Sign in with
Sign up | Sign in
Your question
Solved

Laptop restarts everywhere but Safe Mode

Last response: in Systems
Share
February 2, 2012 10:36:44 PM

So I'm trying to fix a friend's laptop (Acer 8735G-6502) that keeps restarting, but I can't figure out what the problem is.

Symptom: In Windows 7 (or Linux) the laptop suddenly shuts off with no error message then restarts. It appears that it can happen at any time (heavy load, during startup, web browsing, etc.), but might not do it when idling. Also, the DVD drive is not recognized by the bios or Windows. That may be unrelated.

Checked so far:
Power - I've tried it unplugged/plugged with the battery and without. (It doesn't turn on with no battery when not plugged in ;) ). I've tried numerous power strips and outlets.
RAM - I've ran about 4 hours of Memtest86+ from USB with no errors.
DVD - It does the same thing with or without the DVD drive connected.
HDD - I've removed the HDD, swapped in a new one w/ fresh Windows install, and ran Linux installations from USB with the HDD and DVD (all the SATA devices) removed. It still crashes just the same.
Graphics - Discrete GeForce GT 240M. I have not tested this.
Motherboard - I've eliminated SATA ports & SO-DIMM slots as the cause so far.
CPU - Here's where it gets strange. It has, thus far, NEVER crashed in Windows Safe Mode (or Safe Mode with Networking), yet it crashes in Linux and Windows. I've ran LinX to stability test the CPU in safe mode and do not get errors.
Network - It restarts whether or not there is a connection on the Ethernet port.

Summary:
It can restart with no warning in Windows 7 and Linux on USB. It does not appear to ever restart in Windows 7 Safe Mode (even when left overnight running virus scans and defrags), seemingly ruling out an inconsistent power connection as the culprit. No hardware tested appears to be related to errors. I guess I need to try harder to crash it in Safe Mode or try a Windows XP installation?

Any ideas? So far all I've got is a dying motherboard?

I also posted this thread here: http://www.tomshardware.com/forum/75662-35-laptop-resta...
Sorry Mousemonkey if this is not the right section/format.
February 3, 2012 4:34:04 PM

The description of the problem seems to conflict with itself.

It doesn't crash in safe mode. That points to drivers.

However, it does crash in Linux, which is not an easy thing to make a machine even do.

Stab in the dark...

Overheating.
m
0
l
February 6, 2012 3:15:22 AM

Thanks for the reply--I don't know why I didn't get an auto-email?

HWMonitor doesn't show high temps while it's still on in normal Windows. If overheating were the issue, it would be more stable on first startup and decrease in stability with time (they might be A LITTLE to that, let me double check). And it never feels warm.


Safe Mode/Drivers--yeah, I'm pretty pissed about it shutting of in Linux. That's why I removed the SATA drives in case it was a hardware short and ran Linux of a 16GB USB stick.

Because of the Linux crashing, I'm inclined to think it's a physical interruption of power to the motherboard or CPU. But why doesn't it ever crash in Safe Mode? AHH!!! So frustrating!

I guess I'll try safe mode a 10th time. If I can make it crash, then I can at least hone my search to a hardware problem.
m
0
l
Related resources
February 6, 2012 4:55:28 AM

Linux is quite tolerant of faulty hardware according to my knowledge and my experience matches up with it.

I have a drive that reports errors every single time that the computer boots with it installed. It transfers data at about half of the speed it should and often errors show up during the transfer.

My Windows XP and 7 partitions wouldn't read the drive if I just turned the computer on and booted into Windows.

If I booted into Linux, though, it would work pretty much normally.

Linux is just built to be a whole lot more tolerant of faulty hardware.

I haven't seen Linux crash due to software at all, but I haven't spent the most time with it either.

Anyway, if it were crashing inside Windows regular mode I would think it were drivers, bad PSU, or heat. If it was Linux, the same drivers wouldn't apply, so it should work.

That would get me thinking either heat or bad PSU, or both (excessive heat internal to the PSU degrading its performance).

Anyway, if I were you, I would open the side of the case and aim a regular AC oscillating fan in it and see if stability improves.
m
0
l
February 6, 2012 4:56:37 AM

dalauder said:
Thanks for the reply--I don't know why I didn't get an auto-email?


I think it only emails you if you are quoted.

This should be able to prove my theory.
m
0
l
February 6, 2012 9:18:53 PM

Hmmm...I must've setup the "email me" thing wrong (failed to check the box). It's a laptop, so the PSU is the power adapter. It crashes both on it and the battery, so it would have to be the power connections on the motherboard (not impossible, but HARD to fix).

I've gotta boot to safe mode again and see if I can get it to crash. If it won't crash on LinX in Safe Mode, it's not CPU heat. But I guess that doesn't rule out GPU heat.
m
0
l
February 7, 2012 10:27:34 PM

Well it's apparently Prime95 Blended Test stable in (Normal) Windows--a couple hours in so far and only hitting 77C, safe for a mobile CPU. However, it was crashing before I started Prime95--possibly randomly or possibly when accessing the HDD.

How can something be Prime95 stable but crash without warning!!!?
m
0
l
February 7, 2012 11:00:42 PM

Normally you'd have to say that was a graphics driver problem , but the multi OS and reinstall means its a faulty graphics chip or motherboard
m
0
l
February 8, 2012 4:40:08 AM

Update: So far it's 8 hours Prime95 stable (up to 80C) and I'll run it overnight.

So it definitely could still be the graphics chip or motherboard. The Prime95 stable, but consistently crashing when I touch something else thing just feels wrong--very frustrating.
m
0
l

Best solution

February 8, 2012 6:25:49 AM

I have never used prime95 [or any other stress testing software because it usually doesnt actually tell you anything you dont know to start with ]

But I can speculate . If its stressing your cpu and maybe the 2D sections of the gpu it wont be telling you if the 3D portion of the GPU is functioning when you are running in DX9 [XP?]

If thats the problem and you installed Vista or later with aero turned on expect it to crash
Share
February 8, 2012 4:25:55 PM

I would suggest trying to stress test the video card.

I don't remember exactly, but I think Prime 95 does the processor and RAM primarily and other things focus more on video cards like Furmark and Passmark.

I would give Passmark a try and if it stays stable consider trying Furmark for a short period. I do mean short too, because Furmark has been known to destroy perfectly good components from all the strain.
m
0
l
February 8, 2012 8:28:40 PM

23 hours into Prime95 I decide to stop prime95 and insert a flashstick with updated drivers and MSI afterburner. But it crashed as soon as Prime95 stopped, which I thought was funny.

Afterburner can do mobile GPUs, right? If so, I want to try underclocking the card. I'll also turn off all Aero effects and Windows graphics stuff and see if that helps. Can you edit mobile graphics bioses with Nibitor and reflash them? It'd be awesome if that fixed it.

Don't worry about Furmark, I'm well aware of its heat generating abilities and always watch my temps. I'll attempt to underclock first though. Hopefully Linux was also attempting to launch graphics drivers.
m
0
l
February 13, 2012 9:20:51 PM

Hmmm, I flashed the graphics bios (downloadable from Acer's website and already installed via previous update) to a reduced clock one I edited in Nibitor, but GPU-Z is reading out 0MHz on all the clocks, so I don't know if it took and it's not stable enough to open MSI Afterburner. I've had a lot of problems since installing the updated drivers, so I think I'll roll back to pre-driver to check if I can read clocks.

I'm pretty sure the GT 240M is dying. I guess I'll try baking it since this Acer GT 240M is on its own PCB. That should be interesting :) .
m
0
l
February 13, 2012 9:25:08 PM

Definitely doesn't sound like that driver is the one you want.
m
0
l
February 16, 2012 9:59:53 PM

Well the GT 240M seems to fail whenever attempts to load drivers occur--now even in safe mode (and of course still Windows & Linux). The graphics consistently work in the bios and Memtest86+, but that could just be because those environments are resistant to graphics failure and automatically restart.

I baked the graphics card, as Acer's have removable GT 240M's. That didn't change much, if anything. It may have made it slightly less reliable. Occasionally, I get one long and two short beeps on power on with no post. What I've found online so far indicates that as a graphics failure.

My last check is gonna be some DOS-based CPU tests after I get it turning on again. I might go ahead and order an MXM-A slot GT 240M soon. There's one on ebay for $85 OBO, which likely will mean $55, which isn't bad at all.
m
0
l
February 20, 2012 9:51:53 PM

Best answer selected by dalauder.
m
0
l
January 18, 2013 1:40:25 AM

I believe that I've found the answer. Notsosure's link to the Toshiba thread really helped in figuring out the logic behind this mess.

What's happening is that certain capacitors on the motherboard that regulate the current of the CPU have started to fail. So whenever the CPU changes its clock speed or multiplier, the capacitors do not deliver the required voltage to it and the CPU malfunctions. This causes the system to reboot. Some users have fixed the issue by replacing the faulty capacitors (they are located directly behind the CPU socket on the motherboard).

Intel refers to this throttling technology as Speedstep. Windows Safe Mode seems to not use this CPU function which is why the system doesn't reboot in that mode. Disabling all the processors in Device Manager prevents normal Windows mode from using it as well. As long as the CPU doesn't use Speedstep, no spontaneous reboots will occur.
m
0
l
January 18, 2013 7:10:32 AM

Wow, that sounds like a good solution. I'll try to disable speedstep...although now it won't seem to boot consistently, so it might be more general motherboard failure...also, I don't work with the guy whose laptop I was fixing anymore. But I'll try to get at it if I get a chance.
m
0
l
!