On xr500 open beta - traffic Connection Benchmark. Ds for speed and Ping Test.

[Soupdragon]

Hi guys.

Always get the feel I've got sudden bursts of lag or data loss where I press buttons in Modern Warfare multi/Warzone and they don't register. Had things like the C4 throw animation starting but not throwing after a button press, and stun grenades landing directly on people in the killcam but not doing anything, shotgun shots at point blank not even hitmarking despite multiple perfect shots at 0 range with them standing still.

Ran the new connection test on the 3.0 beta for XR500 and found I had great results on the Ping Test (A+), but some spikes... suspect they're nothing serious though.

Ping Test and Speed Test were both at D though. Are there any obvious causes for the poorer results in those (and hopefully, any easy fixes to deal with them?)

Thanks in advance for checking this out.

Cheers,

Rob. 

[Netduma Fraser]

What you need to do is use Congestion Control and find the percentages that best lower the ping under load results so would suggest starting at 90% on Always, doing a test, reducing by 10% etc rinse repeat until you find a percentage that gets it quite low, then try 5% either side of that value. Also please post on the Netgear beta forum for beta issues/queries.

[Soupdragon]

Thanks Fraser. Will try that out and report back! 

[Soupdragon]

1 hour ago, Netduma Fraser said:

What you need to do is use Congestion Control and find the percentages that best lower the ping under load results so would suggest starting at 90% on Always, doing a test, reducing by 10% etc rinse repeat until you find a percentage that gets it quite low, then try 5% either side of that value. Also please post on the Netgear beta forum for beta issues/queries.

Just to check a couple of things:

- When I'm running these tests, should I make sure nothing heavy is running in the background? (ie. wife is watching Netflix right now - is it better to run the tests when nothing's on?)

- Do I try reducing the Download and Upload proportionately? (ie. 90% on DL and 90% on UL first, then 80/80 etc.)

Anything else I need to keep in mind or tweak when running these tests? (and is there a lower limit you would recommend I avoid going below... like 30 or 20%?

[Netduma Fraser]

To answer your questions:

• The test does show what your Ping would be like when under load so I would suggest trying as much as you can to prevent anything else going on just to ensure a fair test.
• Yes do it proportionally at first, then if you notice upload ping go way down for example then try to fine tune that then go back to download.
• Not that I can think of no, for lower limit it is really dependent on what your speeds are and how much you need for general everyday use. If you have 1Gbps for example then even 10% might be plenty for you but if you had 100Mbps then 10% might not be enough.

[Soupdragon]

15 hours ago, Netduma Fraser said:

To answer your questions:

• The test does show what your Ping would be like when under load so I would suggest trying as much as you can to prevent anything else going on just to ensure a fair test.
• Yes do it proportionally at first, then if you notice upload ping go way down for example then try to fine tune that then go back to download.
• Not that I can think of no, for lower limit it is really dependent on what your speeds are and how much you need for general everyday use. If you have 1Gbps for example then even 10% might be plenty for you but if you had 100Mbps then 10% might not be enough.

Thanks Fraser. I tried this and I found that ping during upload only started to decrease around the 20% mark, and it only decreased from around 400ms to 200ms at that stage. Once I had it down at 5% (which resulted in around a 0.4mbps available for upload), it reduced to around 100ms.

Still only achieving a D grade for ping under load with that - and it seems very low to set it to in order to only achieve 100ms during upload, considering the ping during download etc. wasn't too bad, and I think that actually started to reach around 7-10ms for idle and download as I decreased both sliders too, so this upload figure stands out. 

Was just about to run a test with all QoS turned off to see what ping during upload is like without it, but the tool for these tests has stopped working on my router now - understandable as it's a beta. Is there an external tool I can use to keep testing/running these checks with the different QoS settings?)

[Netduma Alex]

21 hours ago, Soupdragon said:

Is there an external tool I can use to keep testing/running these checks with the different QoS settings?)

Well, Connection Benchmark uses M-Labs servers to perform the tests, so perhaps M-Labs tests available online might give you similar results?

The only issue is that Connection Benchmark runs on the router rather than your PC, and you can't really emulate that.

[Soupdragon]

On 10/1/2020 at 10:44 AM, Netduma Alex said:

Well, Connection Benchmark uses M-Labs servers to perform the tests, so perhaps M-Labs tests available online might give you similar results?

The only issue is that Connection Benchmark runs on the router rather than your PC, and you can't really emulate that.

Cheers Alex.

Generally, if someone is experiencing a huge jump in ping in the upload test, and lowering the QoS % for upload doesn't really affect the high ping until it gets to unreasonably low percentages - what would you recommend testing/changing? Is there a common cause for this?

[Netduma Fraser]

32 minutes ago, Soupdragon said:

Cheers Alex.

Generally, if someone is experiencing a huge jump in ping in the upload test, and lowering the QoS % for upload doesn't really affect the high ping until it gets to unreasonably low percentages - what would you recommend testing/changing? Is there a common cause for this?

I'd suggest setting Congestion Control to auto enable so it only throttles when you're actually playing a game. Otherwise try to find a higher percentage that maybe not as good but still better than the other options.