Sick laptop

I have an HP Pavilion laptop that I inherited from a friend a few years
ago. It's got Windows 7 64-bit on it. It's been running pretty well all
along, though it tends to run very hot.

I either shut it down last night or closed the lid and let it go into
hibernation. Today when I turn it on, I get a garbled screen on bootup,
before Windows even starts. I can read the screen well enough to
navigate around a bit, and even the HP options and BIOS screens are
garbled. I hooked the laptop up to an external monitor and it's still
garbled. I can let Windows try to boot up part way then bluescreen. I
can boot to Safe Mode. I booted up with Linux on a USB stick and the
display was still garbled, but I was able to see a message there about
it running without hardware acceleration. Because of all that, I assume
it's a hardware failure, not anything to do with the OS. Here's a short
video showing what the display looks like

https://youtu.be/7wvFT3UzFu4

I was able to see well enough to run HP's built-in RAM test, and it
passed that.

So, do you think this machine is toast? Any ideas about what's going
wrong and if there's anything I can fix or replace to get it going
again? I plan to open it up and just re-seat things, jiggle wires and
cross my fingers. Otherwise I don't know what else to do.

Nil wrote:
I have an HP Pavilion laptop that I inherited from a friend a few years
ago. It's got Windows 7 64-bit on it. It's been running pretty well all
along, though it tends to run very hot.

I either shut it down last night or closed the lid and let it go into
hibernation. Today when I turn it on, I get a garbled screen on bootup,
before Windows even starts. I can read the screen well enough to
navigate around a bit, and even the HP options and BIOS screens are
garbled. I hooked the laptop up to an external monitor and it's still
garbled. I can let Windows try to boot up part way then bluescreen. I
can boot to Safe Mode. I booted up with Linux on a USB stick and the
display was still garbled, but I was able to see a message there about
it running without hardware acceleration. Because of all that, I assume
it's a hardware failure, not anything to do with the OS. Here's a short
video showing what the display looks like

https://youtu.be/7wvFT3UzFu4

I was able to see well enough to run HP's built-in RAM test, and it
passed that.

So, do you think this machine is toast? Any ideas about what's going
wrong and if there's anything I can fix or replace to get it going
again? I plan to open it up and just re-seat things, jiggle wires and
cross my fingers. Otherwise I don't know what else to do.

Would this be one of the machines with the NVidia GPU problem ?

There was a problem with solder balls. NVidia provided something
like $300 million to OEMs, to cover the cost of repairs.

Some (disreputable) OEMs decided that instead of fixing the
GPU (hot air machine, remove and replace), they would release
a BIOS update that runs the fan more aggressively. The intention
was, to cool the circuit down enough, it would fail
outside the warranty! Very clever.

So before doing anything, I would be reviewing what GPU is
in there.

*******

A second possible answer, is the heatpipe has come loose
from the GPU, the GPU overheated, and is now "crazy". On
desktops, when the GPU overheats, sometimes it gets hot
enough to melt the plastic on the fan. Once the die goes
over 135C, there could be some permanent damage. And if a
hot chip loses its cooler, it can shoot up to 200C in
a couple seconds.

One failure scenario with no visible symptom, is if the
heatpipe springs a leak, the working fluid escapes - this
causes the heatpipe to conduct heat poorly, and both
the CPU and GPU could get hot. If the CPU gets hot enough,
it will turn off the computer via the THERMTRIP signal.
The GPU on the other hand, usually isn't protected like that.
The GPU is just supposed to burn up or something.

Paul

On 15 Mar 2017, Paul wrote in
alt.comp.hardware.pc-homebuilt:

Would this be one of the machines with the NVidia GPU problem ?

I'll check on that. I've been trying to keep a record of all hardware
in all computers under my control... but I don't keep it up well
enough.

One failure scenario with no visible symptom, is if the
heatpipe springs a leak, the working fluid escapes - this
causes the heatpipe to conduct heat poorly, and both
the CPU and GPU could get hot. If the CPU gets hot enough,
it will turn off the computer via the THERMTRIP signal.
The GPU on the other hand, usually isn't protected like that.
The GPU is just supposed to burn up or something.

This does make some sense. One thing this computer does sometimes is
that, when I close the lid and it's supposed to suspend, it doesn't.
I've come back into the room a couple of times and found it still
cranking away even with the lid closed. I wondered whether it did this
again and it overheated during the night and cooked itself to death. I
thought the CPU should shut down in time, but if the GPU didn't, that
might explain things.

Thanks for the tips. When I have time to perform surgery on it now I
have some things to look for.

On Wed, 15 Mar 2017 20:30:26 -0400, Nil
wrote:

This does make some sense. One thing this computer does sometimes is
that, when I close the lid and it's supposed to suspend, it doesn't.
I've come back into the room a couple of times and found it still
cranking away even with the lid closed.

I get that on a desktop, rather rare but not unheard of;- the oldest
flat panel display in existence: a 32 Olevia LCD that's maybe 15 years
old. A software vector, though, part of a display power-out utility,
which I can correct with an OS restoration. A good feature for times
I forget, although, as a rule, the remote's eye is placed facing
forward just under the left side of the display receiving unit, and I
usually remember to push it off (or the display's off button) when
leaving it unattended (rest of the computer is left on 24/7).