Another way to fix hitreg and card draws! Best settings!

[capy2008]

This are new method to improve hitreg to be perfect like shroud play or Ninja(pubg,fornite,realm royale) and card draws(hearthstone)

Posibble the best settings I found so far:

Atention, this is not the same settings as TcpAckFrequency, they act different!
Before you do this remove any tcpnodelay in msmq or any optimization you done to network card(TcpAckFrequency,MTU,DefaulTTL,TCPWindowSize and others)!!!

Code: Select all

Windows Registry Editor Version 5.00

[HKEY_LOCAL_MACHINE\SYSTEM\CurrentControlSet\Services\Tcpip\Parameters]
“DelayedAckFrequency”=dword:00000001
“DelayedAckTicks”=dword:00000001
“CongestionAlgorithm”=dword:00000001

You can try also this settings:

Code: Select all

Windows Registry Editor Version 5.00

[HKEY_LOCAL_MACHINE\SYSTEM\CurrentControlSet\Services\Tcpip\Parameters]
“DelayedAckFrequency”=dword:00000000
“DelayedAckTicks”=dword:00000000
“CongestionAlgorithm”=dword:00000000

[link removed by admin]

[capy2008]

Sadly my edit session expire , so i post here next tweak!

Another command you can try, not listed everywhere on internet, you find it only here exclusively and on my website( it is listed in my profile About me/Statistics/ )!

Code: Select all

Windows Registry Editor Version 5.00

[HKEY_LOCAL_MACHINE\SYSTEM\CurrentControlSet\Services\Tcpip\Parameters]
“MultihopSets”=dword:0000001e

MultihopSets = maximum hops / 15 decimal for fps games and 30 decimal for all games, depend on distance you can try different settings , I recommend to everyone to start with 30 and if it work very good leave it 30. You will fell move and shots more responsive!
DeafulTTL = fixed hops / Don't use DefaulTTL with MultihopSets, just delete the DefaultTTL from registry!

You can try MultihopSets work more better than DeafultTTL, and is more safe cause you can put it with value of 1

My name in games: GrandpaRO
Twich: Top1Genius

[Philip]

"MultihopSets" ? Where does this come from, is there some Microsoft reference for the setting? How do you know what works? If this is somehow related to EBGP Multi-Hop, and assuming it does anything, I think it is only related to BGP TTL security, not performance. Essentially, if you get packets with longer TTL value, they would get dropped. I could be wrong, I am just asking for more info.

[capy2008]

All settings are from reverse engineering to find what settings influence network , all settings are tested by me! I do this after I beg microsoft support to help me, simply their ignorance push me to find this by my self and help everybody to tweak settings on our own! In my opinion this is the freedom microsoft should give to us!

[Philip]

"reverse engineering" is fine, but how do you know exactly what it changes/means and what the valid ranges are?
What Operating System exactly? What happens with/without the setting, what are the results?
In general, a bit more detail may help others try to test and see if they can repeat and confirm your results.

[capy2008]

Tested on Windows 10 x64 1803
All this settings are forced by windows

Like i think you know, this settings in my opinion change values for all windows congestion templates, most of the settings are forced by windows for cmd and powershell but you can manipulate them from registry!

CongestionAlgorithm - from microsoft
Reserved for future use.
If you are an independent hardware vendor (IHV) and you want to implement nondefault congestion control algorithms, for example Compound TCP, contact Microsoft at External TCP Offload Triage.

-DelayedAckFrequency - from microsoft
Gets and sets number of unresolved acknowledgments (ACK).

-DelayedAckTicks - from microsoft
Gets and sets the TCP delayed ACK timeout, in milliseconds.

- MultihopSets
This set maximum ttl value


On games seems faster with this settings(snnapy aim and movement like lan feeling), it also fix hitreg issue(other people think it is bloom).
I tested only with games
I need more time for complex tests, maybe tomorrow!

[Philip]

I know about the first three, I am just not sure that "MultihopSets" value simply sets max TTL. And TTL only sets the max value before packets get discarded, if you are within those 15 hops/seconds, it does not affect anything else. I think Multihopsets may be something else and not TTL though, like EBGP Multi-Hop.

[zimessi]

what is ur MTU? capy2008...you use some program to optimize ur pc?

[capy2008]

I don't have any programs, I use only this 3 parameters DelayedAckFrequency, DelayedAckTicks, CongestionAlgorithm and MTU is 1500 everytime. For me, mtu not have any change to set it lower!

[zimessi]

Interfaces \ {NIC ID} add some parameter ? capy2008 ur fix is nice!!!.----- algun other parameter for fix!!

[capy2008]

I test many combinations with this settings but nothing work, this settings (DelayedAckFrequency, DelayedAckTicks, CongestionAlgorithm) work on clean install without any optimizations! Take care not add DelayedAckFrequency, DelayedAckTicks, CongestionAlgorithm in Interfaces \ {NIC ID}, they should go only in HKEY_LOCAL_MACHINE\SYSTEM\CurrentControlSet\Services\Tcpip\Parameters to have effect!

You must understand i don't use MultihopSets, I add here just like another tip for tweak network, everyone decide it how is better for him!

[st1cky]

I test many combinations with this settings but nothing work, this settings (DelayedAckFrequency, DelayedAckTicks, CongestionAlgorithm) work on clean install without any optimizations! Take care not add DelayedAckFrequency, DelayedAckTicks, CongestionAlgorithm in Interfaces \ {NIC ID}, they should go only in HKEY_LOCAL_MACHINE\SYSTEM\CurrentControlSet\Services\Tcpip\Parameters to have effect!

You must understand i don't use MultihopSets, I add here just like another tip for tweak network, everyone decide it how is better for him!

I tested it, no improvement.

I was changing that behavior already on Windows Server Edition 2016, these Values are not protected.
I had no improvement, but maybe because bad ISP or Copper Line in the Ground. ( Villiageline )

I was running some One-Click Optimizer for Testing purposes, but i did ->

TCP Default Settings how you recommendet and ->

netsh int ipv4 reset
netsh int tcp reset
netsh int ipv6 reset
netsh int httpstunnel reset
netsh int portproxy reset
netsh advfirewall reset
netsh winsock reset
netsh http flush logbuffer
netsh nap reset configuration
netsh branchcache reset
netsh lan reconnect
ipconfig /flushdns

But it wasn't working, i will try maybe tomorrow with fresh Installation.

How did you test it, for working?

Because Powershell shows old values, you just play Online-Games and feel a difference?

In General whats the best way to check Tweaks of Performance? ( Registry-Tweaks,... )
And im always asking my self, why isnt it possible to see all DWORDS in the Registry, i mean there are hidden ones running in the background. I asked Microsoft they said it isnt possible but i dont believe them, because the engineering crew of Microsoft definitely got some ways to see them.



@Capy thanks for your reply, i will test it tomorrow. ( ****ing Microsoft, even support is so lame. There is not much information, about Settings up Windows for prefect gaming. Just a lot Guides in the Internet but not everything is right )

[capy2008]

I tested it, no improvement.

I was changing that behavior already on Windows Server Edition 2016, these Values are not protected.
I had no improvement, but maybe because bad ISP or Copper Line in the Ground. ( Villiageline )

I was running some One-Click Optimizer for Testing purposes, but i did ->

TCP Default Settings how you recommendet and ->

netsh int ipv4 reset
netsh int tcp reset
netsh int ipv6 reset
netsh int httpstunnel reset
netsh int portproxy reset
netsh advfirewall reset
netsh winsock reset
netsh http flush logbuffer
netsh nap reset configuration
netsh branchcache reset
netsh lan reconnect
ipconfig /flushdns

But it wasn't working, i will try maybe tomorrow with fresh Installation.

How did you test it, for working?

Because Powershell shows old values, you just play Online-Games and feel a difference?

In General whats the best way to check Tweaks of Performance? ( Registry-Tweaks,... )
And im always asking my self, why isnt it possible to see all DWORDS in the Registry, i mean there are hidden ones running in the background. I asked Microsoft they said it isnt possible but i dont believe them, because the engineering crew of Microsoft definitely got some ways to see them.

Try on clean windows install, you can't see information of commands changes in cmd or powershell like ttl value, if you change in registry ttl to 64 in cmd you still see 128 value.
What I observe, windows put some protection on this values, if you change them more than 3-4 times, windows will reset value then lock value, this why you need to reinstall windows.

[capy2008]

Still work!

[st1cky]

Still work!

What Operating System you currently use?
What Mainboard you have (AMD/INTEL)?
What Drivers?
What Tweaks you Applied?

Greetings St1cky

[capy2008]

If you really want to tweak PC for best experience possible you need to have something like this in latency moon after hours of gaming, should be stable and almost same values! Fast internet not depend of only network tweaks,you need to proper installation drivers and good windows troubleshooting with windows performance analyzer to fix all problems. Do this and if graph look like mine i guaranteed 100% you will have best experience possible. How I do this? Learn and practice!

[Philip]

Those measurements seem to be in mu-seconds (millionth of a second). That is, 1000 μs = 1 millisecond... all your measurements are less than 0.4 milliseconds, that doesn't have much relevance onto the internet, since any switch/router you go through will add 5 times as much latency, and you will easily have 50+ times that latency when gaming online (just 20 milliseconds ping is 50 times, 5000% the latency you are showing).

In general, the median human reaction time is 200+ milliseconds. You will be hard pressed to find anyone that has less than 100 milliseconds reaction time. We try to tweak to reduce gaming latency by a few milliseconds, but, I am not sure why those local 0.4 milliseconds make any perceivable difference at all.

[capy2008]

Philip, can you explain to me ,why when you have high values, you have sound problems in games, or lag or fps drop? Here is not only about measurements of mu-seconds, this is about a good health and work of drivers. In teory you have right with measurements but if you have already high mu-seconds for drivers in you pc this things add even more latency + router add 5 times more and other hops. Then I ask you, it is good to have on pc good latency to sustain other works? Reaction time have nothing to do with speed of sound or video processing or computer processing , in back stage things work more crazy faster than what we see on monitor. Let's say we make a program how make a calculation, you hit a run debug button and the program instant show you the result of calculation, you can see that calculation? No, this why you must have other things work smooth in back stage to offer you a good experience with what you real see without a big delay on monitor.

[Philip]

I agree it would show if you have some real issues with your drivers. I was just trying to point out a couple hundred mu-seconds would be unperceivable.

[st1cky]

My DPC is under 10µs and i already know LatencyMon, still doesnt fix anything Internet related.

[st1cky]

Has nothing to do with eyes can see, its the responsiveness from the System. Mouse is smoother.

You cant see but you can Feel

[st1cky]

[capy2008]

You have this wile you play all the time? Just do 1 hour gaming with latencymon open and post screenshot. Also take care of that processor he will go really hot. Reduce in bios ppl oc voltage to 1.100.
.

[st1cky]

You have this wile you play all the time? Just do 1 hour gaming with latencymon open and post screenshot. Also take care of that processor he will go really hot. Reduce in bios ppl oc voltage to 1.100.
.

Of course its almost impossible to have DPC like under 10µs, when you start playing games. But for a good start its prefect when its under 10-20µs.
Next Time i play i will give you feedback.

2 Screenshots with Stresstest of DPC Latency ( CPU Only ).


CPU STRESSTEST



CPU NO STRESSTEST

[capy2008]

This tests don't have any relevance, just do test while gaming!

[st1cky]

I don't agree with you, when you play Games like PUBG or CSGO PC is not Fully occupied.

Anyway here is Screenshot Ingame.

[capy2008]

What is occupied? You don't need to see processor performance here, you must see drivers performance how can affect you internet data transfer! Make a screenshot to device manager with all installed drivers ! You have ahci driver installed? I want a screenshot or regfiles for tcpip paramenters, interface parameters, afd, lanmanserver and PriorityControl.

[st1cky]

What is occupied? You don't need to see processor performance here, you must see drivers performance how can affect you internet data transfer! Make a screenshot to device manager with all installed drivers ! You have ahci driver installed? I want a screenshot or regfiles for tcpip paramenters, interface parameters, afd, lanmanserver and PriorityControl.

If you do stress tests your DPC increase, thats what i mean. Because with some games your PC is not using all Resources. So you mean all Drivers affect internet data transfer? I was testing... normally i use Windows default Drivers.

you must see drivers performance how can affect you internet data transfer

How can i check that? Because DPC Latency is checking kernel timer latencies...only?

You have ahci driver installed?

I'm still testing between them.
- Intel(R) 300 Series Chipset Family SATA AHCI Controller [NOW]
- Intel(R) 200 Series Chipset Family SATA AHCI Controller [DEFAULT Driver from ASUS]
- Standard SATA AHCI Controller [WINDOWS DEFAULT]


Screenshots, i changed some values ->

https://imgur.com/a/7qB8EjA


I am currently sick so it may take sometime to answer.

[capy2008]

If you do stress tests your DPC increase, thats what i mean. Because with some games your PC is not using all Resources. So you mean all Drivers affect internet data transfer? I was testing... normally i use Windows default Drivers.


How can i check that? Because DPC Latency is checking kernel timer latencies...only?


I'm still testing between them.
- Intel(R) 300 Series Chipset Family SATA AHCI Controller [NOW]
- Intel(R) 200 Series Chipset Family SATA AHCI Controller [DEFAULT Driver from ASUS]
- Standard SATA AHCI Controller [WINDOWS DEFAULT]


Screenshots, i changed some values ->

https://imgur.com/a/7qB8EjA


I am currently sick so it may take sometime to answer.

Use Intel(R) 300 Series Chipset Family SATA AHCI Controller , it is the best driver, but do manually install and uninstall other drivers saved! I suggest only use provider drivers! Just clear any tweaks to tcpip cause when you test many things you can't find if driver is good or not. I don't use any tweaks now! My network card is TG-3468, for this card you can use this settings also in

Code: Select all

Windows Registry Editor Version 5.00
[HKEY_LOCAL_MACHINE\SYSTEM\CurrentControlSet\Control\Class\{4d36e972-e325-11ce-bfc1-08002be10318}\0002\Ndi\Interfaces]
"UpperRange"="ndis4"
"LowerRange"="broadband"

More information here https://docs.microsoft.com/en-us/window ... interfaces

[st1cky]

Use Intel(R) 300 Series Chipset Family SATA AHCI Controller , it is the best driver, but do manually install and uninstall other drivers saved! I suggest only use provider drivers! Just clear any tweaks to tcpip cause when you test many things you can't find if driver is good or not. I don't use any tweaks now! My network card is TG-3468, for this card you can use this settings also in

Code: Select all

Windows Registry Editor Version 5.00
[HKEY_LOCAL_MACHINE\SYSTEM\CurrentControlSet\Control\Class\{4d36e972-e325-11ce-bfc1-08002be10318}\0002\Ndi\Interfaces]
"UpperRange"="ndis4"
"LowerRange"="broadband"

More information here https://docs.microsoft.com/en-us/window ... interfaces

Thank you for providing Information, One Question here. Do you know how to use RSSv2 with NDIS 6.x?

LowerRange= ? broadband

Lower edge for a broadband adapter. Should't it be Ethernet?

UpperRange= ? ( ndis2, ndis3, and ndis4 should no longer be used) Safe to use it?

My currently Default Settings are.[sstpminiport] ( I guess its the wrong Path, because its not my Ethernet Adapter )
LowerRange: nolower
UpperRange: ndiscowan

I need to use

Code: Select all

KEY_LOCAL_MACHINE\SYSTEM\CurrentControlSet\Control\Class\{4d36e972-e325-11ce-bfc1-08002be10318}\0001\Ndi\Interfaces

Or im wrong here?

Just clear any tweaks to tcpip

You mean Interf & Para. ?

[capy2008]

Thank you for providing Information, One Question here. Do you know how to use RSSv2 with NDIS 6.x?

LowerRange= ? broadband

Lower edge for a broadband adapter. Should't it be Ethernet?

UpperRange= ? ( ndis2, ndis3, and ndis4 should no longer be used) Safe to use it?

My currently Default Settings are.[sstpminiport] ( I guess its the wrong Path, because its not my Ethernet Adapter )
LowerRange: nolower
UpperRange: ndiscowan

I need to use

Code: Select all

KEY_LOCAL_MACHINE\SYSTEM\CurrentControlSet\Control\Class\{4d36e972-e325-11ce-bfc1-08002be10318}\0001\Ndi\Interfaces

Or im wrong here?


You mean Interf & Para. ?

Settings are safe to use. Just as a attention, if you test other settings ndis 3,2 or nolower, network card can be disappear from network connection, but it is just visual,you can still find it in device manager and make from there changes, network card will work just normal, if you do this and you want to go back just change settings back and will appear.
You must find you network card like this

Yes, it is better to have clean install and test each settings one by one, cause don't have any sense to use settings how not change anything! If is not too much for you ,you save to usb, reg files with you tweaks and test with new windows install what we talk here.

[capy2008]

For Win32PrioritySeparation you can use this as guide.
I use 28 hex, 40 decimal

2A Hex = Short, Fixed , High foreground boost.
29 Hex = Short, Fixed , Medium foreground boost.
28 Hex = Short, Fixed , No foreground boost.

26 Hex = Short, Variable , High foreground boost.
25 Hex = Short, Variable , Medium foreground boost.
24 Hex = Short, Variable , No foreground boost.

1A Hex = Long, Fixed, High foreground boost.
19 Hex = Long, Fixed, Medium foreground boost.
18 Hex = Long, Fixed, No foreground boost.

16 Hex = Long, Variable, High foreground boost.
15 Hex = Long, Variable, Medium foreground boost.
14 Hex = Long, Variable, No foreground boost.

Code: Select all

Win32PrioritySeparation   Foreground   Background
0x28, 0x29, 0x2A                   18          18
0x18, 0x19, 0x1A                   36          36
0x24                               6            6
0x25, 0x14                         12           6
0x26                               18           6
0x15                               24           6
0x16                               36           6

[n1kobg]

St1ky, I send you on private message several programs, whith one of them you can see the hidden parts of the Registry. GMP
https://www.wagnardsoft.com/forums/view ... 0df880bc95

TTL affects internet when browzing. Not sure about gaming. I stopped using Nagles Algoritm Tweak. It can only messed up my settings with combination with something else. There is no difference on the games I play.

[limeay]

Wait what? a new way to configure...?

[n1kobg]

" If is not too much for you ,you save to usb"

I feel st1cky can handle more than what you show him here. Dont'wory he can keep up

Ofcourse ACHI is the best option. Its nobrainer.
About these tweaks in TCPIP. They are not in the Registry anymore for a reason. Its handled by the drivers. And while you can override some of them from the Regitry most of the times drivers just ignore them.

Just random link:

https://docs.microsoft.com/en-us/window ... parameters

For Win32PrioritySeparation you can use Bitsum's TweakScheduler as well.

BTW LanmanWorkstation tweaks have nothing to do with gaming. I disable the service. If you dont have workstation enabled the tweaks are not relevant anyway.

[n1kobg]

Hi there st1cky, Its good thing you post these screenshots for comparisson.
These are some of miine:



https://gyazo.com/6663fee24091ca133829abc033858f02


Do you use Defender? It seems to have noticable impact in FPS shooters. I see my frametimes are lower so im asking whats differend in your Windows. For 500fps you should have 2ms average if you do the math.
I added option in G-Booster to "pause" Defender when gaming.

[sonexs]

Hi there st1cky, Its good thing you post these screenshots for comparisson.
These are some of miine:



https://gyazo.com/6663fee24091ca133829abc033858f02


Do you use Defender? It seems to have noticable impact in FPS shooters. I see my frametimes are lower so im asking whats differend in your Windows. For 500fps you should have 2ms average if you do the math.
I added option in G-Booster to "pause" Defender when gaming.

Hello N1ko,

heres mine https://gyazo.com/ed61be976c8d975f97d6236cf5b520ed

i have 240hz zowie benq monitor

what does "max frametime" mean?
is higher the better, or lower the better?

[yarr]

will any of these work with Intel(R) Ethernet Connection (2) I218-V?

[n1kobg]

I dont think TcpAckFrequency works in Windows 10, maybe 8 or 7. Not going to work as far as Im concern.

[n1kobg]

The lower frame times the better. Thats the time between every frame. Lower means smoother picture & lower input lag.
4ms one is with K-Booster. Not sure if it helped or its just benchmark variation. My card is bios flashed anyway.