What's new

Beta Asuswrt-Merlin 386.1 Beta is now available

  • SNBForums Code of Conduct

    SNBForums is a community for everyone, no matter what their level of experience.

    Please be tolerant and patient of others, especially newcomers. We are all here to share and learn!

    The rules are simple: Be patient, be nice, be helpful or be gone!

Status
Not open for further replies.
I think ‘off’ is the default.
Switching it on has made no changes here, except a considerable CPU temperature drop.
Hours later, about 15° lower!
Yes, I have been running an AX88U with cpuwait enabled for around a year.
When it first appeared (long ago), Merlin mentioned it was added by Asus due to some errata concerning AX88U V1.0 hardware.
My take is that YourMiliageMyVery - it really depends on your hardware.
I decided to try with it on - if I experienced any odd issues or router reboots, I would set it back to off.
My luck of the HW draw is that it doesn't seem the have any issues - other then reducing CPU temps.
 
Edit: I figured out my issue. I used the steps from L&LD to fully reset the router which is letting me downgrade the firmware

I'm having issues trying to revert back to a previous firmware on my AC-3100. My 2.4 is acting up, and I want to revert back to 384.19 (or even sock ASUS) to see if that fixes it, and I keep getting the error:

Invalid Firmware Upload
To comply with regulatory amendments, we have modified our certification rule to ensure better firmware quality. This version is not compatible with all previously released ASUS firmware and uncertified third party firmware. Please check our official websites for the certified firmware.
 
Last edited:
Finally broke down and flashed beta3... all is well so far, thanks RMerlin!

Other than I dont use guest networks or anything, yet I see dnsmasq allocating 192.168.101.x and 192.168.102.x networks now :confused:
Capture.PNG

And I had some interesting discoveries regarding temps Ill make a post about (edit, up... here)
 
Last edited:
Here's the colour code:

  • Red: The wire in red is used to pass positively-charged electric power. It has 5V direct current electricity.
  • Black: The wire in black is the ground wire.
  • White: The wire in white is used for positive data (D+).
  • Green: The wire in green is used for negative data (D-).
Sometimes you encounter these colours:

  • Orange: The wire in orange is used to pass positively-charged electric power. It has 5V direct current electricity.
  • White: The wire in white is the ground wire.
  • Green: The wire in green is used for positive data (D+).
  • Blue: The wire in blue is used for negative data (D-).
I cut of a few inches of an old mouse cable, cause it's flexible and thin.
Read somewhere it's better to connect to the USB2.0 (black) port and not the USB3.0 (blue) port.
How about a USB cooling fan with a USB connector? Already wired and ready to go. Here is an example: 2-Pack 40 x 40mm USB Cooling Fan
Just watch the current draw (this one is 200mA). There are limits to the current draw on USB ports: USB Power

Note a fan speed switch will not reduce the current load (just adds resistors in series to reduce voltage to fan keeping total current consumption the same) but would slightly reduce fan noise at lower speeds.

How you place/attach/tape the fan will be important to get air flow inside the case. Otherwise you are only cooling the case which will do little to cool the internal CPU. I have a RT-AC86U and I do not see an easy way to open the case (If the back came off, a fan could blow directly onto the circuit board / heat sink / CPU).

Take care if you do choose to tape the fan over the router case vent holes. If the fan stops. for any reason, the normal passive cooling of the router will be affected and the CPU temperature will almost certainly shoot up a lot.
 
Last edited:
I found a short explanation for CPU wait states and why they are used. I thought I would share: Wait state - Energy conservation.

By removing wait states (idle states) the device (router), has less latency for the CPU receiving/sending data between itself and peripherals at the expense of continuous tight-loop polling checking when the read/write is complete. The tight-loop continuous polling makes the router faster but the CPU hotter.

I suspect ASUS must have added enough new code that the router does not hit top performance in certain use cases. But I would rather have a cooler CPU. The hotter the CPU, the sooner it will fail (as would any hot IC).
 
Last edited:
I thought that the AX88Us were supposed to have the cpu wait set to off?

All these temperature post I find amusing considering much of this was discussed years ago and everyone was saying ignore the temps its normal.

Anyways it depends on the revision for the AX88U version 1.0 CPU wait is off by default. On Version 1.1 cpu wait is on by default.
 
I have a RT-AC86U and I do not see an easy way to open the case (If the back came off, a fan could blow directly onto the circuit board / heat sink / CPU).
There is a vid on Youtube on how to do it.

40mm is a little big. I used a 30 mm: https://www.amazon.com/gp/product/B072FW3DDQ/?tag=snbforums-20, the first fan I used on an 87U. The pins would work on the 86U jumpers. Mine came with self tapping screws too, so I could use them on the diagonals without opening the case or taping. 7/64 bit, I think.
 
Clean reset after BETA4 - put all settings in by hand. No JFFS backups used as reset. No addins. Not lying here, as I know NOT doing this can cause issues that aren't issues with the firmware.

For me, QOS does not seem to work (saw this in previous releases too) . Two new AX86U routers. One at one house, one at the other. Both reset and BETA 4, same config near enough except for DHCP reservations and MAC clone address.

When I start streaming media, no traffic on either router is observed (on the previous AC86U I would see it) in QOS pie charts OR at a device level (being server streaming). I don't see upload or download data for device or QOS category increasing and the GUI graphics dials for upload and download stay at zero.

I have an upstream of 38mbps with no ISP saturation issues - previously worked fine for 2 years. With an AX86U on the downstream receiving end I get stop-starts to the media even though the media files have always played fine. I swap out the AX86U at the receiving end back to the ISP's cable router (Virgin Media, UK), and its fine again!

I've previously used two AC86U on last Merlin production release without any issue - one for sending upstream, one for receiving downstream with pretty much the same configuration and both wired with CAT8. Both sender router and receiving router are in same town, on same ISP.

So for me, QOS and Traffic Analysis seems to be broken; at least on AX86U?
All of the Trend Micro components will be non-functional in any firmware you compile yourself.

I can confirm that Adaptive QoS under Asus Stock firmware [Version 3.0.0.4.386.41535] released on 2020/12/30 with manual settings works straight out the box - and is the first stock supplied QoS which has given me a triple A+ rating with settings up and down at 95% of full speed.

I have no doubt that when RMerlin gets to release an updated beta for the RT-AX86U ... based on the above stock firmware it will do the same. NB - he has already indicated earlier in this thread that beta3 for the AX86U won't achieve much given that he did not have a GPL for that model at the time he was compiling b3 for the current model set.
 
believe it or not, I probably have everything I need without having to feed The Bezos...and if I'm mistaken about the voltage of the fan I believe I have, I'd prefer to get off my butt and head down to support the local electronics supplier/maker shop so I can keep enjoying the luxury of a 5 minute drive and not having to wait overnight.
 
I can confirm that Adaptive QoS under Asus Stock firmware [Version 3.0.0.4.386.41535] released on 2020/12/30 with manual settings works straight out the box - and is the first stock supplied QoS which has given me a triple A+ rating with settings up and down at 95% of full speed.

I have no doubt that when RMerlin gets to release an updated beta for the RT-AX86U ... based on the above stock firmware it will do the same. NB - he has already indicated earlier in this thread that beta3 for the AX86U won't achieve much given that he did not have a GPL for that model at the time he was compiling b3 for the current model set.
Me too. My full speed is 840Mbps something. I set Adaptive QOS at 825Mbps and the bufferbloat meter doesn't turn to red. Now, I'm waiting for new Merlin Beta that has working bandwith monitor for AX86U.

DSC_04921.jpg
 
Fun times today... Went from Merlin 386 Beta 3 to Asus 386 RC10 and back. Interesting journey, learned a few things.

To go into rescue mode, I need to disable the HyperV Virtual Adapter/Switch and any other adapter. It just won't work if those are enabled. Having tried rescue mode so many time that I lost count only to have it work the moment I disabled all the other adapters was priceless. Why rescue mode, you need rescue mode to go from Asus 386.rc10 to anything else on the AX88, on the AC5300's it asn't a problem. Went to Asus 386 rc10 to see how some of the errors I was seeing in the logs and the temperature anomalies would manifest, instead my Internet bandwidth over the LAN tanked (from 930/930 to 120/120 on the PC and 98 to 32 on the AndroidTV boxes - Speedtest on Router was consistent as before but over wired or wifi, it was way off). A reported issue a few are seeing too.

Going back to Beta3, had a heck of a time with Aimesh. Turns out you need to disable Ethernet Back haul first, to get the nodes back on. Once on, then enable Ethernet Backhaul assuming your Aimesh nodes have wired uplinks.. Added one node, would stay on for a few then drop, couldn't add the other until I disabled Ethernet Backhaul, and it took me awhile to figure that out too.

Having do a hard facory reset, formatted to JFFS and USB, and starting completly from scratch on the AX88 and both AC5300's a few thing have changed, for the good. Temps on the CPU stay within 78c-80c. The odd spikes on the 2.4mhz, randomly from 40c-49c to almost always 76c are gone. Instead of a heatbeat, my temperature graph shows a flatline. Memory consumption much more manageable and consistent. Same AMTM addons (Entware, Skynet, conmon, ntpMerlin, scMerlin, scribe/uiscribe) but much more free memory and staying free. Now about ~390MB free, before I was dipping into the swap file regularly.

Also found somwhere how to check, then set the log level from 6 to 5, commit it to NVRAM and reboot to quiet the logs down a bit. To my surprise, the checking of the log level (nvram get log_level) returned a 7, not 6. I set to 5 just the same. Dare I say ignorance is bliss. I'm now seeing items in the log that weren't there before (dnsmasq-dhcp, solicitation, offer, request, discover, etc.), not sure if the setting of the log level or the hard factory reset is responsible for that.

Also the errors related to DHCP and devices (wired or wireless, on the router, or one of the nodes) not being able to renew their address, gone now. The incorrect labling of of devices/addresses in the client list of devices static, manual, dhcp, also fixed, issue gone.

All in, long journey, but in a much better place now.
Never underestimating the power of a hard factory reset ever again...
 
believe it or not, I probably have everything I need without having to feed The Bezos...and if I'm mistaken about the voltage of the fan I believe I have, I'd prefer to get off my butt and head down to support the local electronics supplier/maker shop so I can keep enjoying the luxury of a 5 minute drive and not having to wait overnight.
I know there's a ton of possible fans and even complete sets out there, but i found it fun to do it myself with the stuff i had laying around the house :)
 
Note a fan speed switch will not reduce the current load (just adds resistors in series to reduce voltage to fan keeping total current consumption the same) but would slightly reduce fan noise at lower speeds.
What do you mean?
Adding resistance to a circuit with a fixed voltage source will definitely reduce the current. Ohm's Law I=V/R.
Reducing the current will also reduce the power of the circuit. P=V x I
 
All these temperature post I find amusing considering much of this was discussed years ago and everyone was saying ignore the temps its normal.
It's normal up to the point that it stops being normal. In my instance the temperatures were pushed beyond the thermal shutdown limit, which is definitely not the intent.

Mine is sitting at 85 at the moment on beta2, but will go up to 90ish when the ambient temperature rises. On beta 3, it was getting to 100 degrees and halting a core. Even re-enabling the CPU wait state on beta 3 caused the core temp to rise into the danger zone eventually, so not sure if beta3 was simply doing more work in addition to that change.

I may need to crack the router open at some point to make sure the heatsinks have not gummed up or something.
 
It's normal up to the point that it stops being normal. In my instance the temperatures were pushed beyond the thermal shutdown limit, which is definitely not the intent.

Mine is sitting at 85 at the moment on beta2, but will go up to 90ish when the ambient temperature rises. On beta 3, it was getting to 100 degrees and halting a core. Even re-enabling the CPU wait state on beta 3 caused the core temp to rise into the danger zone eventually, so not sure if beta3 was simply doing more work in addition to that change.

I may need to crack the router open at some point to make sure the heatsinks have not gummed up or something.
Asus nor any other manufacturer wants their products to last a lifetime.
They're meant to drop dead just outside the warranty period, so you can buy another. :rolleyes:
 
It's normal up to the point that it stops being normal. In my instance the temperatures were pushed beyond the thermal shutdown limit, which is definitely not the intent.

Mine is sitting at 85 at the moment on beta2, but will go up to 90ish when the ambient temperature rises. On beta 3, it was getting to 100 degrees and halting a core. Even re-enabling the CPU wait state on beta 3 caused the core temp to rise into the danger zone eventually, so not sure if beta3 was simply doing more work in addition to that change.

I may need to crack the router open at some point to make sure the heatsinks have not gummed up or something.
Have you seen this thread yet? It might interest you.
 
Status
Not open for further replies.

Latest threads

Sign Up For SNBForums Daily Digest

Get an update of what's new every day delivered to your mailbox. Sign up here!
Top