn't smooth in 2009." It was so iconic that it sparked the entire community conversation around desync, input lag, hit registration issues (bullets doing zero damage), "Ferrari peeking," and that awful heavy mouse feeling. It made you feel like you were trapped in a 4:3 stretched jail—your heart racing just trying to aim, while other players looked completely relaxed, jumping around and enjoying a flawless, crisp connection. That thread was amazing; it had thousands of comments analyzing different PC builds, power conditioners, and tweaks.Yet, here we are today, still facing the exact same weird issues. We all know that rare feeling when the game suddenly behaves perfectly: footsteps are crystal clear, utility trajectories look incredibly smooth, your reaction time peaks, and enemies seem to move in slow motion, making them easy to aim at.I’ve been down the ultimate rabbit hole trying to fix this. I used to mod BIOS files, revive bricked motherboards using CH341A programmers, desolder chips, mess with USB latencies, disable Windows background services, and tweak OS kernels and timers like crazy. I’ve formatted Windows thousands of times, but nothing ever provided a permanent fix. I spent years studying netcode, frame rates, frame pacing, 1% and 0.01% lows, and tools like Perfmoon. Honestly, I'm grateful that Counter-Strike 1.6 turned me into a tech geek.At this point, I believe it’s a combination of multiple factors:Network Jitter and UDP Packets: This is the hardest variable to measure. It is the hidden enemy of online gaming. Many ISPs actively prioritize UDP packets for streaming services like Netflix over the tiny, time-sensitive data packets used by competitive games like Counter-Strike.Offline Judder and Visual Lag: This issue exists even offline. If you search YouTube for the "Bruce Willis Juddering" clip, you’ll see a perfect example of what this visual stuttering looks like. It completely destroys competitive gameplay.Years ago, I used to have more "good gaming days." Nowadays, it feels like it only happens 1 out of 30 days, or even 1 out of 300. It’s incredibly rare to get a flawless session. Sometimes you can temporarily fix online issues by reducing Total Harmonic Distortion (THD), but the fix never lasts.For instance, you can temporarily clear static and residual energy from your PC by unplugging the PSU, removing the 24-pin motherboard cable, and holding down the case power button so the PSU capacitors drain the motherboard. You can then completely wipe the NVRAM by removing the CMOS battery and shorting the CLRTC jumpers. I only resort to this when my local power grid is so dirty that it corrupts the GPU or monitor OPROM, causing it to fail UEFI recognition. While this helps in both offline and online environments, it still doesn't permanently solve desync.If you look at professional gaming houses, they have a "perfect" environment. They use a high-end modem in bridge mode (zero NAT) paired with an enterprise-grade router running isolated VLANs and a single DHCP setup to completely avoid double NAT or CGNAT. They implement strict QoS/VLAN rules to prioritize game executables (hl.exe, hl2.exe, etc.) over any Windows background data.Furthermore, they operate on pristine electrical grids. Developed countries heavily regulate Total Harmonic Distortion (keeping it around 3-5%), whereas in Venezuela and other parts of LATAM, THD can easily spike to 15%, 20%, or even 30%. In environments like mine, you absolutely need a high-end, double-conversion online UPS with a strict 2% THD output, proper grounding that avoids daisy-chaining, and deep knowledge of your hardware limitations.To summarize, for a game to feel truly smooth today, we need:A strict 2-5% Total Harmonic Distortion (THD).Stable electrical voltage (110-120V or 220-240V).A rock-solid grid frequency that doesn't fluctuate (50Hz in Europe, 60Hz in LATAM). If you disassemble a monitor, the internal labels show their sweet spot for power frequency, usually allowing only a tiny 4% margin of error. If your power grid constantly fluctuates outside of that margin, it directly degrades the clarity of the image on your screen.It’s an incredibly complex mix of variables. Unfortunately, living here in Venezuela makes it very difficult and expensive to acquire the high-end equipment needed to test and isolate all of these electrical and network issues.It’s incredibly frustrating because it completely drains the fun out of gaming. I can’t even enjoy offline games anymore because I constantly notice the visual judder, micro-stutters, and desync.God bless you all. I'm not a religious person, but I just wanted to share my story and send some love to this community! <3"

PS: had to help myself with the AI fixing my Grammar, thanks again.

First, it is impressive that you have covered so many topics. I suffered for such a long time myself that I quit competitive gaming back in 2021. I know that feeling when everything is in sync and no one can touch you unless they outplay you. But while I agree with all your points on the network side (VLANs, QoS at the proper interface for outbound and inbound, and making sure your connection isn't CGNAT), I am not really convinced of this "dirty electricity" concept causing desync. Now, I am not saying it won't cause *any* issues, but if it did, you would know. Your clocks, voltage, BCLK, and in extreme cases, your monitor would fluctuate, and your frametimes would show symptoms of it at the very least.

Game netcodes have come a long way, so even subtle network jitter will only put you behind by one tick (which is still not great, but so much better than the desync described by most people here). While not every pro starts from a gaming house, I have known and seen a lot of players with incredible gaming experience. There are many layers where testing can go wrong, and a lot of popular guides already cover them.

Since you are already too deep into it:

Hardware Side:

CPU & RAM: Make sure your clocks are stable and that the system is not thermal or power throttling. Ensure your RAM is running in 1:1:1 (though technically, a 1:1 UCLK to MEMCLK ratio is more important, along with finding a stable FCLK for your system on Ryzen). Also, remember that finding stability depends heavily on your motherboard's VRM quality.
Network Adapter: A NIC with built-in hardware queues works best.

Measuring on the Windows Side:

ISR / DPC: LatencyMon. While a lot of people concentrate on the highest DPC values, the equal important thing is your average DPC values aswell.
RTTS & CapX: Use these for frametimes, 1% and 0.1% lows, and your game thread clock speed.

I will share one observation I made while trying to help a friend: his P-state in the BIOS was set to P-state 1, which played a huge role in poor 0.1% lows that weren't really visible on the monitor except for a few hitches. But it's not just the main game logic thread that spoils the experience; the network thread can also be affected, adding another layer to the poor experience.

WPT (Windows Performance Toolkit): This is the most interesting and revealing part if everything else looks fine. Here, you can see all your game threads and what cores are being used for them (checking whether some other process or app is using the same threads as the game). The tricky part is that while processes are named, game threads themselves are not explicitly named (they could be the game network thread, main game logic thread, or render thread), so you have to use logic to isolate them.

For example, to isolate and crosscheck the network thread, you have two options:

1. Under your game process threads, you can check for threads that have had interrupts matching your game server Hz (you would have interrupts every 8.3ms if its 120h server , so you check in a variance close to your network jitter, and you will find your network thread easily).
2. The easiest way is to look under your network driver process (usually NDIS), where you can see interrupts with connection IPs. Note the time and check which game thread matches that timing. This gives you a more realistic way to measure if a packet that hit your PC is being delayed or instantly delivered to your game process.

While WPT is really for advanced users, it will help you isolate problems specific to drivers, OS scheduling of your game process, network thread delays, ISR times, and saturation.

Network Side:

If you want a simple way without a lot of hassle, record a Wireshark capture while playing a game, then isolate the downstream of your game traffic and give it to Claude along with your game server tick rate so the AI can check for packet pacing jitter. It will tell you if feeling behind the server is due to network jitter, which causes the game client to set a high interpolation buffer. Of course, this will only tell you the downstream quality, not the upstream. If you want to measure both, iperf3 measures it somewhat accurately.

If you want to fully isolate it, we have to split it into:

1. Internet - Your WAN Device (Your First Router)
To test this, I personally use a 60Hz UDP test script I coded to test pacing from and to servers via my first WAN device. It helps me measure pacing, gaps, and packet batching. I run it on a server that mimics similar routing to and from game server locations (some say routing is different so it doesn't tell the whole story, but ignore them—most saturation occurs at the OLT collection point anyway, which is before peering; otherwise, gpns like ExitLag would have already fixed it). If this gives you perfect values (generally a good jitter below 2ms), then your internet connection is fine.

2. Your Local Setup (Router to PC)
If you have a router running a Linux OS like OpenWrt, it is easy to measure using tcpdump. The best run is when playing a game. Run tcpdump on the WAN interface and run Wireshark on your PC at the same time. Make sure you isolate your game connection either by IP or port before running it, otherwise it will be a mess.

Then, compare them both, find the packet ID, and crosscheck the timings. While you can't perfectly match timings (since there will be a slight difference the moment you start the test on both sides), the proper way to check is by analyzing the pacing through interpacket gaps, which will give you a clear picture of your local network quality and bufferbloat. Though remember, bufferbloat can also create queuing on the WAN interface itself for downstream traffic, so this isn't a completely reliable way to measure bufferbloat entirely.

But properfix for local bufferbloat would be Limiting your bandwidth with an qos and dscp values for traffic priority.

This post assumes familiarity with these technical terms, it's not aimed at basic users. I'd like to turn it into a full guide at some point, but that's a much bigger undertaking it needs a lot of background explanation for everyone who don't already have this knowledge, and that takes real time to put together properly.

Hello , i would like to know if changing my ethernet card "speed & duplex" from as example " 10gb/s to 100mb/s is the same as "limiting my bandwidth with a qos please"?
If it's not the same , what do you recommend as a bandwidth limit with the qos?

No. They are different layers and as far as my knowledge, lowering your Speed & Duplex will have no effect on gaming. Maybe if your rss isnt configured properly and you are saturating your speed on pc then setting it lower may help. But still we are talking about you saturating your speed like downloading or uploading heavy when playing which is rare case.

[quickmess]

My problem is i have a sub par cable internet connection verified by internet provider and my main games are cs2 and pubg.
Ive tested every tweak, all kinda different drivers, all bios settings from default to whatever and always doing fps benchmarks and feel test on same DM server.
I did some testing once again on ethernet properties and while i have tried all the main ones like interrupt moderation, flow control, disabling all offloads and while they do have some sort of effect but only in a way to make my game feel less stable where the sprays etc feel very off sometimes.
So i some tweak guides say just disable all everything except ipv4 / ipv6 and qos packet scheduler, i tested them all out and i disabled QOS packet scheduler completely leaving only ipv4 and ipv6 enabled and what i noticed is my input latency got much lower and even my mouse movement felt A LOT snappier on server the exact same as my own hosted server against bots in cs and even in pubg my movement and moving the mouse felt so much faster.
So care anybody try this tweak and report back if you can notice a difference ?
Disable QOS packet scheduler completely and tell me if you notice anything.