merlin-firmware performance drop from 378.54_2 onward?

[Rippo]

I have a T-Mobile Cellspot (Asus 68u) and have been happily running asus-merlin for over a year. However, every time I upgrade from 378.54_2 (to any of the later asus-merlin revisions) I see a 25% drop in download speeds on the 2.4 network. I am doing factory resets after every firmware upgrade/downgrade and my results have been repeatable about a dozen times now. Did any default settings change after 378.54_2 that I could be missing? I am stumped...

 

[Etz]

How do you measure performance and have determined that it is actually 25% worse?

 

[Rippo]

The first time I upgraded firmware from 378.54_2 to 378.55_0 I noticed that streaming video was choppy/buffering, downloads were taking more time than before the upgrade in revision, and my VPN client was dropping connection every ten minutes or so. This prompted me to try speedtest.net to see if there was a more measurable difference- using 378.54_2 I get 54 Mbps download (on the 2.4 wifi network) vs. 40 Mbps when using any later merlin firmware release. I have upgraded/downgraded about a dozen times now and the speedtest results are consistent as well as the buffering, download speed, and VPN client issues. Note that with 378.54_2 I never buffer, downloads are swift, and I never drop VPN client...

Any ideas?

(Edit: I should add that my Windows clients are the most obviously affected (Windows 8). I have not tested using IOS clients.)

 

[Etz]

Actually Speedtest is not quite adequate way to test throughput, too many variables.

iperf would be far more accurate, also Wireless tests should be run inside Faraday cage, to rule out any interference issues which can also cause different results.

So far I have discovered only speed and stability improvements, with newer firmwares on my RT-AC68U.

 

[joegreat]

I am doing factory resets after every firmware upgrade/downgrade and my results have been repeatable about a dozen times now.

Are you also doing manual reconfiguration (not loading old settings)?

 

[Rippo]

Actually Speedtest is not quite adequate way to test throughput, too many variables.

iperf would be far more accurate, also Wireless tests should be run inside Faraday cage, to rule out any interference issues which can also cause different results.

So far I have discovered only speed and stability improvements, with newer firmwares on my RT-AC68U.

I will look into iperf and compare results. The dozen times I have upgraded/downgraded, I have done so within a forty minute period and without exception I get the buffering, slower downloads and dropped VPN connection only with the later revisions of firmware. I am not suggesting that I think the firmware has a problem- more likely that a default changed or something that I am not understanding in configuring the later revisions... I just upgraded /downgraded again with the same results

 

[Rippo]

Are you also doing manual reconfiguration (not loading old settings)?

Yes, doing a manual reconfiguration each time. I had tried using the ad-blocking functionality in the revision that offered it (later removed) but followed the steps Merlin outlined in the readme for those who had enabled it.
There are three Windows 8 machines I use on the 2.4 wifi network each with a different network adapter- all experience the same issues when I upgrade firmware... I am considering tomato to see if I find any difference. I have no experience with that whatsoever. I prefer merlin's approach/philosophy however.