Having a web browser "just sitting there" does not mean that it is idling. Too many web sites have very VERY bad javascript perpetually running on the browser, sending your CPU usage through the roof. I find that you have to kill firefox every day or so to keep it in control. Right now my system is pretty much "idling", but Firefox, doing apparently nothing, is pulling between 50 and 90% of one core of my CPU.
I think that everybody yapping about thermal paste is on crack. If manually running the fan can put the temperature into a reasonable state, then messing with thermal paste is pointless. Frankly, everybody who messes with thermal paste and sees a significant temperature change did NOT achieve it as a result of the goo, but as a result of cleaning the cooling fins. Take that layer of matted sticky cat hair and cigarette smoke off the cooling fins and you can see a dramatic drop in temperature. Seriously, a teeny bit of marketing from the thermal goo manufacturers and suddenly a bunch of wanna-be geeks think they can magically turn their CPU's into ice cubes. Not going to happen.
As far as this specific computer's rebooting problem goes, I think that the first step is to determine what, precisely, is responsible for the reboot. Computers don't just reboot -- even if they give the appearance of doing so -- there is some routine in there somewhere that is watching for some conditions to occur, and then causing the reboot. Enable kdump/crashkernel and set heavy debug spew and see what comes out after it goes off. Once you know precisely WHAT is going on, you can address the specific problem. It could be as stupid of a problem as a bad thermal sensor driver that occasionally pulls a crazy value like maybe "ffffffff" = 4 billion degrees.
If nothing shows up in the crash dump and all the thermal sensor readings stay reasonable, then it pretty much *has* to be a defect, and the firmware is setting it off for an invalid reason.
I think that everybody yapping about thermal paste is on crack. If manually running the fan can put the temperature into a reasonable state, then messing with thermal paste is pointless. Frankly, everybody who messes with thermal paste and sees a significant temperature change did NOT achieve it as a result of the goo, but as a result of cleaning the cooling fins. Take that layer of matted sticky cat hair and cigarette smoke off the cooling fins and you can see a dramatic drop in temperature. Seriously, a teeny bit of marketing from the thermal goo manufacturers and suddenly a bunch of wanna-be geeks think they can magically turn their CPU's into ice cubes. Not going to happen.
As far as this specific computer's rebooting problem goes, I think that the first step is to determine what, precisely, is responsible for the reboot. Computers don't just reboot -- even if they give the appearance of doing so -- there is some routine in there somewhere that is watching for some conditions to occur, and then causing the reboot. Enable kdump/crashkernel and set heavy debug spew and see what comes out after it goes off. Once you know precisely WHAT is going on, you can address the specific problem. It could be as stupid of a problem as a bad thermal sensor driver that occasionally pulls a crazy value like maybe "ffffffff" = 4 billion degrees.
If nothing shows up in the crash dump and all the thermal sensor readings stay reasonable, then it pretty much *has* to be a defect, and the firmware is setting it off for an invalid reason.
Comment