Is the RT-AC68U a good replacement for a RT-N66U?

[oguime]

My RT-N66U is a few years old and it's 5GHz band has never been strong enough to reach the whole house. My wife and daughter's iPhones sometimes loose connection even on 2.4GHz.

Would the RT-AC68U improve signal strenght and connection quality?

Thanks,
Gilson

 

[pete y testing]

Would the RT-AC68U improve signal strenght and connection quality?

the answer would prob be not a great deal as there isnt a great deal of difference in transmission levels / coverage between the two nor will there be getting any bigger or faster claiming routers that are now coming onto the market as we have reached max eirp ( transmission power levels ) according to the fcc and its rules

caveat : there will be those that will tell you adding higher gain antennas will work or pushing the router past its max transmission levels and yes you can do it but it breaks the rules and puts your health at risk as the max eirp rules are there for a reason

so changing up to an asus rt-ac68u will gain you a little but not a great deal as it sounds like your environment and construction materials are the issue and a second access point located in another part of the house connected back to the first via ethernet is the best course of action and if you cant run ethernet for some reason then look at power line / EOP adapters as they even have ones with wireless at the secondary end

in my case i run my main router down one end of the house where the fiber comes into the house in the study , its an asus rt-ac87u and covers the study / family room / dinning room / living room quite well but the bedrooms down the other end of my long narrow house suffer so what i have done is have ethernet run to the linen cupboard in the hallway near the bedrooms and run an asus rt-ac68u in AP mode in that location and it covers all 4 bedrooms very well with max signal to all , as we move from one area to the other the client devices roam quite well after adjusting the transmission power of each router so there was not a great overlap , i use the same ssid and passphrase and just different channels for each band and router

so my suggestion is keep the rt-n66u and use it fir the access point in another part of the house and yes get the rt-ac68u for your main router

sure try the rt-ac68u on its own first and see how it covers but i think you will end up needing the n55u as an access point anyway

pete

 

[Nullity]

I would not buy it for signal strength, though AC should offer improvements with even N clients. The RT-N66U is well respected for it's signal reach.


I would buy the AC68U for the improved CPU. My RT-N66U maxes out at ~300Mbit wired (intra-LAN), which sucks. The RT-N66U has a 600Mhz (overclockable to 662Mhz) MIPS CPU, while the RT-AC68U has dual 800Mhz (overclockable to ~1.4Ghz or more?) ARM CPU. The AC68U also comes with a newer Linux kernel, I think.

 

[ColinTaylor]

I would buy the AC68U for the improved CPU. My RT-N66U maxes out at ~300Mbit wired (intra-LAN), which sucks.

That's weird. My N66U (not overclocked) does inter-LAN at near gigabit speed as expected.

C:\Utils\iperf>iperf -c micro -w 128k -i 5 -t 60
------------------------------------------------------------
Client connecting to micro, TCP port 5001
TCP window size:  128 KByte
------------------------------------------------------------
[  3] local 192.168.1.55 port 50826 connected with 192.168.1.50 port 5001
[ ID] Interval       Transfer     Bandwidth
[  3]  0.0- 5.0 sec   566 MBytes   950 Mbits/sec
[  3]  5.0-10.0 sec   567 MBytes   951 Mbits/sec
[  3] 10.0-15.0 sec   565 MBytes   947 Mbits/sec
[  3] 15.0-20.0 sec   567 MBytes   951 Mbits/sec
[  3] 20.0-25.0 sec   565 MBytes   947 Mbits/sec
[  3] 25.0-30.0 sec   566 MBytes   950 Mbits/sec
[  3] 30.0-35.0 sec   565 MBytes   948 Mbits/sec
[  3] 35.0-40.0 sec   566 MBytes   949 Mbits/sec
[  3] 40.0-45.0 sec   564 MBytes   947 Mbits/sec
[  3] 45.0-50.0 sec   566 MBytes   950 Mbits/sec
[  3] 50.0-55.0 sec   566 MBytes   950 Mbits/sec
[  3] 55.0-60.0 sec   566 MBytes   950 Mbits/sec
[  3]  0.0-60.0 sec  6.63 GBytes   949 Mbits/sec

Windows file transfers are ~840Mbps (and that's probably limited by how fast I can write to the disk)

 

[Nullity]

That's weird. My N66U (not overclocked) does inter-LAN at near gigabit speed as expected.

C:\Utils\iperf>iperf -c micro -w 128k -i 5 -t 60
------------------------------------------------------------
Client connecting to micro, TCP port 5001
TCP window size:  128 KByte
------------------------------------------------------------
[  3] local 192.168.1.55 port 50826 connected with 192.168.1.50 port 5001
[ ID] Interval       Transfer     Bandwidth
[  3]  0.0- 5.0 sec   566 MBytes   950 Mbits/sec
[  3]  5.0-10.0 sec   567 MBytes   951 Mbits/sec
[  3] 10.0-15.0 sec   565 MBytes   947 Mbits/sec
[  3] 15.0-20.0 sec   567 MBytes   951 Mbits/sec
[  3] 20.0-25.0 sec   565 MBytes   947 Mbits/sec
[  3] 25.0-30.0 sec   566 MBytes   950 Mbits/sec
[  3] 30.0-35.0 sec   565 MBytes   948 Mbits/sec
[  3] 35.0-40.0 sec   566 MBytes   949 Mbits/sec
[  3] 40.0-45.0 sec   564 MBytes   947 Mbits/sec
[  3] 45.0-50.0 sec   566 MBytes   950 Mbits/sec
[  3] 50.0-55.0 sec   566 MBytes   950 Mbits/sec
[  3] 55.0-60.0 sec   566 MBytes   950 Mbits/sec
[  3]  0.0-60.0 sec  6.63 GBytes   949 Mbits/sec

Windows file transfers are ~840Mbps (and that's probably limited by how fast I can write to the disk)

I always thought it was weird as well... I never checked into it further because I only have 2 (Intel) GbE NICs, and both are installed in my gateway pfSense router (for driver purposes).

Maybe it is CTF/HW-accel related, since I always disable it? IIRC, when it maxed at 300Mbit, the softirq was at ~100%. Maybe I had it misconfigured...

 

[ColinTaylor]

Maybe it is CTF/HW-accel related, since I always disable it? IIRC, when it maxed at 300Mbit, the softirq was at ~100%. Maybe I had it misconfigured...

I've seen that sort of problem with softirq's as well but only on WAN to LAN, not LAN to LAN.

 

[Nullity]

I've seen that sort of problem with softirq's as well but only on WAN to LAN, not LAN to LAN.

It's not a problem by itself, since the same condition arises when you max it out the device with WiFi (LAN/WLAN to WLAN) traffic. Though, I would expect wired-only LAN-to-LAN to have much less soft interrupts since the dumb switch should be the main mover of traffic. Dunno... I wish I could do more testing, but I don't have the hardware.

I get ~180Mbit with wireless 2x2 clients, which is my main concern.

 

[pete y testing]

I get ~180Mbit with wireless 2x2 clients, which is my main concern.

just for reference i have just tweaked a few things in my lab setup and now have a ssd in both my synology ds 514+ and my test pc with an asus pce-ac68 installed

am getting 110MB/s throughput read and write over ethernet and 100MB/s read and 75MB/s write with the asus pce-ac68

I get ~180Mbit with wireless 2x2 clients, which is my main concern.

how are you testing this ? as that equates to just over 20MB/s or so and is extremely slow for 2 x 2 AC , at what distance are you measuring this at as well and what sync is the adapter showing during the test

 

[Nullity]

just for reference i have just tweaked a few things in my lab setup and now have a ssd in both my synology ds 514+ and my test pc with an asus pce-ac68 installed

am getting 110MB/s throughput read and write over ethernet and 100MB/s read and 75MB/s write with the asus pce-ac68


how are you testing this ? as that equates to just over 20MB/s or so and is extremely slow for 2 x 2 AC , at what distance are you measuring this at as well and what sync is the adapter showing during the test

I have the RT-N66U as an AP, meaning 3x3 can reach a link-rate of 450Mbit, but the best real-world bitrate I have seen is ~270Mbit, so my max of 180Mbit with 2x2 seems optimal. I have gotten ~180Mbit with my Nexus 9, my Asus USB-N53 (Linux), and my TP-Link TL-WN822Nv3 (Linux).

The only 802.11ac compliant device I have is a Nexus 9 (which I think is 2x2).

I test with iperf.


PS - Are using that PCE-AC68 with Linux? If so, does it work well?

 

[pete y testing]

PS - Are using that PCE-AC68 with Linux? If so, does it work well?

nope just a windows 10 pc with 8 gig ram

the pce-ac68 works well with all the asus routers i have here , i also have a tp link t8e which doesnt work so well esp with its stock antennas , the t9e aparently works better but is limited to euro wifi channels

I test with iperf.

suggest as a better throughput real world measurement you test by just transferring a vid file over sambe from one client on the wifi and one client connected directly to the router via ethernet

 

[Nullity]

suggest as a better throughput real world measurement you test by just transferring a vid file over sambe from one client on the wifi and one client connected directly to the router via ethernet

All my regular tranfers matched the iperf results. IIRC, the multi-stream iperf tests were negligably different from single-stream.

Recently, non-iperf (netcat) file transfers with my Nexus 9 result in ~22Mbyte/sec file transfers, which mirrors my ~180Mbit iperf results with 2x2 WiFi clients. I have not yet gotten a skewed result from iperf (version 2 or 3) over the past 2 years of usage. Benchmarks usually replicate normal traffic, in my limited experience.

Have you had troubles with iperf?

 

[L&LD]

My RT-N66U is a few years old and it's 5GHz band has never been strong enough to reach the whole house. My wife and daughter's iPhones sometimes loose connection even on 2.4GHz.

Would the RT-AC68U improve signal strenght and connection quality?

Thanks,
Gilson

That is true. The RT-N66U does not have the prowess of a three year newer design AC1900 class router. And the RT-AC68U is an excellent example of an AC1900 class router (with the appropriate firmware).

I just upgraded my RT-N66U for an RT-AC68U and saw an immediate effect on my networking tasks. Not 'huge', but significant enough. On 2.4GHz I am finally able to hit my ISP speeds (50/10) and even surpass them (57/12). On the 5GHz band, I have more range and throughput vs. the RT-N66U router (both placed in the same location and antennae orientation).

Wired or wireless, webpages are noticeably faster to load (I would guess the latency to be half of what it was) and network shares seem directly connected to any computer I am using.

Both routers are running RMerlin firmware (380.57), but have 3 antennae/3 streams but a single 600MHz core vs. dual 800MHz cores are noticeable and appreciated. I can only imagine how the RT-AC88U and the RT-AC5300 with dual 1400MHz cores would feel like in my environment.

But, before you go buying a new router, what have you done to maximize the RT-N66U's potential?

Is it located centrally to the area that is needed to be covered? Is it placed at least 10' (3 metres) above ground level (not just the floor)? Does it have at least 3' (1 metre) clearance all around and above it too?

Have you tested channels 1, 6 and 11 in the 2.4GHz band and each channel in the 5GHz band for highest throughput and lowest latency?

Have you fully reset the router to factory defaults and then manually and minimally configure the router to secure it and connect to your ISP?

Which firmware and version are your running? Have you tried john9527's or hggomes firmware to see if you see any benefits in your environment?

Unless and until you do the above, buying and testing a new router is not recommended. You will not know where you were with your existing equipment and any benefits the shiny new router offers (or not) may be an unfair comparison of new vs. old too.

http://www.snbforums.com/threads/no...l-and-manual-configuration.27115/#post-205573

http://www.snbforums.com/threads/asuswrt-merlin-378-55-3_hgg-final-mod.26524/page-2#post-199549

http://www.snbforums.com/threads/asuswrt-merlin-3-0-0-4-374-38-is-out.14691/page-19#post-99500



As you can see from the links above, an RT-AC56U may be a suitable upgrade too. Depending if you have any 3 antennae clients that you need to have at full wireless performance (the RT-AC56U has internal 2x2 antennae).

If the RT-AC56U is an option for you (depending on your clients), then note that it has the same core hardware as the RT-AC68U and has just as much of an recommendation from me as the RT-AC68U.

 

[pete y testing]

Have you had troubles with iperf?

i prefer not to use it as it doesnt use the same protocols as standard smb / samba and the results and a max average and not a means average of a transfer over the time taken to transfer , it is just my choice to give real world MB/s results that 90% of users can expect when they do a transfer over windows and looking at the windows transfer window , plus i use networx speed meter to record means average throughput in MB/s

 

[Nullity]

i prefer not to use it as it doesnt use the same protocols as standard smb / samba and the results and a max average and not a means average of a transfer over the time taken to transfer , it is just my choice to give real world MB/s results that 90% of users can expect when they do a transfer over windows and looking at the windows transfer window , plus i use networx speed meter to record means average throughput in MB/s

iperf supports both the UDP & TCP protocol. Does samba use some other protocol? Throughput benchmarks are most useful when they the most universal, low-level protocol and avoid higher-level protocol overheads.

Not everyone uses Windows OS. (For example, does any common, consumer router run Windows?)

 

[pete y testing]

Not everyone uses Windows OS.

but over 80% do

(For example, does any common, consumer router run Windows?

its the clients that are important when measuring throughput , eg the end points , what the modem runs is irrelevant

Does samba use some other protocol?

yes

see

https://en.wikipedia.org/wiki/Samba_(software)

 

[Nullity]

but over 80% do

True, but ~100% of internet users use TCP or UDP.

its the clients that are important when measuring throughput , eg the end points , what the modem runs is irrelevant



yes

see

https://en.wikipedia.org/wiki/Samba_(software)

All modern implementations of SMB run over TCP/UDP.


I think we have gone off topic though.

 

[oguime]

so my suggestion is keep the rt-n66u and use it fir the access point in another part of the house and yes get the rt-ac68u for your main router

Hi Pete,

I believe you are right when you point to my environment as the main issue, since I live in a brick walls apartment. I do however have ethernet on most of the rooms and will probably add an AP.

Not sure if I should use the RT-N66U though... How well would an AC router hand out to a N AP?

Thanks,
Gilson

 

[oguime]

Is it located centrally to the area that is needed to be covered? Is it placed at least 10' (3 metres) above ground level (not just the floor)? Does it have at least 3' (1 metre) clearance all around and above it too?

Have you tested channels 1, 6 and 11 in the 2.4GHz band and each channel in the 5GHz band for highest throughput and lowest latency?

Have you fully reset the router to factory defaults and then manually and minimally configure the router to secure it and connect to your ISP?

Which firmware and version are your running? Have you tried john9527's or hggomes firmware to see if you see any benefits in your environment?

Hi L&LD,

It's as centered as Heatmapper suggested and the WAF allowed! But it is only 1.5m high...

Both 2.4GHz and 5GHz bands were configured on Merlins 378.56_2.

I hope a new AC router is able to give me just a slight increase on reliability and performance, otherwise I'll try an AP on another room.

On another topic, did you use the USB 3.0 port to add shared storage? I was tinking if it would be able to replace a dedicated NAS for now, but it seems the eprformance is not good...

Thanks,
Gilson