Call of Duty online gaming more responsive w/ QoS Off

[Statix]

The title is pretty much self-explanatory. I've been struggling with the latest Call of Duty game (Black Ops 2, PS3 version) for about a month now, and after trying a variety of different configuration tweaks, I found that disabling QoS has had the greatest effect on mitigating my "lag" issues when playing this game.

I have struggled with frustrating lag issues with the new Call of Duty game, Black Ops 2, for about a month, having a generally horrible experience playing the game and trying to be successful/competitive. I've struggled to simply maintain an average performance ("average" equating to approximately a 1.0 kill-to-death ratio--i.e., killing about as many people and the number of times I die in the game). After turning the QoS feature off entirely (both upload and download QoS), my kill-to-death ratio statistic has risen dramatically to about a 2-to-1 or greater kill-to-death ratio pretty consistently, over the course of the last 15-20 matches I have played thus far since disabling QoS; my overall empirical feeling is that the game feels significantly more responsive and smooth with QoS off.

This result is very much in-line with my previous experiences with turning QoS on and off on other routers and types of firmware I have owned (Linksys WRT54G, Buffalo WHR-HP-G300N, D-Link DIR-655, and lastly, Netgear WNDR3700v2). In all cases--and counter to my expectations--disabling QoS has usually benefited my gaming experience with Call of Duty games, enhancing either the responsiveness with which my bullets successfully register when shooting enemies, and/or the smoothness with which I see enemies move on-screen. My statistical performance has usually improved markedly with QoS off.

I hypothesize and attribute this overall improvement to the fact that QoS, even Gargoyle's implementation, actually increases the overhead, amount of processing, and overall feeling of latency (actual RTT ping times and ping tests remain mostly unchanged) of one's network packets, at least in terms of the Call of Duty games. Disabling QoS appears to make online gaming more responsive in Call of Duty, and thus enhance one's performance.

My performance and experience in another online game, Battlefield 3, seems largely unaffected by whether QoS is on or not. I do not know whether this is due to the fact that Battlefield 3 is less of a twitch-based, reactionary game than Call of Duty--thus less affected by slight differences in packet delay and processing--or whether the fact that it runs on a dedicated server network model (in contrast to the peer-hosted model of Call of Duty games) has any relation to my findings.

Fyi with regards to my network setup, I currently use a Netgear WNDR3700v2 router, Gargoyle 1.5.8, connected to the Motorola SB6120 Docsis 3.0 cable modem, and subscribe to Comcast cable internet, with download and upload bandwidth of approximately 16 megabits downstream and 2 megabits upstream.

[Statix]

I've been suffering from laggy gameplay in Battlefield 3 (PS3 version) for a few months now--the usual issues such as bullets taking very long to register, or not registering at all, opponents seemingly a second or two ahead of me at all times, etc. I've tried limiting my max download and upload to varying degrees, and everything from prioritizing my PS3's IP address to prioritizing the specific UDP game ports, to foregoing prioritization altogether.

After a few matches of frustration today, I decided to turn off QoS altogether again--both upload and download. Suddenly, my bullets are registering like they should be, seemingly almost instant bullet registration and kills. This went on for the next 3-4 games, with minimal lag and responsive gameplay. I no longer felt a second or two behind my enemies when fighting them--I actually felt like I had a fair chance to react and shoot them before they killed me.

Fyi, I am in California, and play on West Coast Battlefield 3 servers. I have Comcast Xfinity Internet (16mbps down/ 2mbps up tier).

I'd like to hear about other peoples' experiences with the QoS options of Gargoyle, and how their gaming is positively or negatively affected by it. In my experience, the QoS feature has had mostly had a negative effect on my gaming and lag. One positive, for what it's worth, is that enabling Upload QoS reduced my upstream bufferbloat dramatically, according to Netalyzer tests. Without outgoing QoS, my bufferbloat latency tested to be about 500-1000ms. It drops to ~60-70ms of bufferbloat when upload QoS is on. Download bufferbloat is seemingly unnaffected by QoS--it is ~60-70ms whether QoS is on or off.

[pbix]

I play Halo online a lot myself which was one of the main reasons I developed Gargoyle QoS to begin with. Once properly configured there will only be positive effects from enabling QoS for online gaming.

The key is to use the Minimum Bandwidth setting and set it large enough to cover the bandwidth needs of your game. Leave the maximum bandwidth setting as unlimited and the percentage setting really does not matter. Then check the minimum RTT box and enable the congestion controller.

If you are experiencing lag in your gaming then the only QoS related problems could be that you did not set your minimum bandwidth with sufficient bandwidth or that you do not have your gaming traffic routed into the correct QoS class.

Poor game play can be caused by devices outside your gargoyle router such as the other guys router, or anything in between. It can also be due to the fact that you just suck at the game

[Statix]

Hmm, so you say that the Minimum Bandwidth setting is more important than the Percentage BW setting? I will try cranking up Minimum Bandwidth and see if it helps anything.

A couple questions:

1) When configuring the classification rules, would it be best to set a priority for my PS3's device IP as a whole (e.g., 192.168.1.50, the PS3's static IP)? Or should I be more specific and also choose Transport Protocol: UDP and/or the specific game ports that the game uses (in the case of Battlefield 3, it would be UDP port 3659)?

In other words, how specific should I be? Does it really matter?

2) Why is Active Congestion Control only available for the download QoS and not the upload QoS?

[pbix]

I have my XBOX setup to classify UDP packets coming from the IP of my XBOX into my gaming class. This is because the important packets for online gaming in XBOX live are the UDP ones. I suspect most online games are the same.

The advantage of this is that lots of other things can happen on the XBOX like web browsing, downloading new maps and watching movies. This traffic is not time critical like gaming traffic is so it gets put in the normal class.

If you will read the QoS Wiki you will see that every class is allocated its minimum bandwidth first. Then what is left is split according to the other parameters. Minimum bandwidth should be used sparingly.

ACC works the way it does because that is what works best for most people.

[GargoyleNoob]

I have my up and down gaming classes set to 5% and 95% for normal. Max down for gaming class is 768kbps and max up is 512kbps. Neither have no limit. I still have well over 2000kbps available bandwidth when I check in Black Ops 2. Other QOS at least limited my bandwidth to what I set. Gargoyle's QOS doesnt seem to do that.

[pbix]

There are no known problems with max download settings in Gargoyle.

It appears your rules are not classifying traffic into the classes you think they are.

You should post a screen shot of the QoS download page which shows a class allowing more bandwidth than it is set for if you feel otherwise.

[Statix]

Thanks for the tips, pbix. It's just great and refreshing to see a router firmware developer who's as focused on the online gaming aspect as much as other tasks. Not enough firmwares out there put an emphasis on gaming latency and performance (which is probably the most latency-critical activity there is).

With that said, is QoS really going to benefit me if I am the only one using the internet in my apartment? My provisioned internet bandwidth never gets fully saturated; my roommate only does web browsing and never downloads or does anything bandwidth-intensive while I'm gaming. Wouldn't QoS simply add more unnecessary processing and delay to my network?

I have my up and down gaming classes set to 5% and 95% for normal. Max down for gaming class is 768kbps and max up is 512kbps. Neither have no limit. I still have well over 2000kbps available bandwidth when I check in Black Ops 2. Other QOS at least limited my bandwidth to what I set. Gargoyle's QOS doesnt seem to do that.

Fyi, Black Ops 2's bandwidth display shows the available upload bandwidth, not download bandwidth, of your network.

I've personally never had any problem with Gargoyle failing to cap my upload/download speeds. Whatever I set my upload bandwidth maximum to, it displays properly in Black Ops 2 and in speed tests.

[pbix]

It is true that if your line is never saturated then QoS will not have any benefit. I just find it hard to believe that your line is never saturated. Even watching one youtube video will saturate most lines but if you are sure then your right.

But it is also true that QoS does not add any delay to your packets and you can use it without fear of this. Packets move through your router very fast indeed, that is as long as they do not need to wait because the WAN link is full of other traffic.