Asusmerlin version 42 slower than stock 5656

[tommyv2]

.42 is definitely faster than 5656 because Merlin lets me do a 1-button overclock

 

[Biite]

Make sure you disable unnecessary features such as VPN servers, IPTraffic (which would force HW acceleration to be disabled) and IPv6 tunneling (Asus disables HW acceleration when using 6in4 tunnels on the N66 and AC66).

Hmmm. that explain some of my problems when using 374.42 . When I enable a 6in4 tunnel to he.net my maximum speed is about 200Mbit on a 300Mbit line with a RT-N66U.
When using 374.41 with the same configuration it maxes out on 300Mbit.

Something must have changed between .41 and .42 with ipv6 tunneling and the HW acceleration. Is there a check (e.g. on the CLI) to find out if HW acceleration is disabled/enabled?

 

[RMerlin]

Hmmm. that explain some of my problems when using 374.42 . When I enable a 6in4 tunnel to he.net my maximum speed is about 200Mbit on a 300Mbit line with a RT-N66U.
When using 374.41 with the same configuration it maxes out on 300Mbit.

Something must have changed between .41 and .42 with ipv6 tunneling and the HW acceleration. Is there a check (e.g. on the CLI) to find out if HW acceleration is disabled/enabled?

Due to compatibility issues, Asus disables HW acceleration when you use a 6in4 tunnel on the RT-N66U and RT-AC66U. This would explain your performance drop when using a tunnel.

 

[cofetym]

One man manages to get credit from Asus in their builds and that's RMerlin.

 

[viperzxz]

No, different issue (sorry for the thread hijack).

I haven't tested wifi on the AC56 recently, but from what I remember I'm using the same driver Asus is using in 5656, so I would expect the same performance. Try the usual debugging steps: setting a static channel, re-creating the wireless profile on your computer, etc...

Two additional settings that has made a different for other users were Beamforming (try disabling them both), and the 802.11d+h setting (this one especially impacts Apple devices).

I'll try to check next weekend in case the wireless driver found in Asus's GPL might not be the same they are using in the released firmware itself - it has happened in the past.

I tried the usual debugging steps you mentioned before I posted. I'll try the beamforming and the 802.11d+h. For sure, beamforming is enabled.

Every time I restarted the router I also changed the SSID's password so that I would make sure only one device was connected at a time.

Thanks for the help. Will try your suggestions.

 

[MonkY]

I love RMerlin too, but i was forced to give up on version 374.42 for AC66U because of slower internet speed on PPoE using Gigabit WAN connection.

Strange thing with these upgrades:
374.40 - arround 750Mbps
374.41 - arround 600-650Mbps
374.42 - under or arround 500Mbps

Please don't let 374.43 go to 400Mbps or under...

Your ISP is pushing 750 Mbps over PPPoE? Talk about a waste...

My ISP is actually pushing over 950Mbps over PPPoE and that's probably why Romania is the third country in the top provided by netindex.com (speedtest). I don't see why is this a waste... but it's ok by me.

You might be using features that forces part of that traffic to not go through CTF, which is essential for WAN speeds beyond 150 Mbps on the RT-AC66U. Make sure you disable unnecessary features such as VPN servers, IPTraffic (which would force HW acceleration to be disabled) and IPv6 tunneling (Asus disables HW acceleration when using 6in4 tunnels on the N66 and AC66).

I have checked and double checked. I have no VPN, IPTraffic or IPv6 enabled.

It might also be worth upgrading to a router with a faster CPU, such as either the RT-AC56 or AC68 to better handle such WAN speeds.

Really? I'm just showing you how this one can get upto 750Mbps with a different firmware but 500Mbps with the latest one and you're telling me to buy another router? It's like you're not trusting your own work.

Regarding the sollution to get a router with a faster CPU, let me tell you that RT-N56U (which has a much slower CPU than AC66U) with padavan firmware is getting over 850Mbps, in my location. So no, faster CPU doesn't have to mean faster download speed.

Also, there is a chinese router from TP-Link, called WDR3600, that's actually like 4-5 times cheaper than AC68 or AC66U... and it can get 950Mbps DL on the same provider. I will post a screenshot if you can't believe it.

I think Asus is starting to lag behind much cheaper ones... and the firmware updates seems to go worst in terms of speed, instead of getting better.

 

[RMerlin]

My ISP is actually pushing over 950Mbps over PPPoE and that's probably why Romania is the third country in the top provided by netindex.com (speedtest). I don't see why is this a waste... but it's ok by me.

Lots of overhead, both in traffic (from the PPP encapsulation) and CPU-wise (for the need to process the PPP packets).

Really? I'm just showing you how this one can get upto 750Mbps with a different firmware but 500Mbps with the latest one and you're telling me to buy another router? It's like you're not trusting your own work.

Regarding the sollution to get a router with a faster CPU, let me tell you that RT-N56U (which has a much slower CPU than AC66U) with padavan firmware is getting over 850Mbps, in my location. So no, faster CPU doesn't have to mean faster download speed.

CPU-wise, the RT-AC66U is only capable of pushing around 150 Mbps. Anything beyond that is achieved through Broadcom's closed-source CTF (so-called "hardware acceleration"), which does it at the expense of preventing many features from working properly - in fact, anything that relies on the FORWARD chain.

Asus's workaround for this limitation is to mark certain packets so they bypass CTF. So if you have a large portion of that traffic bypassing CTF, that portion of the traffic will seriously impact your global throughput capabilities as it will increase the load on your CPU. For example, any traffic going through a port forward will seriously impact the router's global throughput, as that traffic bypasses the HW acceleration, and becomes a CPU burden.

Also, there is a chinese router from TP-Link, called WDR3600, that's actually like 4-5 times cheaper than AC68 or AC66U... and it can get 950Mbps DL on the same provider. I will post a screenshot if you can't believe it.

That is an Atheros device, not Broadcom. Just like the RT-N56U, it doesn't suffer from Broadcom's shennanigans that they must use to compensate for the fact that their SoC can't really process gigabit throughputs.

Anything at those speeds will always be iffy on Broadcom's devices. That's why I recommend using a device with a faster CPU if you wish to stick with Broadcom devices, otherwise it gets very easy for things to seriously hamper the router's throughput.

I think Asus is starting to lag behind much cheaper ones... and the firmware updates seems to go worst in terms of speed, instead of getting better.

Just go with either their faster Broadcom-based routers (which was my suggestion), or their Ralink-based devices, such as the RT-N56U and RT-N65U.

Broadcom will always be a problem when you start offering more advanced features, as it will interfere with their ability to accelerate the throughput.

 

[RMerlin]

BTW, here's a benchmark run between WAN and LAN on my RT-AC66U:

F:\Share>iperf -s
------------------------------------------------------------
Server listening on TCP port 5001
TCP window size: 64.0 KByte (default)
------------------------------------------------------------
[232] local 192.168.10.100 port 5001 connected with 192.168.10.227 port 1222
[ ID] Interval       Transfer     Bandwidth
[232]  0.0-30.0 sec  3.27 GBytes    935 Mbits/sec

The client was a laptop on the LAN side of the RT-AC66U, and the server was a desktop on the WAN side of the RT-AC66U. (the reason the client shows as being on the same subnet is because of NAT - that's the WAN IP of the RT-AC66U. Client is actually in the 192.168.1.0/24 subnet).

This was (obviously) with HW acceleration enabled, and DHCP (not PPPoE). The RT-AC66U was running 374.43-alpha2.

 

[_robodoc_]

@merlin

Those limitations of the Broadcom SOC you mentioned above; can this be the reason for the lack of DFS support for the RT-AC66U ?

 

[icece]

@merlin

Those limitations of the Broadcom SOC you mentioned above; can this be the reason for the lack of DFS support for the RT-AC66U ?

Nope. DFS relies on a different functionality of the wireless driver.

 

[RussellInCincinnati]

Original Post describing an unknown router model: "Asusmerlin version 42 slower than [Asus] stock [firmware] 5656"

Not enough details to define an issue that is worth Merlin's or anyone else's time. It's like saying "Merlin firmware is only 50% as good as Asus stock firmware, what should I do?"

So am hoping Merlin will not even considering spending time on this situation until the original poster clarifies:
(1) what router are you using?
(2) which settings have you changed, if any, compared to all the default settings in Merlin version ...42?
(3) Please clarify how long it takes to copy a file and/or complete some sort of Internet speed test, telling us what your measured speed difference is between Merlin and Asus stock firmwares. So far you are only telling us about some sort of "reported link rate" like "866mbps" which does not tell us whether the Asus stock firmware link is usefully twice as fast as the Merlin code. Reported link rates do not tell you what the stability and/or error-and-retry rates are going to be over a given connection. Maybe an error-free connection at a lower link rate will be usefully faster than an error-prone one at a higher nominal rate.

If you want full functionality and stability, better stick to OFW [Asus Original Firmware].

This is usually poor advice, to advise the original poster to stop using a piece of software with hundreds of features and behaviors, even before you have a useful definition of what one single "practical problem" is.

Merlin's build not only add features, but also break some OFW function. After all it's not official...Like "PPPoE + ipv6 tunnel + CTF resulting not being able to access ipv6 sites" on AC68.

It is odd that you would mention an obscure problem that you are having with Merlin firmware on the not-yet-maturely-implemented-throughout-the-world-internet ipV6 protocol, connecting to websites you do not specify, using an internet service provider you do not specify, using a host of Merlin non-default settings that you also do not specify...as if that has the slightest relevance to the original poster's observation of reported 802.11ac WiFi link rates.

It would be much more logical for you to post your observed problem in a separate discussion thread, than to lead this thread off-topic.

maybe there added feature and modification does conflict with something ASUS modified. It's known to all especially to coders that added features can always potentially introduce some bugs.

This is off-topic to the question of whether or not Merlin firmware on an unknown router with unknown settings and unknown clients gives faster 'ac connections than stock Asus firmware, as well as being a uselessly vague observation.

And user base on OFW is way much larger than 3rd partys', so the probability of finding bugs in OFW is far greater than in 3rd partys'.

Since Merlin might be properly understanding and fixing (say for example) 10% of the "bugs" reported, and Asus for all we know may be choosing to fix no more than 1% of the bugs reported, this is not a useful (or on-topic) observation.

Further more, ...But one can't complain to 3rd party firmware author when something went wrong.

Ah, you have deduced from theoretical principles, that this forum does not exist.

For me, without this problem, I definitely love Merlin's build over OFW, but with this problem presented and I have no right to complain to a voluntary 3rd party FW developer, I can only go back to OFW.

You are among other things implying that the number of important problems caused by Merlin's non-ASUS code additions that Merlin could fix, but doesn't, is a bigger deal than the number of ASUS code problems that Merlin fixes.

Why don't you start a separate thread where such an observation would be on-topic? Be sure to provide a decently large and specific list of both Merlin pro's and con's. Without such a list of pro's and con's, vague speculations as to why Merlin's code "should be more buggy" than Asus's sounds like non-useful "searching for negatives".

 

[RMerlin]

Hi!

Experiencing issues since .42. Wan disconnect 3-4 times/week, Wi-Fi stay stable
i called my ISP videotron to ask them if they see something wrong with my modem and they got nothing on their logs.I loose WAN 1 or 2 mins
I tried everything, RESTORE factory default, WPS trick button...etc.
First time i got an issues with merlin

RT-n66u running now since 2 weeks on OFW .5517 and no disconnects at all.

No issue here with 30 Mbits Teksavvy Cable (which actually is on Vidéotron's network). My modem is a DCM475.

Check your system log. It will tell you if the issue is the router is losing sight of the modem for 2-3 mins (which means your modem either rebooted, or lost sync). I would also suggest replacing the Ethernet cable between router and modem.

I would also recommend resetting your connection by unplugging the power of your modem for 5-10 mins (if you have their VoIP combo device, also pull the battery from it).

 

[damir]

I am running Start's cable 60\10 (Ontario) , no issues as well.

 

[jgrove]

No issues with me, i use a 152 meg downlink with a 10 meg uplink. This firmware works fine, much better than the Asus firmware.

 

[Grievous Angel]

And user base on OFW is way much larger than 3rd partys', so the probability of finding bugs in OFW is far greater than in 3rd partys'.

May be larger, but I doubt this is true. This "much larger" user base includes your average joe who buys his router from Best Buy, can connect to Facebook, and therefore never touches it again.

The nerds all congregate here.

And what are you going to do if you find a bug in OFW? Contact ASUS support? Do you KNOW how painful and slow it can be dealing with them?

Merlin is unpaid and yet is very responsive to issues, releasing speedy updates to problems.

I haven't had an issue with any of his builds (AC66U) and have updated compulsively.

 

[IAAI]

My ISP is actually pushing over 950Mbps over PPPoE and that's probably why Romania is the third country in the top provided by netindex.com (speedtest). I don't see why is this a waste... but it's ok by me.

I have checked and double checked. I have no VPN, IPTraffic or IPv6 enabled.

Really? I'm just showing you how this one can get upto 750Mbps with a different firmware but 500Mbps with the latest one and you're telling me to buy another router? It's like you're not trusting your own work.

Regarding the sollution to get a router with a faster CPU, let me tell you that RT-N56U (which has a much slower CPU than AC66U) with padavan firmware is getting over 850Mbps, in my location. So no, faster CPU doesn't have to mean faster download speed.

Also, there is a chinese router from TP-Link, called WDR3600, that's actually like 4-5 times cheaper than AC68 or AC66U... and it can get 950Mbps DL on the same provider. I will post a screenshot if you can't believe it.

I think Asus is starting to lag behind much cheaper ones... and the firmware updates seems to go worst in terms of speed, instead of getting better.

why did u post the pic ? I m offended [emoji23]

Looks like i have Dial-UP [emoji13]

How much do you pay for the service ?

 

[MonkY]

why did u post the pic ? I m offended [emoji23]

Looks like i have Dial-UP [emoji13]

How much do you pay for the service ?

Sorry, didn't mean to make you feel bad.

The full price is 59 RON (VAT included). That's arround 13.42 EURO / 18.27 USD.

But i'm paying 49 RON (arround 15 USD) because i also have TV cable from them. They also throw a free UNLIMITED USB stick with 3G (works only national - perfect inside my iPad).

It's this package: http://www.rcs-rds.ro/internet-digi-net/fiberlink?t=internet-fix&pachet=digi_net_fiberlink_1000

 

[IAAI]

Sorry, didn't mean to make you feel bad.

The full price is 59 RON (VAT included). That's arround 13.42 EURO / 18.27 USD.

But i'm paying 49 RON (arround 15 USD) because i also have TV cable from them. They also throw a free UNLIMITED USB stick with 3G (works only national - perfect inside my iPad) with Gigabit connection.

It's this package: http://www.rcs-rds.ro/internet-digi-net/fiberlink?t=internet-fix&pachet=digi_net_fiberlink_1000

f**k i m paying 65 for 20/2 internet with slandered tv

could you please try Merlin's build 43-2 and give us your feedback ?

 

[MonkY]

could you please try Merlin's build 43-2 and give us your feedback ?

Well, i'm happy! 42 was really messed up for me (upto 500Mbps on PPoE), but seems like 43 is now back on track (i'm able to go over 800Mbps now). So, thanks Merlin (or Asus) for fixing the issue! Now, let's see if it can go over 900!

I will also attach a screenshot to prove it!

 

[vincom]

Well, i'm happy! 42 was really messed up for me (upto 500Mbps on PPoE), but seems like 43 is now back on track (i'm able to go over 800Mbps now). So, thanks Merlin (or Asus) for fixing the issue! Now, let's see if it can go over 900!

I will also attach a screenshot to prove it!

damn, on internet offerings/speeds/pricing canada is even worse than 3rd world status, we're living in the dinosaur age, scratch that.....single cell age