Page 3 of 20

Re: I finally found the reason behind my input lag after 6 years

Posted: 03 Nov 2022, 20:44
by Chief Blur Buster
Patrykelele wrote:
02 Nov 2022, 20:37
You are 100% right.
I have GPON internet provided by local ISP. They are monopolists, I don't have any other company that provides good internet(10Mbs+). They are using a bunch of traffic algorithms. As you can read in their statute "In order to maintain optimal conditions for Internet access, the Operator applies traffic monitoring and traffic management measures, such as: queuing, prioritization, reservation, temporary transmission allocation, traffic transfers and avoiding congestion. These measures are applied for no longer than It is necessary. The operator allows the use of management measures traffic based on defined traffic categories only in the case of w which proves to be technically unavoidable.". It makes my game terrible. 95% people having internet with perfect speed, latency still have problems due to above algorithims.
They also use some sort of speed stabilization. doesn't mater which hour, which day the speed stay consistent and on the same perfect speed. In my case it's 300Mb/s, always the same value. I don't belive in this. They have to do some tricks...

When I use lte the game is 10 times better than on gpon.
It's not just the ISPs but the playing-field-levelling algorithms by games.

In certain games:
FTTH users are often matchmakered with similar low-ping users.
LTE users are often matchmakered with similar mobile users (or by mobile-users IP address ranges etc)

If your FTTH trafficshaping is worse than the other FTTH gamers on the same matchmaker'd server, you lose.
If your LTE is worse than the other LTE gamers on the same matchmaker'd server, you lose.
Etc.

The matchmaking/compensation algorithms in games are a black box, and they annoyingly vary between the games. Sometimes you need slightly higher network latency to bypass the LPB-handicap algorithms etc in order to gain a competitive advantage -- annoyingly so.

And sometimes least penalty occurs when you're using a median-latency network. Low-latency players are often artificially handicapped (with various flawed latency compensation algorithms) and high-latency players are often artificially helped, to try to level the playing field.

Playing the VPN roulette can occasionally help, e.g. FTTH+VPN (a good paid one, preferably WireGuard with low-latency-jitter) to a remote server to add approximately +20ms or +30ms latency (but try to keep +/- 0ms jitter) can perform better than FTTH to the nearest server, for example! But it's ISP-dependant, VPN-dependant, location-dependant.

Low jitter is quite important, and that's where business fiber subscription (3x more expensive) can sometimes help over residential/home FTTH. Bypass the traffic shaping as much you can. Plus also concurrently getting the best VPN money can buy to add an exact artificial latency offset, to try to get you a very good sweet spot on certain games. A slightly-higher-network-latency zero-jitter connection seems to produce some really interesting improvements for many gamers.

Some games will score best at lowest network latency, and some games will score best at intentionally higher network latency.

Trying to zero-out network latency as much as you can, neglects to consider the infernal playing-field-levelling algorithms built into games (whether be latency compensation quirks or matchmaking quirks, or other).

And it's all a proprietary black box, how the games matchmaker/lag-compensate you, and how buggy or good these algorithms are.

The ISP is definitely a big factor too, but definitely not the only one...

It's crapshoot, really.

Re: I finally found the reason behind my input lag after 6 years

Posted: 03 Nov 2022, 22:43
by Anonymous768119
[one sentence deleted by moderator] If you experience stutters every time you see the enemy it proves that there is big discrepancy between what happens on your screen and what happens on a server. Play offline against bots on any map and it should give you clear singal that it's not about rendering or any other process related to GPU or CPU. If you're not synced properly with a server, which means movement of your model on the server and model on the screen do not overlap as close to each other as possible, it will lead to an avalanche of undesirable effects like inexplicably strange recoil control, impossibility of accurately locating the opponent by the sound, freezes affecting your crosshair movement, impossibility to track and shoot at moving enemies, impossibility to give one taps BECAUSE YOU RECEIVE DELAYED INFORMATION ABOUT ENEMY LOCATION.

Now explain to me, how is it possible, that at first I see the animation of my opponent dying and then the last shot that killed him?
Because I killed him ON THE CLIENT but the information from the server arrived much later. On a decent internet connection both things happen nearly in the same time and you shouldn't be even able to notice it. While having low ping it's not normal in CS:GO that you kill just after releasing mouse button after spray.

If you can, try to insist on your ISP to finally upgrade their networking equipment or move to Denmark or Sweden.

Re: I finally found the reason behind my input lag after 6 years

Posted: 04 Nov 2022, 00:29
by InputLagger
a_c_r_e_a_l wrote:
03 Nov 2022, 22:43
If you're not synced properly with a server, which means movement of your model on the server and model on the screen do not overlap as close to each other as possible, it will lead to an avalanche of undesirable effects like inexplicably strange recoil control, impossibility of accurately locating the opponent by the sound, freezes affecting your crosshair movement, impossibility to track and shoot at moving enemies, impossibility to give one taps BECAUSE YOU RECEIVE DELAYED INFORMATION ABOUT ENEMY LOCATION.

Because I killed him ON THE CLIENT but the information from the server arrived much later.
But, maybe the local PC plays a huge role for this, like it struggle to process information for send / receive to/from the remote server

Re: I finally found the reason behind my input lag after 6 years

Posted: 04 Nov 2022, 08:36
by Anonymous768119
InputLagger wrote:
04 Nov 2022, 00:29
a_c_r_e_a_l wrote:
03 Nov 2022, 22:43
If you're not synced properly with a server, which means movement of your model on the server and model on the screen do not overlap as close to each other as possible, it will lead to an avalanche of undesirable effects like inexplicably strange recoil control, impossibility of accurately locating the opponent by the sound, freezes affecting your crosshair movement, impossibility to track and shoot at moving enemies, impossibility to give one taps BECAUSE YOU RECEIVE DELAYED INFORMATION ABOUT ENEMY LOCATION.

Because I killed him ON THE CLIENT but the information from the server arrived much later.
But, maybe the local PC plays a huge role for this, like it struggle to process information for send / receive to/from the remote server
Measured in hundreds of milisecond? That would be sick system instability.

Re: I finally found the reason behind my input lag after 6 years

Posted: 04 Nov 2022, 09:51
by Wayzer
Yea I have literally the same issue. Tried 3 different ISPs in my flat - no success. Idk what to do.

Re: I finally found the reason behind my input lag after 6 years

Posted: 04 Nov 2022, 10:39
by dervu
The worst thing is that noone really knows at what level you should measure those delays to prove anything. Where to look? Network - noone found anything so far? Hardware stats - noone found anything. Do we need to go into some engineer level access to hardware or ISP or what?

Re: I finally found the reason behind my input lag after 6 years

Posted: 04 Nov 2022, 12:16
by 1000WATT
dervu wrote:
04 Nov 2022, 10:39
The worst thing is that noone really knows at what level you should measure those delays to prove anything. Where to look? Network - noone found anything so far? Hardware stats - noone found anything. Do we need to go into some engineer level access to hardware or ISP or what?
https://www.waveform.com/tools/bufferbloat
Everything is simple. If you don't get results like this guy.
https://forum.openwrt.org/t/rpi4b-with- ... e/87898/16
Either problems in your network hardware, or in the provider's hardware.

Re: I finally found the reason behind my input lag after 6 years

Posted: 04 Nov 2022, 15:58
by espresso
1000WATT wrote:
04 Nov 2022, 12:16
dervu wrote:
04 Nov 2022, 10:39
The worst thing is that noone really knows at what level you should measure those delays to prove anything. Where to look? Network - noone found anything so far? Hardware stats - noone found anything. Do we need to go into some engineer level access to hardware or ISP or what?
https://www.waveform.com/tools/bufferbloat
Everything is simple. If you don't get results like this guy.
https://forum.openwrt.org/t/rpi4b-with- ... e/87898/16
Either problems in your network hardware, or in the provider's hardware.
While my results seems rather underwhelming, i wonder how it would go if the test only uses 1% of the 1Gbit line. Using 100% of the 1Gbit line most certainly reaches the limits somewhere.

Re: I finally found the reason behind my input lag after 6 years

Posted: 05 Nov 2022, 12:45
by triplese
1000WATT wrote:
04 Nov 2022, 12:16
https://www.waveform.com/tools/bufferbloat
Everything is simple. If you don't get results like this guy.
https://forum.openwrt.org/t/rpi4b-with- ... e/87898/16
Either problems in your network hardware, or in the provider's hardware.
Bufferbloat test doesnt show anything related to gaming traffic, not even close.
Its saturating your link fully, while gaming not saturating even to 10 mbit/s.
Here my test, but faceit DE servers is literally unplayable - I'm almost 1 sec behind server. Any other servers is OK.
So its clearly congestion on route to faceit DE and nobody can fix it without ISP, they need buying more bandwidth, but its VERY expensive.
https://www.waveform.com/tools/bufferbl ... b0d6ca5bfd

Re: I finally found the reason behind my input lag after 6 years

Posted: 05 Nov 2022, 13:08
by dervu
triplese wrote:
05 Nov 2022, 12:45
1000WATT wrote:
04 Nov 2022, 12:16
https://www.waveform.com/tools/bufferbloat
Everything is simple. If you don't get results like this guy.
https://forum.openwrt.org/t/rpi4b-with- ... e/87898/16
Either problems in your network hardware, or in the provider's hardware.
Bufferbloat test doesnt show anything related to gaming traffic, not even close.
Its saturating your link fully, while gaming not saturating even to 10 mbit/s.
Here my test, but faceit DE servers is literally unplayable - I'm almost 1 sec behind server. Any other servers is OK.
So its clearly congestion on route to faceit DE and nobody can fix it without ISP, they need buying more bandwidth, but its VERY expensive.
https://www.waveform.com/tools/bufferbl ... b0d6ca5bfd

So you get no signs of loss either in game or pinging those servers, but game feels worse.
There are no signs it being worse in network statistics compared to other servers?
How can this crowding be affecting game when packets are not affected according to stats?
How can they be slowed if there are no signs of it?

I think only way to find out what is happening is to find someone else without issue like yours, but using same routing, same ISP or not.