sudo kate /etc/modprobe.d/thinkpad_acpi.conf
added the line: “options thinkpad_acpi fan_control=1”
This was the very, very strange solution to a very, very strange problem that plagued my (otherwise wonderful) Lenovo Thinkpad T410, I finally found it in a post on http://ubuntuforums.org/archive/index.php/t-836958.html
Here is what happened: I am frequently running extensive (hours), heavy scientific computations (full load on both cores). During these, my core temperature would heat up to 95°C and higher, and eventually shut down due to overheating. Adding the above line to the thinkpad_acpi.conf file fixed it. Let me give you a more detailed description of the problem, and why this is a really strange fix…To start off, I was quite surprised to see my laptop blackout due to overheating. After all, it is Lenovo Thinpad T410 with a dual core i7 M620 processor. So, we are looking at one of the sturdiest machines you can stick in a backpack, equipped with a chipset built for crunching through immense computations, for hours day in day out – not to mention a chipset that draws much of its performance profile from internal heat management, with a safe, working CPU junction temperature (T junction) of 100°C. Really, not the kind of system you would expect to overheat.
Here is the pattern that repeated over and over, till I was just about ready to send my laptop in for having it checked by Lenovo. I would start a two core parallel job in MatLab that would take several hours to finish, running almost always at full CPU load in two threads. Initially, the system runs at full clock speed of 2.66 GHz on both cores, and rather quickly shoots up to 95°C. Then, clock speed is throttled down to 1.73-1.9GHz on both CPUs, system cools down to maybe 85°C, and frequency goes up to full 2.66Hz again. This cycle repeats a couple of times, each time reaching closer to the (admittedly scary) 100°C mark. The moment it hits 100°C, black screen saying: “The system is going down NOW!” And, that is just what followed. Laptop is off. This scenario happened to me probably 200 times, if not more often. In a sense I was amazed – a delayed negative feedback loop, that became unstable and oscillates too much. I was doubting very much that the IBM/Lenovo/intel engineers would have missed that behaviour. The alternative explanation: UBUNTU!
Going through different forums, blogs, mailing lists etc. (open source is nice, but this part of it is not) I found different methods to control the fan speed, which was my first bet. To cut a long story short, adding
options thinkpad_acpi fan_control=1
in the thinkpad_acpi.conf file finally fixed it. It is basically a line that allows the intel i7 CPU on-board machinery to take control of the laptops fan. After I added this, however, it was like I was working on a new machine. Idle temperature of 45-50°C (75°C before), under heavy load 85°C, full load long jobs running stable for hours and hours. Also, the CPU frequencies were responding to core temperature in a totally different way: faster, more accurate. In short, my heat-plagued T60 came back to senses, and is since back in shape, with an outstanding computational performance you would expect from this kind of system.
Now, why giving the acpi control over the fan changed the whole CPU management, I do not understand – as so many other things in Ubuntu. I believe that the fan control switch is in fact switches several, if not all, parts of the intel core i7 on board acpi on. All I do know – my system is working since, and who ever is stuck in the same situation will be deeply grateful to find this information. I would have been. Also – maybe someone developing on Ubuntu – this seems odd.