Jump to content


  • Content Count

  • Joined

  • Last visited

  • Days Won


Everything posted by Bert

  1. These ports mentioned in the starting post don't work. I have UDP SRC 62000-64000 and DST 31000-45000 and the reverse rule into QoS at the moment. For some odd reason I found some communication with the datacenter (in my case Amazon Singapore) at 30110 and 30120 while just sitting in the lobby. With the above ranges it turns on QoS only from the moment it finds a lobby until you return to the lobby. There it stays on for a minute until it doesn't detect any traffic and then switches off. I can't say that it makes a difference though.
  2. Here in Asia you litterally can't play CoD without a Netduma. Well you can but you get connected up to Japan and then it's game over. There is a second of delay between shot and hitmarker LOL.
  3. I use internet sharing and run my console through my PC. But I think you can make it work by renaming the executable file. From what I see you're better off prioritizing something like source 60.000-65.000 destination 30.000-40.000. Better is the actual ports but you have to change your QoS every time you enter a new lobby. I checked on my setup with the default QoS option and the hitreg is really fine. If you experience delays it's in the game not in your router.
  4. These ports quoted technically can't even work. Whenever I watch MW traffic on wireshark I get a source port in the 60.000 range and a destination in the 30.000 range. They are changing every time however source stays the same every time you start the game. So for gameplay traffic implementing 3074 in your ports doesn't seem to make a difference. I think it's only used by Demonware for matchmaking. That's why I just stick with the Games Console classification if it's for PS4 / Xbox.
  5. Bert


    What type of router is it and sort of CPU does it have? CAKE is pretty hardware intensive.
  6. Well depending on your home network you might not need it. Ground war is a bit wonky since the servers run 20Hz. That means that update rates from the server are lower. So you will always have a longer delay between shot and hitmarker.
  7. Issue with MW is that the ports for game traffic keep changing. I have just set it to DumaOS Games console under QoS but for PC there seem to be some issues. Port forwarding doesn't affect latency or hitdetection.
  8. I think you both had the same thought LOL.
  9. They have always done that. So many times when I quit a match they just toss you back in the same lobby. You can run and gun just fine in this game but the more slow playstyle currently has a lot of advantages. However. There is probably 10+ ways a player can improve himself. Most people just stand opposite of eachother, shoot, or strafe a bit and shoot, or at best use cover. If you're in a 1:1 situation you need to figure out other tactics than shoot and aim for the best.
  10. On BO4 for comparison. Playing from Norway at 47ms in game ping: 3-4 frames delay, 100-133ms. Went 31-12 in that game. Lots or foreign players from all over EU though. (And this is on a really good connection, enterprise router with private lease line, there is no fluctuation in in game ping whatsoever) Playing from my place in Asia 55-60ms in game ping: 2 frames delay. Went 42-6 in that game. In the first game my connection felt godlike though although after playing back the footage it suggests that my home connection in Asia is better.
  11. IMHO if you add asymmetric latency on up and download. Adding 50ms of ping in both directions doesn't alter much. I have been testing and looking at others. On YT, clips have a 30FPS format. so we can assume there is 33ms between frames. My own connection, no artificial delay, 60ms in game latency gives a hitmarker after 2 frames, ie 66ms (+/- 33ms). This is what you would expect. Add another 60ms artificial latency so we show 120ms in game, and measure the same, I get a hitmarker after 4 frames, ie 133ms (+/-33ms) also pretty much what you would expect. I played back a game from when I was in Europe a few weeks ago where I show roughly 25ms in game, 8ms on Duma OS and my shots land on the next frame, ie below <33ms. That is what you would expect it to be. 120ms is perfectly playable by the way you really only notice this when you start measuring or if you are a pro from a hitdetection point of view. I just need to find a game with high latency users now and see if it changes, ie are connections equalized or not. Because this is all measured in games where my connection was good. You can replicate this by connecting to a far away server that gives you higher ping but the issue is that there is more hops in between that can mess with your traffic so getting a server close and adding artificial latecy gives a far more accurate result. Like from the past I know that the Japanese server often has huge unexplainable delays. Ping in game in both MW and BO4 are measured directly off the in game UDP packets so is by far the most accurate. The server itself doesn't actually answer ICMP requests so I am guessing that DumaOS goes off the last hop it can get a ICMP response from. These measurements are very rough, ideally you need a capture card. PS4 and YT footage is recorded at 30FPS but 60FPS would produce a more accurate result. I have looked at others as well. On YT, if you pause a video you can play them frame by frame by pressing , and . on the keyboard. Korean Savage has 3-4 frames delay and on Apathy BZ I can see a 3-4 frame delay as well. These are guys I expect to play on a really good connection. What actually surprises me is how good first shot accuracy on Korean Savage is. He almost never misses the first shot. But it indicates a latency of 100-133ms for them, either they are beeing compensated against or their connections isn't as good as you think. My best game yesterday was on regular 40ms yesterday though (60ms in game) so really I don't think it helps much. I think it might just cause the game to be visually more fair when you are at a slightly higher latency, closer to the lobby average. Ie the guy coming around the corner laserbeaming you in a nanosecond would likely still have had you if your ping was higher. Just the killcam would have looked more decent. Conenction balancing would make the game fair but it does obviously affect the quality of the game for low ping users.
  12. I would be careful setting a hard latency value in your router. Because if you do find a low latency lobby you are severely disadvantaged.
  13. The camera's are off in this game. If you peak try to peak around the right side of a corner, because that has the advantage.
  14. LOL that is pretty much a god lobby you have right there. My alt account is level 35 or so and even on that almost all my enemies are level 155. Post a video of a game like that.
  15. If you want to run QoS / SQM then you would want to look into the Edgerouter 4 or higher. 4 doesn't have a switching chip, the 6 and up have I thought. ER-X doesn't really have enough CPU power for high bandwidth connections. I looked at this to run in tandem with my R1 but to be honest I think it's all pissing in the wind. Connect to the closest server and there is not much else you can do. Unless you really need to utilize QoS etc.
  16. You use a PC and a console. The PC needs to have 2 network ports or wifi. You connect 1 port to the Netduma and 1 port to your console. Then turn on Internet Connection sharing. In DumaOS call your PC a games console and add it to the geofilter, port forwarding etc. Forward ports in windows ICS and voila. Now you can use wireshark to capture traffic and you can also run a connection emulator to add additional latency. I use SoftPerfect Connection Emulator. It lets you target a specific protocol like UDP and a specific port if you want to. But CoD MW seems to be changing it's ports every time so I just delay all UDP. You can adjust the latency values while it's running.
  17. I haven't tried this for MW yet, forcing P2P. But we used to do it for WW2 in Asia. And you always had faster hitdetection on a 120ms p2p host compared to the Dedicated server at 107ms. That suggests that this latency balancing is not there in P2P games, only on dedicated servers. P2P games do compensate for host advantage though.
  18. Tbh I think it happens when they have a lot of jitter, because then lag compensation simply can't keep up. But that might be a desync issue with what is drawn on the client screen versus the data the server has, because it came in late. If you watch the video from Battlenonsense, and do some calculations they have basicly brought down the interpolation delay between players to 1 update / frame. Otherwise you can't achieve a 80ms player to player delay with 2 people playing at 27ms ping. If one has jitter the server misses his updates and is guessing at his position. From my testing & reading, there is basicly 2 components to lag compensation. First the one that makes sure you can actually hit your opponent. Ie Player A moves but Player B shoots at where A used to be (because of the delay from A to server and from server to B ) Then the server rewinds time and validates the hit. The issue with this is (depends on who you are LOL) is that is still heavily favors the low latency player. Ie in the following scenario: Player A 100ms RTT Player B 10ms RTT Now when A comes out of cover he has a 55ms window where B cant see him yet and shoot at him. But his shots also take 50ms to reach the server. With a 200ms TTK weapon B could be dead at T = 255ms. However. Assuming B was preaimed, B can shoot back and kill A in 200ms, plus 5ms to the server, and A could be dead in T = 255ms. Pretty fair scenario. Now reverse the above. B comes out of cover, has a 55ms window to shoot at A, without A seeing him. But his bullets hit in 5ms. So A could be dead in T= 205ms. A doesn't see B yet, until T= 55ms, returns fire and has 50ms latency to the host. So B could be dead in T = 305ms. B should win in almost every scenario. But there is another mechanism in place, you can google this 'Artificial Latency Balancing' Basically it means that in the same scenario B simply gets a artificial latency penalty of 45ms applied to his shots, and probably also to his and the other players movements. That's why you get these wonky killcams, the server knew he was there but was delaying this info. There is probably some minimum and maximum value for compensation (like if there is a 200ms player in the lobby they won't count him unless it's a 1v1) This also explains what I am seeing when playing with my ICS setup. When I feel lag compensation is wonky I just dial up ping to 100 or 120 ms and then the game seems to play smoother. But I don't feel any further degradation in hit detection. If I leave that setting and get into a lobby where players have low latency I can sense they are ahead, turn my ping down to 40ms and all is fine again. I have to try more with assymetrical latency and jitter and see what that gives. And maybe measure the amount of frames between shots and hitmarkers. But I think that this latency balancing is the culprit.
  19. What Fraser said. But until it is fixed, the biggest giveaway is the host tickrate. Dedicated server is 60Hz and peer is 10-20Hz or so.
  20. Was in game ping, I can't tell router ping on the company network. It's a HP enterprise router.
  21. The game is using dedicated servers. No playerhosts.
  22. I played last week for a few days and it played really good for me. But that was on a enterprise system. (About 50 users, 150/150 1:1 lease line, wifi, VPN admin network, VPN client network, IPTV etc) 47ms ping not a smidge of lag. Hitdetection as you would expect from 47ms ping. Still was putting out 3KD games while laying down on the sofa, no headset, stock controller, 65" TV etc.
  23. I have that rigged up at the moment, ICS with my playstation connected for some tests, I am probably using the same software as you. The ICS makes my ping somewhat more unstable but that maybe has something to do with using a weak PC and a USB dongle to connect to LAN. What baffles me is that I hardly notice any degradation in hit detection going up to 100ms ping. So really what I am thinking now is that low ping players get a artificial latency penalty, similar to the anti host advantage. Now what I am curious about is how that is determined. Ie do they take the average ping in the lobby or is there a fixed value. I need to explore this more in Europe on a very low ping. Here in Asia half the lobby is usually red barring. I get way less instant death situations by adjusting my ping up to 100ms playing on the singapore server than playing on the Japanese server at 107ms.
  24. LOL if you're in Australia there is no way you would have 170 ping to The Netherlands. Most likely the Jap server.
  • Create New...