[zagmund]

Hi all,

Short story - Toshiba POC is complaining under Windows that it is overheating, and I can't run any diagnostics which will tell me whether this is actually the case or not as it insists on shutting down.

I'm looking for a distro I can boot from CD or DVD which will have a set of tools on it providing things like temperature readings, etc . . . I'm sure the tools are relatively standard, but I just don't know which ones to use.

I plan on investigating properly and getting it fixed as I know the problems which overheating can cause, but I just want to be sure that it actually is overheating before I send it in for repair.

Any suggestions would be welcome.

z

[Johnboy1951]

Most distros will boot from a CD/DVD/USB device.
If they don't have the software included in the release it should be installable while running live ..... would be lost on reboot unless using a persistent liveOS, for instance on a USB flash stick.

An alternative would be to search from distrowatch.com page for the type of distro you require to do the diagnostics.

I could not be sure which would be best, as I use my regular distro with the utilities installed for such purposes, and have no idea what comes pre-installed

[Khannie]

Grab yourself an ubuntu live cd, once booted open up a terminal and type:

sudo apt-get install lm-sensors
sudo sensors-detect
then you'll need to modprobe all of the various modules it detects
e.g. sudo modprobe i2c-sensor
then if you just run "sensors" in a terminal it will spit out the various bits of data. (or even better run "watch sensors" and you can see changes happening in near real time).

If you're certain something's overheating though, you may not even get that far. It might be worth reseating the CPU heatsink first anyway. At a minimum I would underclock it in the bios if that's possible for you.

[zagmund]

Thanks - sensors-detect is finding stuff for me and sensors is giving back temperatures so I'm on the right track now. I'm trying it on a good system at the moment to get the hang of what it can do for me. I'll try tomorrow on the victim machine when the beer effects of Friday night have worn off

z

[zagmund]

Here's what I've got -

radeon-pci-0100
I'm taking this to be the graphics card and/or GPU - zoomed up from 60 to 80, but then stopped and is kind of hovering now around 77 degrees. Initially I took the constant climb to 80 to be indicative of a problem, but since it stopped and dropped I now take this to be indicative of proper behaviour. My concern is whether 75-80 degrees is normal for a graphics card

coretemp-isa-0000
coretemp-isa-0002
These are described as core 0 & core 2, so I'm taking it that these are the CPU cores. Nothing else was discovered by sudo sensors-detect. These are hanging steady at 60 degrees and have a note with them to say high = 95 and critical = 105 degrees.

Meantime the fan is going like mad. It's not really bad, but it's on all the time. This concerns me. However, if the temperature is being maintained, I could live with that.

Does anyone know whether 80 degrees is normal for a GPU (showing the default mostly white Linux Mint desktop) I know they produce a lot of heat normally so this may be within limits.

In addition, does anyone know whether the other two items measured (coretemp-isa-0000/2) are the CPU temperatures ?

I'm turning it off now while I do more research, but if anyone knows the answers to the above do please let me know.

z

p.s. I forgot to add, the temperatures for my ancient HP Pavilion tx 1000 (cool little mini machine) which I'm typing this on and which isn't running the fan are -
core0/1 - around 45 degrees
nouveau-pci-0028 - 67 degrees (assumed to be the GPU)
acpitz-virtual-0 - 42 degrees (unkown virtual power management device)

[Johnboy1951]

CPU temps at 60C with little or no work being done is high, IMO. The fan seems to need to run a lot to maintain that temp

GPU temps used to run at much highertemps; IIRC I had an NVidia card at one time which showed its limit temp at 121C

It normally ran up in the 60C range.

I would suggest you look to cleaning any fluff/dust-bunnies out of the CPU heatsink ..... maybe even taking the heatsink off, cleaning it and the CPU, applying fresh thermal paste, and refitting it.

[zagmund]

Thanks for the advice. I watched a short video (Jedward for the Eurovision . . . . woooh) to see the impact and recovery.

The graphics went up to about 90 and dropped back to 85 for the last 10 minutes
The cores went up to the early 70s and have dropped down to 68

My taken on this is that things are operating mostly normally (you see temperatures go up when stuff like displaying the video is happening) but it's not going down as fast as I would expect.

This looks like it's a heat dissipation issue. I will do the dust and fluff blowing away thing as suggested and see if this helps. If it doesn't help, I will look at the more radical solutions of taking bits off, cleaning and possibly replacing them.

I'm slightly glad to see that it's both the graphics & CPU with the issues - this indicates that it's not one of them actually failing, but just the cooling. However, I am not a scientist . . .

z

[zagmund]

Woot.

I took it apart according to the instructions on www.irisvista.com. When I took the fan out there was a fair amount of fluff between it and the vents which crossed the heatsink. Like a lot of fluff. This clearly explains the loss of heat dissipation.

I took it out and blew through the vents as I didn't have any compressed air to hand, then I put it back together. I've just run it through the same Jedward video test and it's at 65 for the graphics (up only a few degrees compared to idle state) and 49 for the two cores.

So, it looks like I have my problem identified and mostly solved. I'll buy a can of compressed air during the week and give it a good seeing to later.

Thanks for the suggestions and help.

z