ROG Strix G513QM 2021 Laptop freezes with some insights on the issue and possible fix

[Darkosphere]

Hello everyone,

To start I would have been able to reply directly on this thread but since it is closed I needed to make a new one.

Informations about the computer:

ASUS ROG STRIX-G15-G513QM-HF220T

SN: M8NRKD02264134G

OS: W10 Family Edition (the one that cames preinstallated on the computer)

Hardware state:

SMART readings on the NVME SSD shows it is in good condition.

Performing a 4 Pass test on the RAM with MemTest86 showed no errors and not a single bit of error.

The touchpad of this computer has a small defect that makes it's right click to not work.

The fans have been cleaned from dust after seeing a 90°C on the CPU in BIOS after only 2 minutes from boot after that cleaning for about 15 min on the BIOS the CPU reached a 87°C max before going down to 62°C when the fans started, then the CPU temp moved between those two values. 

Description of the problems encountered:

- The computer will fail repetitively on startup blocking on the logo screen or sometimes on the recuperation screen making the access to Windows login screen extremely difficult.

- When the Windows login screen is reached  the computer can then freeze on that screen or a bit later when logged in a user session, it could happen with a light load or no load at all. These freezes happens randomly, when it does happen the computer is totally unresponsive and needs to be forced to shut down by pressing the power button in order to regain a hand on it. Looking in the Windows Event Observer there are no specific errors associated with the freezes except for critical errors linked to the use of the power button.

freezes definition: in this case the freezes are not small freezes but totally block the system when happening.

Explanations:

While it may seems that the problems encountered comes from the same cause it is not as the loading issues could be fixed from flashing the BIOS, and the freezes seems to be caused by unreliable performance profiles thus in some ways it is related to temperatures.

What fixed the problems for some time until a bad manipulation reproduced the situation:

Warning: It is advised to save your files and data before proceeding with this otherwise they will be lost. 

1. Reset the machine to factory settings from the recuperation environment Asus WinRE
2. Proceed with the finalization of the Windows installation, if freezes are encountered you will have to retry untill success. (I didn't connected the computer to internet during the procedure and proceeded with the limited installation to create a local account)
3. Flash the BIOS

At this point and after some fails on startup the system finally stabilized and the fails on startup were mostly fixed it could still happen from time to time but didn't get to the same extremities as before. Now I was able to log into a user session nearly every time but the freezes were still occuring.

I started to open the task manager when arriving on the desktop to have an indication of the system uptime between the freezes and I got ratings from 2 minutes to ~ 11 minutes. Then reading about the fans RPM on this thread I clicked on the Armoury Crate icon that was in the start menu it installed and started the application, navigating in it I managed to select the performance profile "performance" letting the machine run I reached 42 minutes of uptime without any freezes when I decided to restart the machine to reset that uptime count to see if the freezes were gone. That was a bad idea following the bad idea of clicking on the Armoury Crate icon...

Following the installation of Armoury Crate the system forced a functionality update during the reboot of the machine that brought it back to the initial state of problems (I guess this update messed with the BIOS) . So I repeated the previous steps again but this time I flashed the BIOS like 4-5 times because it looked like the start up issue would not resolve. From my attempts I could say that flashing the BIOS do not resolve the issue immediately you will have to proceed through a bit less of 10 start up fails before actually seeing improvements of the start up.

After getting the system up again, the next day I started to test the uptime with the "performance" profile, with this profile the system was able to run for 5 Hours+ without freezing the system now looked to be stable. I then decided to play a bit with the profiles to see what would happen this time I didn't clicked on the Armoury Crate icon but used the fan button next to the rog button on top of the keyboard to switch between profiles. As expected when using the "silent" profile the system started to freeze again but when on the "performance" profile there was no freezes.

I then decided to try an alternative to Armoury Crate : https://github.com/seerge/g-helper

While it is touted on the app page that the fans profiles it uses are the ones that are defined in the BIOS, it seems there are some variations. From the original profiles: "silent", "performance", "turbo", You now have: "silent", "balanced"  and "turbo". The "balanced" profile from g-helper do not match the original "performance" profile so I was getting freezes again. I tried to modify the fans curves of the g-helper "balanced" profile to have the minimum set around 2000-2200 RPM with no success the system was still freezing so I repeated another time my procedure.

After getting the system running one more time I had another occasion to break it again after pressing the rog button out of curiosity it started the installation of Armoury Crate, I then closed immediately the installation windows and killed the associated process but it was too late on reboot the functionnality update that broke the start up previously was started. So I had to repeat the same procedure one more time. 

FIX Resume :

For the failures at start up :

Flash the BIOS, it may not resolve the issue instantly though as you may have to proceed throught a bit less of 10 failed start up until the system starts to stabilize. (I'm unsure of the effect of resetting the system to factory setting here I needed to do it in first place because the Windows installation had some issues).

For the freezes :

Use the original "performance" profile. Below is a little table for a quick overview.

Profile Name	"Silent"	"Performance"	"Turbo"
Freezes status	unreliable	stable	untested

You could move between those profiles using the "fan" button that is next to the "rog" button on top of the keyboard so there is no need to install any software of any sort to do this.

Conclusion :

I hope that this will help other people to resolve the issues they may have with their laptops. It could be interesting to see if what I have described above can be reproduced on other machines. 

I haven't tried to flash the BIOS after the functionnality update that follows the Armoury Crate installation without resetting the OS to factory settings.

Also I could be interested in how to prevent the rog button or any other to start the installation of Armoury Crate as it is pretty annoying to break havoc on a system for a mistake as small as pressing a key by inadvertance.

[ElectroStingz]

Hello,

Does it still crash if you run the laptop on battery only?

If no modifications to the laptop power profiles this should run the CPU without the boost clocks enabled so if it is stable on battery only there is a possibility the issues are with the CPU being unstable when it tries to boost the clock frequencies.

[Darkosphere]

I tried a startup on battery today when logged into the windows session the profile switched automatically to silent mode, looking at CPU clocks it was well under its 3.30 GHz base it was running at around 2.30 GHz.

In this state I've beaten the previous silent mode uptime (with the charger connected) and got 24 minutes before crash I then restarted had to do it a certain amount of times before to log again into a windows session then it crashed after 2 min.

Looks more like the CPU is unstable when underclocked or the silent mode is not able to handle temperatures correctly.

I also tried the "normal usage" power mode that shows when going into the windows power options but can't remember if it was on battery nonetheless it crashed.

So silent mode is crashing with and without the battery. Performance mode is stable without the battery but I haven't tested it yet on the battery.

I also have found a way to not have the rog button try to install armoury crate on press:
Remove or rename C:\Windows\System32\ASUSACCI\ArmouryCrateKeyControl.exe
From the start menu of windows in the recently added program (middle part of the start menu) under the letter "A" right click on "Armoury Crate" and select the option uninstall. 
Note that doing so changes the name of the "performance" profile to "balanced" profile but this one works similarly to the original "performance" profile and no crash occurs when using it.

For now I'm gonna check the stability of this "balanced" profile when using the battery lets see if we can do better than the 24 minutes of the "silent" profile, I will also check the CPU clocks in this case.

[ElectroStingz]

That's a shame, is it possible to get some temperature readings whilst in windows as 87 Deg C is also too high for a BIOS reading.

https://www.cpuid.com/downloads/hwmonitor/hwmonitor_1.57.zip

Hardware monitor will record min, max and current temperatures. If the CPU is idling high in windows or peaking over 90 it will be unstable on any load you apply, and if its on the absolute limit to the point that airflow is having that kind of impact then the other profiles with higher Fan RPM will make a difference.

Within hardware monitor press on the + sign to expand all the sensor fields and post some screenshots when the laptop is idle in performance mode (as you mention this one being stable). (show all the temps, GPU, CPU, Motherboard...)

[Darkosphere]

I got you by speed on that one yesterday I already got a portable version of hwmonitor as I thought it will be easier to have more readings to investigate on temperatures.

I have also installed OCCT and already performed a successfull stress test with the CPU and the RAM. During this test the CPU peaked at 100°C for some extended periods of time but it wasn't that high for the total duration of the test, internal GPU was peeking at 77°C. On the "performance" profile with the charger adaptator connected of course.

Below are HWMonitor combined screenshots  :

Min/Max readings not reset

 Only programs running are OperaGX and HWMonitor, the min/max readings have not been reset so the highest value comes from machine start up.

About the failing startups I start to see a pattern here :
When I shut down the computer normally from the start menu option it will restart without failing most of the time, but if I choose the restart option it will enter in a failing starting loop that may resolve after a certain amount of forced shut downs or restart from the recuperation environment. During the failing loop the computer will either block on the logo screen or on the loading of the recuperation environment but this one happens less often. I precise that FastBoot is enabled.

For the high temps on the BIOS it seems to me that when the laptop is starting the fans are not working directly but starts after some time. Maybe there is another profile of performance with its fans settings that is in play here but there is no way to control it or have a sight on it. Maybe the laptop starts on a silent mode that will later be switched to another profile that may explain the notification of the performance mode like it was selected when arriving on a user session.

Due to the instability of that profile (silent) it could explain the failed start ups. I haven't encountered a single issue while flashing the BIOS, resetting to factory settings, or restorating windows but during all of these operations the fans tend to scream pretty loud. I also have not encountered any issue of stability while in the BIOS.
Issues occurs at start up on the logo screen mostly from a reboot and some times on the loading of the recuperation environment,  otherwise it will happen on the windows login screen, finalization process of a windows reinstall and when logged into a user session.