Waveform or CB for bufferbloat?

[druski]

I've been playing with the CC dials, and have finally managed to get my grade to an A-A+ on waveform. However, when I test in the CB for xr1000, my underload latencies are about 20-30ms on average, peaking into the 35-60ms on occasion, with a 17ms idle. Waveform put me at around +4 to +7ms for both Up and Down. I have been able to tune the CC to have my underload on CB within +1-5ms with 17ms idle, but when on waveform, I lose the low ms spiking to about +10-20ms. I wanted to know which test a more reliable/accurate representation of what my underload latency actually is. I followed a thread in one of the R3 discussions when it came to bettering hit registration  in COD, and everyone in there was using waveform as their representation for "good reg." Just wanted to know if there really any difference, or if one really does matter when trying to accommodate to my situation (that being bettering reg).

 

[druski]

Just now, druski said:

I've been playing with the CC dials, and have finally managed to get my grade to an A-A+ on waveform. However, when I test in the CB for xr1000, my underload latencies are about 20-30ms on average, peaking into the 35-60ms on occasion, with a 17ms idle. Waveform put me at around +3 to +7ms for both Up and Down. I have been able to tune the CC to have my underload on CB within +1-5ms with 17ms idle, but when on waveform, I lose the low ms spiking to about +10-20ms. I wanted to know which test a more reliable/accurate representation of what my underload latency actually is. I followed a thread in one of the R3 discussions when it came to bettering hit registration  in COD, and everyone in there was using waveform as their representation for "good reg." Just wanted to know if there really any difference, or if one really does matter when trying to accommodate to my situation (that being bettering reg).

I dont think the original CB posted. Here it is.

[Netduma Fraser]

They're both using different servers so you'll get different results and while they're both providing an approximation, the best thing to do in my opinion would be testing with downloads while following: https://bit.ly/bufferbloat-test start with a high value e.g 95%, test, decrease 10%, test, decrease 10% etc, until you get to a value that is pretty good, then try 5% either side of that value to see if it can be improved. Important to note that Download/Upload on CC don't have to be the same value & you may have a better experience with differing values. This is more of a real world test as you're actually doing traffic you would normally do.

[druski]

1 hour ago, Netduma Fraser said:

They're both using different servers so you'll get different results and while they're both providing an approximation, the best thing to do in my opinion would be testing with downloads while following: https://bit.ly/bufferbloat-test start with a high value e.g 95%, test, decrease 10%, test, decrease 10% etc, until you get to a value that is pretty good, then try 5% either side of that value to see if it can be improved. Important to note that Download/Upload on CC don't have to be the same value & you may have a better experience with differing values. This is more of a real world test as you're actually doing traffic you would normally do.

I've used this and my ping is pretty stable. The one thing I've noticed when using waveform, which you can see in the screenshot, are those super high max latencies. I've yet to see those on the ping-plotter, and I could only imagine would be responsible for the high jitter/+ms when testing. Are those "anomalies", or is this something that can be targeted? They don't happen every single time, but the do appear often when testing.

[Netduma Fraser]

So bufferbloat is caused by the saturation of your network, with Congestion Control lowered that cannot happen any more, thus bufferbloat is eliminated. So as long as you've found the best results using PingPlotter I would honestly not worry about the waveform test as it's unlikely your ping will ever spike that high now. Also as long as you're not getting it that high when gaming you're all good, you'd just essentially be chasing an arbitrary number.