Bert

You wouldn’t be blaming hitdetection for your loss if you were that good at the game
 
These traffic prio rules only function under the condition that your line is congested. Then it works because you get a buildup of packets in the sending que and it can put game traffic in front of the que. If no congestion occurs then each packet is send inmediately and the rule does nothing. Running QoS requires packet inspection and that does add latency, this is the trade-off for running QoS on your system. So therefore in situations with symmetrical connections with very few devices I would always recommend turning off QoS totally as this gives you the cleanest path for traffic. Ie I have 600/40 connection in one house where that connection services the whole home, there I have QoS on because it’s easy to congest a 40mbit upload line. In my other house I have 2x 600/600 WAN and the connection that is used for gaming doesn’t use QoS at all since I never congest it with just 1 PS4 and 1PC.
 
As for DumaOS classified games vs manual rules. You could argue that other UDP traffic will try to get in front of your gaming traffic (DumaOS classified games simply prioritizes UDP 1024-65535) but in reality this is not noticeable unless you are running something special that generates a huge UDP stream. And here again, it would require the line to be congested in the first place. Also you could add the device as game console and then it simply prioritizes UDP traffic to that device.
 
The only reason I use a manual rule is so that my bandwidth is not restricted when the game is running idle in the menu.

There is really no magic traffic prioritization rule. You will find many topics about it, people have spend countless hours with wireshark etc but none have the magic answer.
 

You need a rig that can produce 540 FPS to be able to take advantage of that, even with a 4090 or 7900XTX you will not see those framerates.
 
Also the difference is getting smaller every time you up the Hz rating:
60 Hz = 16,6ms between frames
120 Hz = 8,3ms between frames
240 Hz = 4,2ms between frames
360 Hz = 2,7ms between frames
540 Hz = 1,9ms between frames
 
For the amount of manipulation happening from the game, ie active latency balancing and skill based matchmaking I would not invest in anything over average hardware as it's simply pointless. Even stuff like routers. Average home net has 1ms between device and router and for a fiber connection you have 1ms between device and ISP node. Do you really think QoS and all this BS matters if the game simply matches your latency in lobby to the other players? You can't even tell if you are playing on a 10ms, 30 ms or 50ms server in game due to the lag compensation systems. 

You need a rig that can produce 540 FPS to be able to take advantage of that, even with a 4090 or 7900XTX you will not see those framerates.
 
Also the difference is getting smaller every time you up the Hz rating:
60 Hz = 16,6ms between frames
120 Hz = 8,3ms between frames
240 Hz = 4,2ms between frames
360 Hz = 2,7ms between frames
540 Hz = 1,9ms between frames
 
For the amount of manipulation happening from the game, ie active latency balancing and skill based matchmaking I would not invest in anything over average hardware as it's simply pointless. Even stuff like routers. Average home net has 1ms between device and router and for a fiber connection you have 1ms between device and ISP node. Do you really think QoS and all this BS matters if the game simply matches your latency in lobby to the other players? You can't even tell if you are playing on a 10ms, 30 ms or 50ms server in game due to the lag compensation systems. 

You need a rig that can produce 540 FPS to be able to take advantage of that, even with a 4090 or 7900XTX you will not see those framerates.
 
Also the difference is getting smaller every time you up the Hz rating:
60 Hz = 16,6ms between frames
120 Hz = 8,3ms between frames
240 Hz = 4,2ms between frames
360 Hz = 2,7ms between frames
540 Hz = 1,9ms between frames
 
For the amount of manipulation happening from the game, ie active latency balancing and skill based matchmaking I would not invest in anything over average hardware as it's simply pointless. Even stuff like routers. Average home net has 1ms between device and router and for a fiber connection you have 1ms between device and ISP node. Do you really think QoS and all this BS matters if the game simply matches your latency in lobby to the other players? You can't even tell if you are playing on a 10ms, 30 ms or 50ms server in game due to the lag compensation systems. 

You need a rig that can produce 540 FPS to be able to take advantage of that, even with a 4090 or 7900XTX you will not see those framerates.
 
Also the difference is getting smaller every time you up the Hz rating:
60 Hz = 16,6ms between frames
120 Hz = 8,3ms between frames
240 Hz = 4,2ms between frames
360 Hz = 2,7ms between frames
540 Hz = 1,9ms between frames
 
For the amount of manipulation happening from the game, ie active latency balancing and skill based matchmaking I would not invest in anything over average hardware as it's simply pointless. Even stuff like routers. Average home net has 1ms between device and router and for a fiber connection you have 1ms between device and ISP node. Do you really think QoS and all this BS matters if the game simply matches your latency in lobby to the other players? You can't even tell if you are playing on a 10ms, 30 ms or 50ms server in game due to the lag compensation systems. 

I tried updating from .134 and that seems to do the trick.
 
I was at a DumaOS 3 beta previously. going intermediate firmware -> RC10 -> RC14 gave issues. Maybe updating needs to be done from .134 and not from a previous beta.

At 1Gbit up and down you should switch off QoS entirely unless you use a huge amount of bandwidth.
 
QoS is the thing that is probably least understood by gamers but it does for sure sell routers. If you are not reaching the maximum of your connection then there is no packets stuck in the queue and QoS will not do anything for you.
 
Also say you have 500mbit. You go testing bufferbloat, adjust it to 400mbit. Great but if you are using the net by yourself and your average traffic use while gaming is 1mbit instead of downloading stuff, here QoS will again do nothing for you.
 
QoS actually slows down your traffic in terms of latency. Generally this little bit of latency is accepted by getting constant performance in return. A household with kids going off a 100/10 connection or so needs QoS but not at 1000/1000.
 
And yes for 1Gbit SQM you need a lot of CPU power. As far as I know only x86-64 routers are capable of doing this at this present time.

Depends on really.
 
The 8Gbit we are supposed to get comes with a router that has 2.5Gbit ports so you can't use 8Gbit for a single device. That's why it's offered cheap as they think that nobody is able to use 8Gbit anyway. (I think XGS-PON can supply 10Gbit to a node so you would be almost saturating the entire node) But offcourse we won't be happy with that  
 
10Gbit infrastructure is not that cheap yet though. If you buy the equipment new. You need a 10Gbit card in every PC or use mainboards that have 10Gbit included ($$$$). 10Gbit capable switches are also on the pricy side. I went with a switch that supports both 10Gbit and 2.5 Gbit as I think 2.5Gbit will gain more traction for home use, you can see PC mainboards utilizing this a lot more over 10Gbit and things like NAS devices as well. It's only a matter of time before TV's, consoles and streaming devices will switch to 2.5Gbit. Reason is that it's a lot more power efficient over 10Gbit. 10Gbit switches need active cooling so are usually on the noisy / power hungry side.
 
Then depending on the usecase, both your PC and your other sources need to be able to go up to 10Gbit speeds. Most NAS devices for home will not make this, hence why 2.5Gbit is rising in popularity. My servers are Intel Xeon with PCIE 3.0 NVME drives in RAID so they have no issue keeping up. And workstation is AMD 5950X with PCIE 4.0 drive, Sabrent Rocket. For a single file, say copy a 20GB DVD file, takes 20 seconds to copy, 100 GB takes under 2 minute, can't beat that with walking around with a external SSD.
 
Also for latency it makes a difference. At 1Gbit, local latency is about 1ms on LAN. At 10Gbit, latency is 0.25ms. While it does not sound like much, working on network shares feels like the drives are physically in your PC, they are much more responsive then on 1Gbit. The real reason for wanting the 8Gbit WAN over 1Gbit WAN though is that I have a second place in another country that already has 1Gbit FTTH (and they are slowly scaling up to 2Gbit), and use a site to site wireguard VPN to access the same server, where as the internet connection at my place where the server is has a low 40mbit upload now that is holding back transfers. For connections over distance it's actually better to have some overhead as latency quickly kills speed.
 
Also what helps loads for big file transfers is loads of RAM. When you download something, the NIC will write to memory and then to the drive. Windows utilizes 10% memory for this. So for a PC with 32 GB memory, it's full at 3GB and then you are most likely to see drops in transfer speed as memory needs to be flushed to the drives. Windows Server uses 50% so a server with 64 or 128 GB memory that can be used for 50% is more likely giving better results. Also for slower PC's RSS (recieve side scaling) needs to be enabled so incoming transfers will be distributed over multiple cores rather than 1 core. And something that is an issue with current retail PC's is the number of PCI-E lanes. Most have 20 so you use 4x for the SSD boot drive and 16x for the Graphics card. So that means that the 10Gbit NIC is generally located on the chipset / south bridge hogging up 4x PCI-E lanes. In reality for fastest transfers you would want both the NIC connected to the CPU and also a 2nd SSD drive connected to the CPU but that brings you into the territory of workstation PC's, Intel Xeon and AMD Threadripper.
 
Going above 1Gbit bandwidth for sustained transfers actually poses a lot of bottlenecks.
 
 

144Hz usually.

Scoreboard is roughly equal to netduma ping. 

What he was referring to is artificial latency balancing.
 
Paper from 2005:
https://www.researchgate.net/publication/220982447_Achieving_fairness_in_multiplayer_network_games_through_automated_latency_balancing
 
Basically they attempt to balance out latency server side by adding additional latency to players with fast internet. And that paper describes tech from 2005.
 
This is also the reason why P2P games always felt more responsive, they eliminated host advantage but didnt balance the clients.
 
So indeed it makes little difference playing on 20ms or 50ms ping etc. 
 

It supprised me actually the R6700XT. Compared to my other system, 5950X + 3070Ti it's getting a bit more FPS but not amazing amounts. I got a 40% off deal on the RX6700XT so for the price can't beat that.

Basically a copy of MW2019 with a new skin. In all aspects, shite maps, SBMM, camping & slow gameplay.
 
If you didn't like MW2019 you won't like this either.
 
99% uses M4 because it's way stronger than anything else and the rest of the weapons is dull and of limited use. No default ghost and dead silence is a field upgrade, plus footsteps are loud AF.

Actually I am playing on my parts bin PC lol (Xeon E5 v4 E2697A4 + RX6700XT)
 
Old server CPU but still getting 120-150 FPS in 1080P.
 
When I switched to PC I played 1 game of Ground war, joined half way, #1 on the leaderboard with 30 kills and straight back to SBMM land lol. 

LOL I have been raging at the clunky aiming only to find out now there is actually a advanced menu.

Something that I have not seen mentioned is that they have removed the aiming curves, linear, dynamic etc. I found aiming a bit more difficult because of that, since I have been playing on the Dynamic setting for ages.

If the connection is spiking at idle no QoS will fix that, you will have to take it up with your ISP.

QoS adds another layer of processing. And really you are trading ultimate low latency for a more stable latency.
 
Also CoD games, at least in 6v6 use something that is called artificial latency balancing. So if your ping is 20 but the lobby average is 100 it gives you a artificial penalty. So really would only advise using QoS if you have real issues like these 1000ms ping spikes.
 
Also noteworthy is that QoS is only useful if you download at the same time. Ie when you are playing a game and not using the internet much QoS will do almost nothing for you.
 
I have it off most of the time.

Nothing, same minimap as in MW2019, so no dots when shooting. UAV is hyper OP.

Do you have IPv6 enabled? A lot of these sites use IPv6 and I found on XR500 that speed is not restricted on IPv6. Might be the same issue with XR1000.

The reason for this is that the router does not know if you use a repeater or LAN connection. It detects the interface of the repeater that is why it's shown as LAN.
 
You need a mesh system or a system with controller for that to work properly. Or possibly a netgear repeater with the smart connect option.
 
The router only knows if it's wifi when the device connects through their own interface.

What kind of SSD is it? I did my testing with a Samsung 860 QVO.
 
The difference in rates might be caused by buffering. The router does not have any sort of data caching. So when you copy to the disk, the router will attempt to write straight to the disk and if this is interrupted, ie by inefficiencies in the USB protocol the transferspeed will stall.
 
If the other way around, you download to your PC. What will happen is that the router will take the data off the disk and your PC writes the entire transfer in memory. It will show as complete but windows will write the data to the disk in the background. This has nearly no restrictions so no drops in data transfer other than caused by the router.
 
For windows 10 systems drive cache is 10% of available memory and for windows server systems it's 50% of available memory.
 
Also if you are attempting these copy tests you need to alternate your testing file as often the file you just copied remains in windows/drive hotcache so when you transfer it again, it's much faster.
 
When I am operating a lot of VM's on my SMB server I can see writing stalling out when the buffers in memory are full (VM's take system memory so less total memory), windows needs to flush the buffers first. This is with big files though like 30GB and transferring via 10Gbit. But it's writing to a NVME SSD RAID cluster so very little restriction there.

I think you are referring to having a disk connected to the router's USB?
 
For me the XR500 seems limited to 80 MB/s in router mode and no restriction (120mb/s) in AP mode. XR700 does 200 MB/s if fed off a SSD.
 
For device to device transfers, the router's CPU has nothing to do with it as there is hardly any routing involved.
 
On SMB networking, just like any other TCP/IP connection it's very sensitive to starts and stops in transmissions. Ie dropped packets it will slow the transmission down. So over wifi this is probably exaggerated. 
 
I would try and use a different tool first, ie Iperf3 and test speed on the actual connection between the laptop and the SMB server and take it from there.

The PON-XGS has been out for some time in France, we have 2 ISPs out of the 4 national ones which offer 8 Gbit in DL and the 2 others which currently only offer 2 Gbit (theoretical speed)..., I would be very interested to see a router signed with DumaOS allowing us to take advantage of it...!

In order to do what Fraser says you need to have a managed switch that allows for port replication. For example Netgear GS305E, typically 30 bucks on Amazon. Hub will also work but not even sure if you can still buy those, if so they will usually only be 10/100 mbit. And you need a second device, ie PC or Laptop running the wireshark software and connect to that port.
 
Alternative is going to the router's debug page (for example 192.168.1.1/debug.htm change IP to yours) and there is an option to replicate the router's WAN port to LAN1 to do a wireshark capture. Same as with a switch you need to connect a PC/Laptop to that and use wireshark.