Investigating Laptop Crashing

[Eluvade]

Hi! I could really use the assistance of someone proficient in troubleshooting both hardware and software points of failure.

My laptop has been powering off suddenly, with no warning or error message displayed for a while now, and I've been, so far unsuccessfully, trying to figure out the cause.

I'm attaching a ThrottleStop log from the other day during which 3 crashes happened. Specifically, they happened while running Unreal Engine 5 and a random video game. Two of those happened under extensive load, with temperatures at ~90°C while throttling CPU clock speed. The third one, however, happened while practically idling, with temps sitting at the low-end and the CPU usage being extremely low.

No matter the ThrottleStop configuration; TurboBoost enabled or disabled, core and cache undervolted or not, bd prochot on or off, whenever I run anything more demanding on the laptop, it will crash eventually. However, I believe I might have found an outlier. This only happens while the laptop is plugged in. I don't think I was able to crash the laptop yet while it's running on battery alone, unplugged. I will continue testing this, in the meantime, does anyone have any advice or idea about what might be the likeliest culprit that makes me unable to work or play video games?

Laptop's specifications, if they matter to anyone:
Geforce GTX 1070
Intel(R) Core(TM) i7-8750H CPU @ 2.20GHz 2.21 GHz
DDR4-2666 16Gb x2 (from 2 different brands, but I've matched the CAS latency and never had issues with it before)
2x 512Gb SSD's: NVMe Samsung SSD 970 & THNSN5512GPU7 TOSHIBA both in healthy condition
64-bit operating system, x64-based processor
Dual Booting Windows 11 Pro, Version 22H2 and Arch Linux

 

[unclewebb]

core and cache undervolted or not

ThrottleStop does not work correctly in Windows 11 until you disable Virtualization Based Security including core isolation memory integrity. If maximum security is important to you then leave this stuff enabled. If playing games is important to you, I would recommend turning this stuff off. Follow the Disable VBS link in my signature. This link is also in the ReadMe file that is included with the ThrottleStop download.

If you decide to disable VBS, reboot and delete the ThrottleStop.INI configuration file. Post an updated FIVR screenshot after you do that.

Your CPU is getting up to 90°C when turbo boost is disabled. That tells me that you have a cooling problem. When was the last time you opened up your laptop and cleaned it? This is normal maintenance for any laptop with cooling issues. You might also have to replace the thermal paste if blowing the dust out does not solve your problem.

DDR4-2666 16Gb x2 (from 2 different brands, but I've matched the CAS latency and never had issues with it before)

I find that most Intel CPUs are fairly stable throughout their life. A memory module that is failing is a more likely source of random crashes. I would try testing one memory module at a time. Run something simple like the TS Bench test. Does this test report any errors? Set the size to 7680M so it can test a bigger percentage of your memory. No need to run the TS Bench test all day. Even 5 minutes without crashing or reporting any errors is better than nothing.

There are lots of programs available when testing memory for errors. I have not had any issues for years so I do not know what program is the best tool for the job. I prefer programs that run before Windows starts up if possible.

When you say you matched CAS latency, did you use a program like Thaiphoon Burner to physically change the memory timings? Explain this further. Different memory modules will have a lot of different sub timings. There is a reason why running two different memory modules is not recommended.

 

[Eluvade]

Thank you for taking the time to review my case and suggest methods of approaching the situation, unclewebb. Also, thank you for creating ThrottleStop, it's an amazing piece of software!

ThrottleStop does not work correctly in Windows 11 until you disable Virtualization Based Security including core isolation memory integrity. If maximum security is important to you then leave this stuff enabled. If playing games is important to you, I would recommend turning this stuff off. Follow the Disable VBS link in my signature. This link is also in the ReadMe file that is included with the ThrottleStop download.

If you decide to disable VBS, reboot and delete the ThrottleStop.INI configuration file. Post an updated FIVR screenshot after you do that.

I've read your guide but VBS was already disabled. I suppose that's because the laptop was purchased in 2018 and I've upgraded from Windows 10 at some point since.

Your CPU is getting up to 90°C when turbo boost is disabled. That tells me that you have a cooling problem. When was the last time you opened up your laptop and cleaned it? This is normal maintenance for any laptop with cooling issues. You might also have to replace the thermal paste if blowing the dust out does not solve your problem.

Apologies for the confusion, the CPU doesn't go over 75°C with the turbo boost disabled. My ThrottleStop configuration was changed several times throughout those logs. I maintain my laptop regularly. As I've been trying to diagnose the cause for these crashes, I've recently disassembled it, cleaned out the fans, reapplied the lubricant and replaced the thermal paste on the CPU and GPU. Then, when the crashing persisted, I second guessed my thermal paste application and I redid it more carefully, but that wasn't the issue.

I find that most Intel CPUs are fairly stable throughout their life. A memory module that is failing is a more likely source of random crashes. I would try testing one memory module at a time. Run something simple like the TS Bench test. Does this test report any errors? Set the size to 7680M so it can test a bigger percentage of your memory. No need to run the TS Bench test all day. Even 5 minutes without crashing or reporting any errors is better than nothing.

That's been my experience too. Testing the modules one at a time is a good suggestion. I'll do this as soon as I'm able. I'm currently on a work-related trip and I don't have my toolkit with me to open up the laptop.

The TS Bench does not report any error. Please keep in mind some of these tests were done on battery or with turbo boost disabled which adds to the time necessary to complete the test, but the key takeaway is that the laptop didn't crash once during these benchmarks.

There are lots of programs available when testing memory for errors. I have not had any issues for years so I do not know what program is the best tool for the job. I prefer programs that run before Windows starts up if possible.

I'll do some research for this once I'm able to test my RAM sticks one at a time. Speaking of programs that run before Windows start up... I've followed along a guide and set up a task in the Task Scheduler that runs ThrottleStop upon Windows logon. While I can see ThrottleStop running in the Task Manager, when it starts up this way, it doesn't show the temperature/its icon in the tray, on the taskbar. In order to tweak the settings or open the user interface, I have to first end the task and then re-run the executable file. Do you have any solutions for that?

When you say you matched CAS latency, did you use a program like Thaiphoon Burner to physically change the memory timings? Explain this further. Different memory modules will have a lot of different sub timings. There is a reason why running two different memory modules is not recommended.

I don't have any experience with programs such as Taiphoon Burner. I simply meant that when I was looking for laptop parts, I made sure the two would match. I've had the laptop for 5 years, and I've never had any issues up until a few months ago.

I'm really starting to suspect the AC adapter at this point. The first time the laptop crashed the way it does daily now was when it was plugged into the same socket in the wall that my old adapter gave out while being plugged into. I rarely use that particular socket, which is why I distinctly remember it. Perhaps I have a faulty wall socket.
I've been running the 7680M sized TS Bench for 18 minutes and counting as I'm writing this reply. The processor is happily chirping at 90-91°C, throttling down to 2.8Ghz from 3.9Ghz. But the laptop won't crash. This makes me think it's not the thermals. If I start up something that requires the dedicated GPU to turn on, in turn using more wattage, the laptop will crash. But since it runs just fine on battery, unplugged, I'm assuming the GPU is all well and healthy. Is there any way to test the AC adapter before I order a replacement?

I'm attaching one more screenshot now that the TS Bench has finalized.

I'm sorry for the long post, thank you for baring with me. I truly appreciate your effort!

 

[Eluvade]

Speaking of programs that run before Windows start up... I've followed along a guide and set up a task in the Task Scheduler that runs ThrottleStop upon Windows logon. While I can see ThrottleStop running in the Task Manager, when it starts up this way, it doesn't show the temperature/its icon in the tray, on the taskbar. In order to tweak the settings or open the user interface, I have to first end the task and then re-run the executable file. Do you have any solutions for that?

After taking a look at this thread, I too found multiple ThrottleStop icons in the "Other Icons" of the "Taskbar Settings":

Since I'm not the only one with this issue, and it's my understanding you've developed the software on Windows 10 and it hasn't yet been extensively tested on Windows 11, please don't waste your breath addressing this issue to me personally. I wanted to take a look at the source code to help resolve the bug, but I quickly realized the code is proprietary. Let me know if I can help anyhow though!

 

[Eluvade]

Update: It turns out it does sometimes crash while being on battery and unplugged. I was just doing some work in Unreal Engine 5 when it died on me. I'm just unable to reproduce the crash at will.

 

[Eluvade]

Update #2: I ran the RAM test and everything seems to be fine.. which isn't good.. because I still don't know what's causing this

 

[kilo]

Hard power off without BSOD is likely power delivery. Remove your battery and test it on just the AC adapter. If it's fine, reinsert your battery and make doubly sure it's inserted fully. Loose connection there might not be obvious by sight alone.

 

[Eluvade]

Hard power off without BSOD is likely power delivery. Remove your battery and test it on just the AC adapter. If it's fine, reinsert your battery and make doubly sure it's inserted fully. Loose connection there might not be obvious by sight alone.

That's actually a pretty good idea! I'll try it out as soon as I get my hands on a screwdriver. Thank you

 

[Eluvade]

I installed a few GPU benchmark/stress testing tools and the first couple of times I ran them, the laptop would crash, but I was just able to complete one without crashing.. I don't really understand what's happening :/

 

[kilo]

Repaste the CPU and GPU. Other than that it's possible you have a bad PSU and battery connection. You may have to take it into a shop.

 

[unclewebb]

In the above screenshot, the cores are equally loaded but look how big the temperature difference is. One core at 67°C while the one beside it is at 91°C is not a good sign.

Uneven maximum core temperatures can be a sign that the heatsink is not sitting square to the CPU or the thermal paste was poorly applied or the thermal paste has pumped out. Some heatsinks are so flimsy that they can become warped. Random shut downs can happen when cooling across all cores is inconsistent.

VBS was already disabled

Once upon a time VBS might have been disabled. Your ThrottleStop screenshot shows me that VBS is enabled. Some Windows updates can enable VBS behind your back so you have to keep an eye on things. Windows Update likes to enable core isolation memory integrity every chance it gets. Post a screenshot of Windows Features if you are not sure why VBS is enabled. If you see a bunch of odd looking voltage numbers (0.3799) in the FIVR monitoring table, that is a sign that VBS is enabled.

 

[Eluvade]

Once upon a time VBS might have been disabled. Your ThrottleStop screenshot shows me that VBS is enabled. Some Windows updates can enable VBS behind your back so you have to keep an eye on things. Windows Update likes to enable core isolation memory integrity every chance it gets. Post a screenshot of Windows Features if you are not sure why VBS is enabled. If you see a bunch of odd looking voltage numbers (0.3799) in the FIVR monitoring table, that is a sign that VBS is enabled.

Does this not mean VBS is disabled?

 

[unclewebb]

You cannot use Windows Subsystem for Linux and ThrottleStop FIVR voltage control. These two are not compatible.

You might have to disable Hyper-V as well.

If you make any changes, reboot and delete the ThrottleStop.INI configuration file. Some users have reported that this file can get corrupted.