ASUS RT-N66U Firmware Version 3.0.0.4.374.2239

[bluepoint]

I have just been looking for improvements to the 5GHz band but maybe that's not going to happen. I also noticed Wi-Fi performance degrade when upgrading to iOS 7.0... I have a feeling Wi-Fi will improve with iOS v7.1.

What channels are you using? Try the lower 5GHz channels @40MHz width.

 

[got_milk]

The 5GHz performance on this build is unbelievable.

I tested a 945GB transfer from a wired gigabit desktop to my wireless MacBook Pro about 25 feet away, with a thin wall obstructing line of sight.

I had to redo the test a few times with a stopwatch and do the throughput calculations myself because I was becoming more and more sure that the OS X' Activity Monitor was giving me wrong information.

The 945MB transfer completed in 24.09 seconds, working out to an average throughput of 313.82Mbps or 39.228MB/s.

The highest I've ever seen on prior firmwares is about 28-29MB/s.

Edit: As evidence to prove I'm being serious:

 

[L&LD]

got_milk,

These are the kind of results I am getting too with v2239. Very, very impressed with Asus right now. Of course, this is closer to 'short/medium' distance performance results - but this is still an upgrade.

Did you notice that 2.4GHz was slower (or at least fluttered more) with this build? Or, like me you don't care about 2.4GHz anymore.

The MBP is a three antenna design, correct? (I can't get those speeds with my dual antenna AC7260 client).

 

[got_milk]

Did you notice that 2.4GHz was slower (or at least fluttered more) with this build? Or, like me you don't care about 2.4GHz anymore.

I didn't notice any slowdowns on 2.4GHz, I was averaging around 17-19MB/s (136-152Mbps) in the same conditions as my above post. There were very few instances where throughput suddenly dropped but as I'm surrounded by plenty of 2.4GHz networks it's not surprising to me and not something I'd directly attribute to the router itself.

The MBP is a three antenna design, correct? (I can't get those speeds with my dual antenna AC7260 client).

Correct - Apple has been using three antenna designs for the past few years.

 

[L&LD]

Thank you, then it must have been my 2.4GHz environment at that time or my client.

There really doesn't seem to be a downside to this new beta - hope all future firmware offer improvements like this.

 

[tpackage]

What channels are you using? Try the lower 5GHz channels @40MHz width.

I had the routers channel setting on "Auto" and it was on channel 157, channel bandwidth is set to 20/40MHz. I changed it to channel 36 and tested again. It made a big difference... I had a dead spot in one room before that now appears to be gone, getting max ISP speeds in that room now with 5GHz radio set to 80mW. That room is about 50 to 60 feet away from the router one floor below the router. Thanks bluepoint. I'll let you know if I see less dropped connections and more consistent speedtests.

On the a 2.4GHz radio, a lower channel (aka frequency) is better too from what I read, correct?

 

[bluepoint]

On the a 2.4GHz radio, a lower channel (aka frequency) is better to from what I read, correct?

You will need to use a tool to survey the wireless environment surrounding you like "inSSIDer" to have a good idea of the channels your neighbors are using. The lesser noise or overlaps from your neighbors is the ideal channel. First channels to try are 1,6,11 at 20MHz bandwidth since those channels are the least overlapped channels. However, if you have plenty of neighbors that are close to you and 1,6,11 does not give you the best throughtput, you might have to try each channels one at a time to choose the one that will give you the best throughput.

 

[tpackage]

You will need to use a tool to survey the wireless environment surrounding you like "inSSIDer" to have a good idea of the channels your neighbors are using. The lesser noise or overlaps from your neighbors is the ideal channel. First channels to try are 1,6,11 at 20MHz bandwidth since those channels are the least overlapped channels. However, if you have plenty of neighbors that are close to you, you might have to try each channels one at a time to choose the one that will give you the best throughput.

I use iStumbler to survey and I did that already... there are three SSID's on Channel 1, five SSID's on channel 6 and two on channel 11. I tried channel 2 and extended above... is that a good idea? Is it best to leave channel width at 20/40MHz or force 40MHz?

 

[bluepoint]

For 2.4GHz band and there are other wireless stations around you, it's best to use fixed channel @20MHz channel width. Try each channel(1 to 11) and stay with the channel that gives you the best throughput. Can you see from iStumbler how close your neighbors are(RSSI)? Stay away from a channel your neighbor is using with the strongest signal. Sometimes even if your neighbors are using the same channel as yours but they are far from you, it will be fine. Your throughput will determine the channel that is best for you.

 

[L&LD]

I learned something today with regards to 5GHz channels (lower=better).

In my experience though (with a few different routers, clients and wifi environments), the higher the 2.4GHz channel, the better the throughput - even if neighbors are on the same channel (11).

Time to test...

 

[L&LD]

Well, here are my results testing 5GHz channels with my RT-N66U and Asus beta v2239.

~35 feet away from router - two walls in between.

Note: although this is at the same distance as my previous tests around these forums, the orientation of the laptop is different as is the file used for the file copy testing.


Channel 161 Control Channel, 165 Bonded Channel

inssider shows 100% @ -50dBm - Base score

Ookla speed test 52.78/11.46 @ 16ms latency. - Base score

File download ~12MB/s avg. - Base score



Channel 149 Control Channel, 153 Bonded Channel

inssider shows 100% @ -50dBm - no change

Ookla speed test 52.93/11.50 @ 17ms latency. - significant? No.

File download ~10MB/s avg. - throughput - worse by 2MB/s



Channel 48 Control Channel, 52 Bonded Channel

inssider shows 100% @ -49dBm - significant? Maybe.

Ookla speed test 52.99/11.50 @ 16ms latency. - significant? No.

File download ~25MB/s avg. - throughput - MUCH better by 13MB/s



Channel 36 Control Channel, 40 Bonded Channel

inssider shows 100% @ -49dBm - significant? Maybe.

Ookla speed test 52.10/11.48 @ 16ms latency. - significant? No.

File download ~6.5MB/s avg. - throughput - MUCH worse by -5.5MB/s



Channel 40 Control Channel, 44 Bonded Channel

inssider shows 100% @ -50dBm - no change

Ookla speed test 52.92/11.54 @ 17ms latency. - significant? No.

File download ~23MB/s avg. - throughput - MUCH better by 11MB/s



Channel 44 Control Channel, 48 Bonded Channel

inssider shows 100% @ -48dBm - significant? Maybe.

Ookla speed test 52.91/11.46 @ 16ms latency. - significant? No.

File download ~29MB/s avg. - throughput - MUCH, MUCH better by 17MB/s




Well, a very productive night! Hanging out with my snb friends.

I more than doubled my actual throughput (what I use my wireless network for, besides internet access) while getting almost no indication of 'better' or 'worse' from simply running the Ookla speed tests OR using inssider from a stationary position/orientation with my laptop.


This is a good lesson for me to not be complacent once more. I thought I had learned this lesson; that 'the higher the channel, the better the WiFi' - but that only applies to 2.4GHz, I guess.

It was also interesting to see that the highest 'low channel' was also not the best either (as I would have predicted/betted).


Thanks again to Tim Higgins for this great forum.

Performance without spending more dollars is the best performance increase of all.

 

[RogerSC]

L&LD,

How many times did you run each of these tests, and how large were the test files? Wireless is pretty much up and down, lots of variability, so one test on each channel combination doesn't really prove anything, unless the files were GB's in size...not sure how many tests you need for each channel combination to get a statistically significant result, though, haven't looked at that. Did you use something like jperf so that you could actually look at the variations during the downloads via graphical results? That can also be significant...if you're streaming media, for example, you want as little variation as possible to avoid loss of streaming. You really don't know about that sort of thing unless you look.

Just curious how you approached this?

Thanks.

 

[L&LD]

The file transfer tests were run with a single file (my backup Outlook .pst file, specifically) of almost 3GB.

This was from a QNAP 4x 3TB RAID5 NAS to an AC7260 Windows 8.1 Pro client with a fast SSD. Both of these are very consistent in transfer speeds. I was also rebooting between each run to not be caching the file; need to be careful with 16GB RAM on board.

I used FreeFileSync to copy from the NAS to the laptop and I could see any variability as the copy was completing. The best runs had the most consistency.

The slowest time was around 7 minutes and the fastest was just over a minute (from memory - should have also written the times down in my post).

I did check/repeat the first couple of tests 3 times each; but they didn't change significantly - besides, the inssider graph shows no other 5GHz AP's in sight so I didn't continue this for the remaining runs.

Also, inssider was shut down while I was running the Ookla and file transfer tests - I have noticed that inssider otherwise affects the results (negatively).


Thanks for asking - I should have included this info in the original post.

 

[RogerSC]

The file transfer tests were run with a single file (my backup Outlook .pst file, specifically) of almost 3GB.

This was from a QNAP 4x 3TB RAID5 NAS to an AC7260 Windows 8.1 Pro client with a fast SSD. Both of these are very consistent in transfer speeds. I was also rebooting between each run to not be caching the file; need to be careful with 16GB RAM on board.

I used FreeFileSync to copy from the NAS to the laptop and I could see any variability as the copy was completing. The best runs had the most consistency.

The slowest time was around 7 minutes and the fastest was just over a minute (from memory - should have also written the times down in my post).

I did check/repeat the first couple of tests 3 times each; but they didn't change significantly - besides, the inssider graph shows no other 5GHz AP's in sight so I didn't continue this for the remaining runs.

Also, inssider was shut down while I was running the Ookla and file transfer tests - I have noticed that inssider otherwise affects the results (negatively).


Thanks for asking - I should have included this info in the original post.

Sounds good, but again your results are total transfer time, not looking at the fluctuations during the transfers...that would have been interesting to me. For me, the failures of wireless haven't been in average transfer rates, they've been in fluctuations that have affected the ability to stream media to my TV, and so on. But doing that is much more time consuming, I do get that.

 

[L&LD]

Maybe I wasn't clear enough; the worst scenarios had the most fluctuations - the transfer rate might vary as much as 10MB/s.

The best scenarios had very, very little variance. I would guess 2MB/s maximum and closer to 0.5MB/s avg. Taking into account that the difference between the worst to the best was over 22MB/s avg. speeds - you can see that the variance (absolute and percentage wise) is much less with the best channel used.

 

[bluepoint]

If I have time I'll test this beta or the next Asus release which probably is soon and let you know.

I tested .2239 beta very quickly and here's the comparison to EM.

a. EM's signal is better than .2239 by 5-8dBm on both bands
b. Range of EM is farther than .2239(tested from 200ft)
c. .2239's 5GHz band has showed good throughput as EM's.

Conclusion: Asus newer SDK6 improved but not as good as the EM build.

Beta .2239's driver's version is the same as Merlin's 6.30.163.2002(r382208) but different built date(1/14/2014) while Merlin's was built 12/12/2013.
BTW, "regulation mode" was removed in this build.

 

[RMerlin]

Beta .2239's driver's version is the same as Merlin's 6.30.163.2002(r382208) but different built date(1/14/2014) while Merlin's was built 12/12/2013.

The driver I use is a custom build they provided me with.

BTW, "regulation mode" was removed in this build.

Regulation mode is something I implemented, it's not present in the stock FW.

 

[RMerlin]

When testing throughput between two devices, I recommend using a tool like iperf or Lanbench. It will remove a bunch of variables from the equation that might be introduced when copying a file.

 

[tpackage]

I tested .2239 beta very quickly and here's the comparison to EM.

a. EM's signal is better than .2239 by 5-8dBm on both bands
b. Range of EM is farther than .2239(tested from 200ft)
c. .2239's 5GHz band has showed good throughput as EM's.

Conclusion: Asus newer SDK6 improved but not as good as the EM build.

Beta .2239's driver's version is the same as Merlin's 6.30.163.2002(r382208) but different built date(1/14/2014) while Merlin's was built 12/12/2013.
BTW, "regulation mode" was removed in this build.

bluepoint, I tested both .2239 and 3.0.0.4_374.38_2-em today and got results similar to yours. I uploaded the firmware, did a hard reset on the router and had all clients "forget" the SSID. I used "NetSpot" to measure signal strength.

I found the -em signal to be better than .2239. 40 feet away from and one floor below the N66U router. Here is what I got (worst signal location):

.2239 2.4GHz -57dBm
.2239 5.0GHz -75dBm

-em 2.4GHz -54dBm
-em 5.0GHz -66dBm

I only did Speedtest.net tests and all came back with Max ISP speed. I did not do a throughput WLAN speed test as I use Mac and can't use LANBench and I'm not sure if I can use iperf. Speeds on iOS went from 42Mbps using .2239 to max ISP speed using -em at this location as well. So far -em seems to perform as well as .2239 did in terms consistency but it will take some time for me to reply with those results.

 

[bluepoint]

So far -em seems to perform as well as .2239 did in terms consistency but it will take some time for me to reply with those results.

Good for you! Okay please keep us posted.