The in-game latency graph, for me, is nearly always 200ms, however, when looking at the latency outside of the game, it's usually around 180/185. What is the in-game graph actually showing?
"
|------------------------------------------------------------------------------------------|
| WinMTR statistics |
| Host - % | Sent | Recv | Best | Avrg | Wrst | Last |
|------------------------------------------------|------|------|------|------|------|------|
| 192.168.1.1 - 0 | 535 | 535 | 0 | 0 | 3 | 0 |
| 172.17.24.2 - 0 | 536 | 536 | 1 | 3 | 5 | 5 |
| 109.110.48.205 - 0 | 536 | 536 | 1 | 5 | 5 | 5 |
| Po1-20g.c65.vss.core.vl.podryad.tv - 0 | 536 | 536 | 4 | 8 | 8 | 8 |
| Po1-20g.br1.core.vl.podryad.tv - 0 | 536 | 536 | 4 | 8 | 8 | 4 |
| 83.169.195.61 - 0 | 536 | 536 | 16 | 19 | 20 | 20 |
| 83.169.204.89 - 0 | 536 | 536 | 180 | 183 | 183 | 183 |
|xe-0-0-2.bbr02.eq01.ams02.networklayer.com - 0 | 536 | 536 | 184 | 187 | 190 | 190 |
| ae7.bbr01.eq01.ams02.networklayer.com - 0 | 536 | 536 | 180 | 182 | 183 | 180 |
| ae0.bbr02.xn01.fra01.networklayer.com - 0 | 536 | 536 | 179 | 183 | 183 | 182 |
| ae6.dar01.fra02.networklayer.com - 0 | 536 | 536 | 180 | 182 | 184 | 181 |
| po1.fcr01b.fra02.networklayer.com - 0 | 536 | 536 | 181 | 183 | 183 | 182 |
| d0.52.7a9f.ip4.static.sl-reverse.com - 0 | 536 | 536 | 181 | 183 | 188 | 185 |
|________________________________________________|______|______|______|______|______|______|
"
1 <1 ms 1 ms 1 ms 192.168.1.1
2 4 ms 3 ms 4 ms 172.17.24.2
3 4 ms 3 ms 4 ms 109.110.48.205
4 6 ms 7 ms 8 ms Po1-20g.c65.vss.core.vl.podryad.tv [109.110.48.146]
5 7 ms 7 ms 8 ms Po1-20g.br1.core.vl.podryad.tv [109.110.48.145]
6 22 ms 17 ms 20 ms 83.169.195.61
7 181 ms 181 ms 181 ms 83.169.204.89
8 186 ms 185 ms 185 ms xe-0-0-2.bbr02.eq01.ams02.networklayer.com [50.97.16.42]
9 180 ms 181 ms 181 ms ae7.bbr01.eq01.ams02.networklayer.com [50.97.18.212]
10 180 ms 181 ms 181 ms ae0.bbr02.xn01.fra01.networklayer.com [50.97.18.217]
11 181 ms 181 ms 181 ms ae6.dar01.fra02.networklayer.com [50.97.19.145]
12 181 ms 181 ms 181 ms po1.fcr01b.fra02.networklayer.com [159.122.118.135]
13 181 ms 181 ms 182 ms d0.52.7a9f.ip4.static.sl-reverse.com [159.122.82.208]
"
Pinging 159.122.82.208 with 32 bytes of data:
Reply from 159.122.82.208: bytes=32 time=181ms TTL=44
Reply from 159.122.82.208: bytes=32 time=181ms TTL=44
Reply from 159.122.82.208: bytes=32 time=181ms TTL=44
Reply from 159.122.82.208: bytes=32 time=182ms TTL=44
Ping statistics for 159.122.82.208:
Packets: Sent = 4, Received = 4, Lost = 0 (0% loss),
Approximate round trip times in milli-seconds:
Minimum = 181ms, Maximum = 182ms, Average = 181ms
|
Posted byKellogon Nov 5, 2015, 9:37:24 PM
|
|
The in-game graph is actually the network latency + packet processing time.
They measure latency by sending packets to and from the server. These packets go onto the same queue as all the other packets being processed by the game, so it's a more true representation of actual game latency, not just network.
Network latency is only one portion of the total latency. This is why resource contention and CPU speed/load can actually affect the PoE latency, and why when we get latency posts, we look at more than just the network.
|
Posted byDrakieron Nov 5, 2015, 10:00:47 PMValued Poster
|
"
The in-game graph is actually the network latency + packet processing time.
They measure latency by sending packets to and from the server. These packets go onto the same queue as all the other packets being processed by the game, so it's a more true representation of actual game latency, not just network.
Network latency is only one portion of the total latency. This is why resource contention and CPU speed/load can actually affect the PoE latency, and why when we get latency posts, we look at more than just the network.
Thanks for the reply. The packet processing time, is it server side, client side or a mixture of both?
|
Posted byKellogon Nov 5, 2015, 10:04:39 PM
|
|
Both.
The server has to process the packet to send a reply, and the client has to process the packet to count the time it took.
|
Posted byDrakieron Nov 5, 2015, 10:20:22 PMValued Poster
|
You send Ping packets and receive Pong Packets.
Either the Ping or Pong packet include a ticker in ms.
Normally the Ping Packet includes the ticker and the Pong packet is a 1 byte packet.
Meaning you send something lie (Ping) 12345 and receive (Pong) the ticker stops then and the ticker whent to 12360 the difference gets calculated and that's your Ping in ms.
If the server doesn't constantly receives certain packets including the Ping packet, the server will close the connection.
This isn't limited to Ping packets because you can't be at the grace of one packet only.
If the Pong Packet actually sends you the ticker, the Ping Packet has more than one function.
E.g. Cheat Detection.
"
+ packet processing time.
Which is <1-3 ms unless your PC is under a lot of stress.
"
These packets go onto the same queue as all the other packets being processed by the game
Not necessarily. If the game uses serveral servers they can be processed by different servers.
E.G. Gameserver and Chatserver
If the game uses a launcher the launcher can be used to process the ping and what you see is only a memory read.
Many different Packets usually result in a split of servers and threads handling the packets to keep the game smooth.
|
Posted byHilberton Nov 6, 2015, 1:35:01 PM
|
"
Both.
The server has to process the packet to send a reply, and the client has to process the packet to count the time it took.
Just to finalise. In the previous example, the difference between raw latency and the in-game latency graph, is approximately 15/20ms, which you suggest is caused by packet processing overhead.
If I look at the same situation when playing on a local server, there appears to be little or no overhead...
"
|------------------------------------------------------------------------------------------|
| WinMTR statistics |
| Host - % | Sent | Recv | Best | Avrg | Wrst | Last |
|------------------------------------------------|------|------|------|------|------|------|
| 192.168.1.1 - 0 | 206 | 206 | 0 | 0 | 0 | 0 |
| 172.17.0.2 - 0 | 206 | 206 | 1 | 4 | 2 | 4 |
| 10.8.6.1 - 0 | 206 | 206 | 3 | 2 | 3 | 3 |
| 10.8.102.130 - 0 | 206 | 206 | 4 | 3 | 4 | 3 |
| 10.8.102.129 - 0 | 206 | 206 | 5 | 6 | 6 | 4 |
| 109.110.48.137 - 0 | 206 | 206 | 6 | 7 | 6 | 6 |
| 109.110.48.17 - 0 | 206 | 206 | 5 | 8 | 8 | 8 |
| vtc-vlv-ix.gw.vtc.ru - 0 | 206 | 206 | 6 | 7 | 8 | 7 |
| host-194-22.hosts.vtc.ru - 0 | 206 | 206 | 6 | 8 | 9 | 8 |
| host-167-34.hosts.vtc.ru - 0 | 206 | 206 | 6 | 9 | 9 | 9 |
Last edited by Kellog on Nov 6, 2015, 9:31:40 PM
|
Posted byKellogon Nov 6, 2015, 9:11:59 PM
|
It looks to me like this difference is related to Predictive mode...
In the previous example the server was set to Lockstep and I've tried several times on two different servers to see any significant difference between the latency displayed by the in-game graph and raw latency. However, Predictive mode shows an immediate discrepancy, which appears to be accentuated by distance.
"
------------------------------------------------------------------------------------------|
| WinMTR statistics |
| Host - % | Sent | Recv | Best | Avrg | Wrst | Last |
|------------------------------------------------|------|------|------|------|------|------|
| 192.168.1.1 - 0 | 128 | 128 | 0 | 0 | 0 | 0 |
| 172.17.0.2 - 0 | 128 | 128 | 2 | 3 | 4 | 2 |
| 10.8.6.1 - 0 | 128 | 128 | 3 | 6 | 6 | 6 |
| 10.8.102.130 - 0 | 128 | 128 | 5 | 7 | 6 | 5 |
| 10.8.102.129 - 0 | 128 | 128 | 5 | 8 | 9 | 9 |
| 109.110.48.137 - 0 | 128 | 128 | 5 | 8 | 9 | 8 |
| 109.110.48.17 - 0 | 128 | 128 | 5 | 10 | 11 | 10 |
| vvk06.transtelecom.net - 0 | 128 | 128 | 6 | 9 | 16 | 9 |
| 10.66.99.2 - 0 | 128 | 128 | 116 | 120 | 127 | 122 |
| Oversun-gw.transtelecom.net - 0 | 128 | 128 | 117 | 124 | 124 | 119 |
| 94.198.48.129 - 0 | 128 | 128 | 114 | 118 | 124 | 116 |
| 94.198.52.225 - 0 | 128 | 128 | 116 | 120 | 123 | 123 |
|________________________________________________|______|______|______|______|______|______|
|
Posted byKellogon Nov 7, 2015, 8:28:47 PM
|
Something else. This is using a different ISP. As you can see, there's no overhead shown on the in-game graph. Even running about, the in-game graph shows the latency to be lower by 10 to 15 ms almost constantly.
.
"
|------------------------------------------------------------------------------------------|
| WinMTR statistics |
| Host - % | Sent | Recv | Best | Avrg | Wrst | Last |
|------------------------------------------------|------|------|------|------|------|------|
| 192.168.255.1 - 0 | 122 | 122 | 0 | 0 | 1 | 0 |
| 10.251.134.241 - 0 | 122 | 122 | 3 | 3 | 8 | 3 |
| 10.251.134.240 - 0 | 122 | 122 | 3 | 3 | 28 | 28 |
| 95.167.95.233 - 0 | 122 | 122 | 155 | 156 | 164 | 157 |
| No response from host - 100 | 25 | 0 | 0 | 0 | 0 | 0 |
| ae0.bbr01.xn01.fra01.networklayer.com - 0 | 122 | 122 | 184 | 185 | 205 | 184 |
| ae5.dar01.fra02.networklayer.com - 0 | 122 | 122 | 184 | 185 | 193 | 186 |
| po1.fcr01b.fra02.networklayer.com - 0 | 122 | 122 | 185 | 186 | 194 | 185 |
| cc.52.7a9f.ip4.static.sl-reverse.com - 0 | 122 | 122 | 180 | 184 | 191 | 181 |
|________________________________________________|______|______|______|______|______|______|
|
Posted byKellogon Nov 26, 2015, 1:15:57 AM
|
