AC86U overheats with fw 386

[Poncik]

I assume the CPU is getting hotter because it is doing more processing, has there been any definitive explanation of what extra processing is taking place, especially if Trend Micro is disabled?

I disabled everything to bring the temperature down and CPU was idling most of the time but that didn't help. It only improved after I installed a firmware without the temperature bug and configured the router from scratch.

 

[MarkyPancake]

I didn't get as far as putting 386.x on my AC86U after I noticed my AC68U node was running hotter than usual with 386.1 and Asus' latest firmware.

For comparison, my AC86U is still on 384.19 and the temps look like this:

• 2.4 GHz: 51°C
• 5 GHz: 57°C
• CPU: 67°C

 

[dh99999]

Yeah, 41994 dropped mine from low 90s to low 80s... at 79 now. 15% below 100C is comfortable operation

Could you please let me know what version were you running before 41994?

I was running Merlin 386.1_2 (95 degrees celcius) and had to revert to Merlin 384.19 (83 degrees celcius) -> these temps are for an ambient of around 26 degrees plus high humidity (90%). Resetting/etc did not resolve the issue.

Merlin 386.1_2 is based on 41700, so I am trying to understand if the changes you are seeing in 41994 are also supposed to be in 41700 (and hence Merlin 386.1_2).

If they are, then I'll probably revert to stock 41994 for a while to test it to see if its something Merlin specific.

 

[Poncik]

Could you please let me know what version were you running before 41994?

I was running Merlin 386.1_2 (95 degrees celcius) and had to revert to Merlin 384.19 (83 degrees celcius) -> these temps are for an ambient of around 26 degrees plus high humidity (90%). Resetting/etc did not resolve the issue.

Merlin 386.1_2 is based on 41700, so I am trying to understand if the changes you are seeing in 41994 are also supposed to be in 41700 (and hence Merlin 386.1_2).

If they are, then I'll probably revert to stock 41994 for a while to test it to see if its something Merlin specific.

I don't think it is Merlin specific. I was using "RC2-11 41994" before installing "Merlin 386.1_2" and both were around ~92-95c. In both cases I configured the router from scratch to get rid of the temperature bug. Now with "Merlin 386.1_2" and a fan, the temp is around 60c.

 

[OzarkEdge]

Could you please let me know what version were you running before 41994?

I was running Merlin 386.1_2 (95 degrees celcius) and had to revert to Merlin 384.19 (83 degrees celcius) -> these temps are for an ambient of around 26 degrees plus high humidity (90%). Resetting/etc did not resolve the issue.

Merlin 386.1_2 is based on 41700, so I am trying to understand if the changes you are seeing in 41994 are also supposed to be in 41700 (and hence Merlin 386.1_2).

If they are, then I'll probably revert to stock 41994 for a while to test it to see if its something Merlin specific.

I was running 41634 before 41994.

41700 was suppose to have a fix in it for the CPU temp issue, according to a post here by Merlin, but he did not say what the fix was.

OE

 

[ATLga]

@RMerlin says in this post that it was fixed in 41700 and what the fix was:

Release - Asuswrt-Merlin 386.1 is now available

upgraded from b5 on ax88. no issues. thanks Merlin!!!

www.snbforums.com

"The issue has been fixed in 386_41700, the code that disabled CPU wait has been removed. I actually see the code.

Give it a rest guys, 80C for that CPU is perfectly normal. That CPU has a throttle limit of 100C. Enough already."

 

[OzarkEdge]

@RMerlin says in this post that it was fixed in 41700 and what the fix was:

Release - Asuswrt-Merlin 386.1 is now available

upgraded from b5 on ax88. no issues. thanks Merlin!!!

www.snbforums.com

"The issue has been fixed in 386_41700, the code that disabled CPU wait has been removed. I actually see the code.

Give it a rest guys, 80C for that CPU is perfectly normal. That CPU has a throttle limit of 100C. Enough already."

For a CPU that throttles at 100C, it would be nice to operate about 15% below that in the mid 80s. Operating at 95C in an ambient of 26C/78F doesn't leave much headroom. I'd blame the heatsink except downgrading the firmware drops the CPU to low 80s.

I'd try the firmware with CPU fix again and this time use the SSH pwr show command to see if CPU wait is enabled like it should be.

OE

 

[dh99999]

"The issue has been fixed in 386_41700, the code that disabled CPU wait has been removed. I actually see the code.

Give it a rest guys, 80C for that CPU is perfectly normal. That CPU has a throttle limit of 100C. Enough already."

Yeah I saw that statement from Merlin as well - unfortunately my CPU is 95 degrees, not 80, so something is not right.

I haven't seen anyone report issues with routers other than the AC86U, so maybe it is localised to that hardware, or maybe just mine

OzarkEdge:
I'd try the firmware with CPU fix again and this time use the SSH pwr show command to see if CPU wait is enabled like it should be.

Thanks OE - I did try that before reverting - CPU wait was enabled.

Poncik:
I don't think it is Merlin specific. I was using "RC2-11 41994" before installing "Merlin 386.1_2" and both were around ~92-95c. In both cases I configured the router from scratch to get rid of the temperature bug. Now with "Merlin 386.1_2" and a fan, the temp is around 60c.

Thanks - I'm going to give stock 41994 a go, and if that doesn't work I'll report it to ASUS and see what they say.

 

[Centrifuge]

In both cases I configured the router from scratch to get rid of the temperature bug. Now with "Merlin 386.1_2" and a fan, the temp is around 60c.

What would your temps be without the fan on that version?

For comparison AC86U with 384.18 merlin:

Functional Block Status
CPU Wait ENABLED
Ethernet Auto Power Down ENABLED
Energy Efficient Ethernet ENABLED
Switch Deep Green Mode ENABLED (status deactivated)

• 2.4 GHz: 56°C
• 5 GHz: 61°C
• CPU: 80°C

Ambient temp is 31°C. Router is in AP mode at this point.

 

[Poncik]

It was around 95c in 25c ambient temp without the fan.

 

[mromero]

I don't think it is Merlin specific. I was using "RC2-11 41994" before installing "Merlin 386.1_2" and both were around ~92-95c. In both cases I configured the router from scratch to get rid of the temperature bug. Now with "Merlin 386.1_2" and a fan, the temp is around 60c.

In my experience it is NOT a Merlin issue. It is an Asus issue. Re-enabling the CPU wait took the AC86U down a few degrees. Not good enough. I took a fan out of my pile of good junk and got the temp down. But I do not want a fan. Happy and fanless back on Merlin 384.19 47 °C - 53 °C - 75 °C (ambient temp 28 °C) Have another router in front with latest UNBOUND so the Asus DNSMASQ and other bugs not an issue for my network.

 

[jpthsd]

This is a known issue with the 386 firmware - both the genuine Asus, and Merlin's implementation. It has been widely discussed.

The solution is to purchase a fan or use the previous release, such as Merlin 384.19, or await for a fix from Asus (CPU wait is NOT the problem).

The consensus appears to be that the 386 firmware is more for the AX models.

The 384.19 works very well on the Asus AC-86U router - no overheating. Placing a business class router in front of the Asus will protect your network from unpatched Asus vulnerabilities etc. Like you, I am looking at other brands for my next purchases as they are getting better.

alright,,,this makes me go to my AC86U AiMesh node and reconnect my 2inch fan that's already attached to the AC86U...,,just to make it cool to prolong the electronic devices!

 

[L&LD]

The routers will be obsolete long before the heat gets them. Really.

 

[webdriver]

It worth noticing that the cpu temp reading is changing immediately after issuing cmd to turn on/off "cpu wait on|off" through command line.

I am kind suspecting how this could happen ... isn't that should the device wait a little bit to drop/rise temp? interesting...

---- it looks exact 5 C drop/rise with cpu wait flag ---
---- it looks eee flag doesn't affect the temp reading ---

 

[maxbraketorque]

My impression is that there are at least two things contributing to higher temperatures in the AC86U. One was having CPU Wait disabled. Its unclear what the other item is, but based on what I've seen in Merlin's posts, it sounds like the other contributor is something that ASUS is doing to address a perceived hardware issue with the AC86U. CPU Wait has been reenabled, but this second contributor may be around permanently, so if you want to run a newer firmware and the router temperature is higher than you like, attach a fan to it and/or reapply the heatsink to the CPU. Either one will drop CPU temperature by 20C.

 

[Centrifuge]

two things contributing to higher temperatures in the AC86U. One was having CPU Wait disabled.

CPU wait is a feature to triage and manage processor demands, it has a net effect of slowing down processing which is what generates heat, so what makes me curious is how turning CPU wait off would paradoxically lower temps. Throttling the CPU would make it run cooler of which CPU wait would to tend to do that.

 

[OzarkEdge]

CPU wait is a feature to triage and manage processor demands, it has a net effect of slowing down processing which is what generates heat, so what makes me curious is how turning CPU wait off would paradoxically lower temps. Throttling the CPU would make it run cooler of which CPU wait would to tend to do that.

As I see it, enabling a CPU wait state may slow down processing and lower CPU temperature. I can't imagine that slowing down processing would generate more heat, as you put it.

OE

 

[jpthsd]

As I see it, enabling a CPU wait state may slow down processing and lower CPU temperature. I can't imagine that slowing down processing would generate more heat, as you put it.

OE

Exactly, hope the new firmware release (currently in beta) would fix the issue!!

 

[bbunge]

9.0.0.4.386_41994-g769f84f
79C in a 20C room - no fan

 

[mromero]

Exactly, hope the new firmware release (currently in beta) would fix the issue!!

From the official Asus site, Version 9.0.0.4.386.41994 beta is "Please be noted this is a quick fix beta version for DNSmasq vulnerabilities." No mention whatsoever about the overheating issue of the RT-AC86U router.