Asuswrt-Merlin 3.0.0.4.354.27 Beta 1

[DrTeeth]

I have no complaints about the speed of loading of web pages with H/W accel disabled. Now that I have seen the list of culprits, I may do some experimenting.

By the by, the performance of Firefox (page loading) reduced drastically from v18 (19 + 20 terrible) and I have moved to PaleMoon permanently; and full fast-as-lightening performance is restored.

Thanks

DrT

 

[RMerlin]

After enabling and disabling many settings, it turned out to be QOS, what killing the H/W acceleration. Is there going to be a fix for this later to have QOS and the H/W acceleration working?

The HW acceleration code is closed source, so only Asus could answer that one. We don't even know for sure how this code works or what it actually does.

 

[Dusan]

I agree, with the H/W acceleration enabled I find webpages load much better. Overall better performance. Hope this can be fixed in later builds.

Unless your internet connection is faster than ~150Mb/s, then any "faster webpages loads with H/W acceleration enabled" is probably just the placebo effect. Did you do some real timing tests, or is it just a feeling? (Or is your internet connection really that fast?)

Also, if your computer is connected via wireless then the H/W acceleration really doesn't matter. The benefit only shows on gigabit wired links.

 

[RMerlin]

Whilst we are talking about hardware acceleration, does it actually produce noticeably better performance? It seems a shame that so many useful features, that themselves could possibly benefit from it, disable it.

UPDATE:- Found some info, written by our favourite firmware developer , "HW acceleration is mostly of use if you have a very high speed Internet connection (over 100 Mbits)". He also wrote that the module is closed source.

DrT

Someone did some benchmark a while ago. Seems like you need HW acceleration if your Internet connection is about 200 Mbits or faster. Below that it shouldn't have any major impact, unless your router is highly loaded by other things (such as torrenting).

HW acceleration used to be even more restrictive, it originally didn't even work with port forwarding. Back then it looked more like a marketing gimmick to get higher benchmark numbers in reviews. I'd say it's actually useful now, since people with such fast Internet connection might be less in need of using QoS. Sure would like to see Broadcom/Asus make it compatible with more typical usage scenarios, but not sure if it's technically possible, not knowing how this module works.

 

[DrTeeth]

usage scenarios, but not sure if it's technically possible, not knowing how this module works.

There was an article which said, about H/W acceleration, "this involves the routing chip processing the packets instead of the router’s CPU". It was in a review (for another Asus router) and don't know whether it is rubbish or not.

 

[got_milk]

There was an article which said, about H/W acceleration, "this involves the routing chip processing the packets instead of the router’s CPU". It was in a review (for another Asus router) and don't know whether it is rubbish or not.

That's probably for the RT-N56U, as far as I know it has a dedicated chip for hardware-accelerated NAT (that's how it gets gigabit WAN <-> LAN performance).

I wonder if CTF is simply an implementation of cut-through switching, but for packet routing? Cut-through switching doesn't perform a CRC check on packets and can make forwarding decisions as soon as it knows the destination MAC address. That would explain why it would be shut off for things like QoS, where packet prioritization would need to be occurring, which couldn't be done with such little data.

 

[rumnyc]

Someone did some benchmark a while ago. Seems like you need HW acceleration if your Internet connection is about 200 Mbits or faster. Below that it shouldn't have any major impact, unless your router is highly loaded by other things (such as torrenting).

HW acceleration used to be even more restrictive, it originally didn't even work with port forwarding. Back then it looked more like a marketing gimmick to get higher benchmark numbers in reviews. I'd say it's actually useful now, since people with such fast Internet connection might be less in need of using QoS. Sure would like to see Broadcom/Asus make it compatible with more typical usage scenarios, but not sure if it's technically possible, not knowing how this module works.

I currently have PPPoE and current speed is 100Mbits (copper).

This weekend I am upgrading to 300Mbits (fiber).

I do use port forwarding etc. so will report back my findings. Will do a speed test with h/w acceleration on and off.

 

[Kenneth]

https port

I updated to this firmware version from 270.26b and make a Factory Default Reset in my RT-N66U. All works perfect, it's an excellent work.

I only have a trouble:

I want to setup the "https access" same like previous firmware, but I can't configure the "Web Access From Wan" Port on 8443. A error message says: "This port had been used for HTTP LAN PORT" (See the image)

What am I doing wrong?

Thanks!

Kenneth

 

[RMerlin]

I updated to this firmware version from 270.26b and make a Factory Default Reset in my RT-N66U. All works perfect, it's an excellent work.

I only have a trouble:

I want to setup the "https access" same like previous firmware, but I can't configure the "Web Access From Wan" Port on 8443. A error message says: "This port had been used for HTTP LAN PORT" (See the image)

What am I doing wrong?

Thanks!

Kenneth

Just above you have the WAN port also set to 8443. You probably can't have both on 8443 at the same time.

 

[RMerlin]

Asus devs have confirmed the issue with the 5 GHz band and Intel-based cards on the RT-N66U. They're looking into it.

 

[CD122]

I updated to this firmware version from 270.26b and make a Factory Default Reset in my RT-N66U. All works perfect, it's an excellent work.

I only have a trouble:

I want to setup the "https access" same like previous firmware, but I can't configure the "Web Access From Wan" Port on 8443. A error message says: "This port had been used for HTTP LAN PORT" (See the image)


What am I doing wrong?

Thanks!

Kenneth

I am currently using the stock Asus fw 354 and having the same issue. I was unable to make any changes on the "system" tab unless I changed both the port number in "HTTPS Lan port" and the "Port of Web access from WAN" fields simultaneously. Then before hitting apply, I made changes to other fields I wanted and then hit apply.

So try changing the ports to 8445 or some other number, change any other settings you want, then click apply. It worked for me.

 

[Kenneth]

Just above you have the WAN port also set to 8443. You probably can't have both on 8443 at the same time.

I could do it, but changing fields simultaneously just like user CD122 explains in the other reply. It seems it's present in 354 stock firmware too.

I think it's a bug because don't let you to change any "system" tab setting unless you configure this fields simultaneously.

 

[Kenneth]

I am currently using the stock Asus fw 354 and having the same issue. I was unable to make any changes on the "system" tab unless I changed both the port number in "HTTPS Lan port" and the "Port of Web access from WAN" fields simultaneously. Then before hitting apply, I made changes to other fields I wanted and then hit apply.

So try changing the ports to 8445 or some other number, change any other settings you want, then click apply. It worked for me.

Thanks! It works!

 

[Trent Bates]

It sounds like I might be a bit late with this information since ASUS developers are looking into the 5GHz issues.
I did some testing yesterday with Merlin's 4.354.27Beta1 and then reverted to Merlin's 4.270.26b and ran similar tests. All testing was done with the nvram cleared and only minimal changes to the user settings like SSID name and password, IP address, Time zone, manual channel selection, etc.
I found one interesting behavior with 4.354 on 802.11a that supports what I had noticed a few days ago.

To those that aren't up to speed yet or might get the wrong idea, although I used Merlin's releases for this test, this isn't a problem with his releases. I have experienced the same issues with Asus's versions of the firmware.

I'm not characterizing this as a "broken" driver issue since I don't know if this is unique to my laptop's particular WLAN interface or not. But I am saying that my laptop works much better with the drivers in 4.270 than it does with the drivers in 4.354.
The laptop's device is a Dell 1470 Dual Band Mini-PCI with the BCM4319 / BCM2060 chipset. Drivers are updated to the newest version available dated Oct. 2006.

Here's what I did:
Turned everything else off. (I typically have 22-27 devices connected.)
Started one Desktop PC running 100 Mbps to the RT-N66u.
Started one Laptop running 54 Mbps on 802.11a with other bands disabled.
Both running Windows XP SP3. (Everything I'm working with is older for this test.)

Cleared nvram and loaded 4.354... and then cleared nvram again. (Overkill)
Set necessary settings to use router.

I used the Wireless Card utility in my laptop to monitor signal and connection speeds. This utility is always on in the tray and appears to have no appreciable impact on testing when open.
I also used LAN Speed Test (Lite) 1.2.0 as shown in the screenshot.

Physically I was sitting about 10-12 feet from the RT-N66u on a shelf in a closet. (1 wall between it and I) I tried to stay in the same position throughout the tests so that my own body didn't cause any signal changes.

One big surprise that I really feel is worth mentioning is that channel selection on 802.11a made a noticeable difference for me. I had been using the lower channels (36, 40, 44) in the belief that lower frequencies would travel farther but found that I got a stronger signal on channels 149 and 157. It may have to do with the laptop's antennas so others might notice the opposite.

The difference that I noticed between the two drivers seems to be how they handle 20MHz vs 40MHz.
I've been running 802.11a set at 40MHz for 6 months or more since I'm the only one in the neighborhood that I can detect on 802.11a.

--------
With 4.354... I started with 40 MHz and noticed again that my connect speeds stayed in the 6.0 to 18.0 Mbps range according to the Wireless Utility.
LAN transfers were wildly varying from 1.x to 6.x Mbps. I couldn't get consistent speeds in the tests.

I changed to 20 MHz in various ways. I turned on Legacy mode, or set to 20 MHz in Auto mode and tested a bunch of times. One big clue was that running at 20 MHz doubled my connect speeds. Instead of connecting at 6-18 Mbps, I was consistently connecting at 12-36 Mbps.

Once I got that sorted out, I noticed that every time I started a speed test, the connect speed would drop for the duration of the test. I could be sitting at 36 Mbps and then start the test and see the connect speed drop and fluctuate around 18-24 Mbps. I was getting average throughput speeds of 12-13 Mbps writing and 12-14 Mbps reading.

--------
Then I cleared nvram, loaded 4.270.26b, cleared nvram again, and set the basic settings to get running. I set the 802.11a channel to 149 again and set to 20 MHz.
With 4.270.26b (and Asus's 4.270) I connect at 54 Mbps and usually it stays there. I ran dozens of tests and tweaked a few settings on the RT-N66u to see what difference they made. LAN speed tests are a consistently 25-28 Mbps writing and 22-27 Mbps reading.
The screenshot below is from 30+ feet (through 3 walls) away instead of the 10-12 feet (and 1 wall) I tested with yesterday. (Our rooms aren't that small, I'm talking hallways and corners. )

I also noticed that I could run at 40 Mhz and Auto mode without affecting my connect speeds or throughput. The older driver seems to be much more compatible with my laptop.
--------

For reference, I ran a few LAN speed tests with the laptop wired at 100 Mbps to the RT-N66u as well and got 65-70 Mbps writing and 73-80 Mbps reading. That seems to support that the speeds I was getting on a good 48-54 Mbps connection are within the proper range.


Anyway, I hope this info helps! I'm going to have to stick with 4.270.x unless Asus releases a different driver for 802.11a or I find a newer Mini-PCI card that works better with the new drivers. It's kind of a shame because I like some of the new features I saw in 4.354.

 

[Hothersale]

With respect to 5Ghz channel selection, I believe the lower channels may be limited in their max transmission power. Here's one thread for reference: http://forums.smallnetbuilder.com/showthread.php?t=6430

 

[Trent Bates]

Here's one thread for reference: http://forums.smallnetbuilder.com/showthread.php?t=6430

That makes sense. Thank you!

 

[Reolin]

I've been having USB flash drive installed for a long time. Today I noticed

Apr 10 19:54:54 kernel: scsi 0:0:0:0: rejecting I/O to dead device
Apr 10 19:54:55 kernel: scsi 0:0:0:0: rejecting I/O to dead device
Apr 10 19:54:56 kernel: scsi 0:0:0:0: rejecting I/O to dead device
Apr 10 19:54:57 kernel: scsi 0:0:0:0: rejecting I/O to dead device
Apr 10 19:54:58 kernel: scsi 0:0:0:0: rejecting I/O to dead device
Apr 10 19:54:59 kernel: scsi 0:0:0:0: rejecting I/O to dead device
Apr 10 19:55:00 kernel: scsi 0:0:0:0: rejecting I/O to dead device

in system log, these entries didn't stop appearing after I removed my usb drive, but stopped after reboot.

 

[Werthe]

Hi,

What's up with this General Log...

Apr 10 18:28:45 nmbd[585]: [2013/04/10 18:28:45, 0] nmbd/nmbd_incomingdgrams.crocess_local_master_announce(309)
Apr 10 18:28:45 nmbd[585]: process_local_master_announce: Server ComputerName at IP 192.168.0.101 is announcing itself as a local master browser for workgroup WORKGROUP and we think we are master. Forcing election.
Apr 10 18:28:45 nmbd[585]: [2013/04/10 18:28:45, 0] nmbd/nmbd_become_lmb.c:unbecome_local_master_success(149)
Apr 10 18:28:45 nmbd[585]: Samba name server RT-AC66U has stopped being a local master browser for workgroup WORKGROUP on subnet 192.168.0.1
Apr 10 18:29:04 nmbd[585]: [2013/04/10 18:29:04, 0] nmbd/nmbd_become_lmb.c:become_local_master_stage2(392)
Apr 10 18:29:04 nmbd[585]: Samba name server RT-AC66U is now a local master browser for workgroup WORKGROUP on subnet 192.168.0.1
Apr 10 18:36:43 nmbd[585]: [2013/04/10 18:36:43, 0] nmbd/nmbd_incomingdgrams.crocess_local_master_announce(309)
Apr 10 18:36:43 nmbd[585]: process_local_master_announce: Server ComputerName at IP 192.168.0.101 is announcing itself as a local master browser for workgroup WORKGROUP and we think we are master. Forcing election.
Apr 10 18:36:43 nmbd[585]: [2013/04/10 18:36:43, 0] nmbd/nmbd_become_lmb.c:unbecome_local_master_success(149)
Apr 10 18:36:43 nmbd[585]: Samba name server RT-AC66U has stopped being a local master browser for workgroup WORKGROUP on subnet 192.168.0.1
Apr 10 18:37:03 nmbd[585]: [2013/04/10 18:37:03, 0] nmbd/nmbd_become_lmb.c:become_local_master_stage2(392)
Apr 10 18:37:03 nmbd[585]: Samba name server RT-AC66U is now a local master browser for workgroup WORKGROUP on subnet 192.168.0.1


Any help?

 

[RMerlin]

Hi,

What's up with this General Log...

Apr 10 18:28:45 nmbd[585]: [2013/04/10 18:28:45, 0] nmbd/nmbd_incomingdgrams.crocess_local_master_announce(309)
Apr 10 18:28:45 nmbd[585]: process_local_master_announce: Server ComputerName at IP 192.168.0.101 is announcing itself as a local master browser for workgroup WORKGROUP and we think we are master. Forcing election.
Apr 10 18:28:45 nmbd[585]: [2013/04/10 18:28:45, 0] nmbd/nmbd_become_lmb.c:unbecome_local_master_success(149)
Apr 10 18:28:45 nmbd[585]: Samba name server RT-AC66U has stopped being a local master browser for workgroup WORKGROUP on subnet 192.168.0.1
Apr 10 18:29:04 nmbd[585]: [2013/04/10 18:29:04, 0] nmbd/nmbd_become_lmb.c:become_local_master_stage2(392)
Apr 10 18:29:04 nmbd[585]: Samba name server RT-AC66U is now a local master browser for workgroup WORKGROUP on subnet 192.168.0.1
Apr 10 18:36:43 nmbd[585]: [2013/04/10 18:36:43, 0] nmbd/nmbd_incomingdgrams.crocess_local_master_announce(309)
Apr 10 18:36:43 nmbd[585]: process_local_master_announce: Server ComputerName at IP 192.168.0.101 is announcing itself as a local master browser for workgroup WORKGROUP and we think we are master. Forcing election.
Apr 10 18:36:43 nmbd[585]: [2013/04/10 18:36:43, 0] nmbd/nmbd_become_lmb.c:unbecome_local_master_success(149)
Apr 10 18:36:43 nmbd[585]: Samba name server RT-AC66U has stopped being a local master browser for workgroup WORKGROUP on subnet 192.168.0.1
Apr 10 18:37:03 nmbd[585]: [2013/04/10 18:37:03, 0] nmbd/nmbd_become_lmb.c:become_local_master_stage2(392)
Apr 10 18:37:03 nmbd[585]: Samba name server RT-AC66U is now a local master browser for workgroup WORKGROUP on subnet 192.168.0.1


Any help?

There is nothing wrong here. The master browser of a network is determined through an election. If there's a conflict, then a new election gets done to chose the master browser. This is simply the logged activity related to that process.

 

[RMerlin]

I've been having USB flash drive installed for a long time. Today I noticed

Apr 10 19:54:54 kernel: scsi 0:0:0:0: rejecting I/O to dead device
Apr 10 19:54:55 kernel: scsi 0:0:0:0: rejecting I/O to dead device
Apr 10 19:54:56 kernel: scsi 0:0:0:0: rejecting I/O to dead device
Apr 10 19:54:57 kernel: scsi 0:0:0:0: rejecting I/O to dead device
Apr 10 19:54:58 kernel: scsi 0:0:0:0: rejecting I/O to dead device
Apr 10 19:54:59 kernel: scsi 0:0:0:0: rejecting I/O to dead device
Apr 10 19:55:00 kernel: scsi 0:0:0:0: rejecting I/O to dead device

in system log, these entries didn't stop appearing after I removed my usb drive, but stopped after reboot.

Means that your router lost access to the USB disk. Physically removing the disk won't remove it logically (software-wise), that's why a reboot was required.