Super User is a question and answer site for computer enthusiasts and power users. Join them; it only takes a minute:

Sign up
Here's how it works:
  1. Anybody can ask a question
  2. Anybody can answer
  3. The best answers are voted up and rise to the top

I recently moved in to an apartment with out air conditioning. This is fine most of the time as I am in upstate New York. It only ever gets above the high 70s during the hottest of the summer months. And when it does, I'm stubborn enough that I'll just deal with wearing minimal clothing around the house.

However, I'm worried about my computers. I'm a software developer and gamer, so many of my machines are very high powered. And at least one of them is a server that must be left on 24/7 (not just a game server - also serves multiple websites). I've never before had to worry about the heat too much, as I always lived in buildings with central air. The in building temperature rarely got much above 70 F. All of the machines I built had good enough air cooling that I never saw a problem.

Now the temperature in building is pushing 100F and I'm worried that the machines will not be able to keep themselves cool enough by simply blowing already hot air over themselves. The hottest of them I've turned off. However, the server I cannot. It's an old Dell (not custom build) that runs on a Pentium 4 (2.2GHz). It only has a single hard drive, integrated video. And it'd not running any processor intensive servers. Just basic LAMP. It used to run a MUD server, but that's off for now. So it should be idling most of the time.

I haven't been able to find any sort of built in temperature sensors in the hardware... at least not any that the programs I've found in the Debian repository can read. And it's an inherited machine to which I do not have the full specs, so I don't know the tolerances anyway. How worried should I be about it melting down on me? How worried should I be about the hard drive melting or becoming corrupted?

To generalize the question for other people, what are the safe temperature tolerances for most machines. How widely does it vary, and how does one go about determining when their machine is running too hot and needs to be shut down?

share|improve this question
up vote 9 down vote accepted

In general - it depends. ;)

All CPU's and GPU's should have a maximum thermal limit specified on the manufacturers website. For example, this processor has a recommended maximum thermal limit of 75C. Although some GPU's don't specify, a generally acceptable maximum upper limit for GPU temperature is 90C (even though certain CPUs/GPUs can technically go higher, this is probably way too hot to sustain under load).

If your temperatures are really concerning you, then you should consider cleaning all of the dust buildup in the case, adding another fan (or repositioning the older fans), and either getting a new CPU heatsink, or re-seat the old one with new thermal paste (which dries out over time).

An absolutely essential tool is HWMonitor - it will provide you various temperatures of your computer from almost all sensors inside of it, even hard drives.

Also, remember - your fans should suck air OUT of the case. Try to get a cross-ventilation happening, with air moving through the case from one side to the other.

Back to the CPU's now. Pentium 4's have an upper thermal limit of 65-78C, with the lower limit obviously preferable. If you hit the limit, there's a few things you can do:

  • *under*clock the processor
  • reduce the core voltage (may have to underclock or lower the multiplier to stabalize it)
  • get a new heatsink (see above)
  • put more fans blowing air onto/moving air away from the CPU

Since you use Debian, you might want to try the GNOME Sensors Applet, although I'm sure there is a million other temperature monitors out there that you could use.

share|improve this answer
Good answer. Just to add, hard drives usually operate in temperatures up to 55C or 140F, very hot. Breakthrough, since the OP mentioned LAMP/Debian, can you suggest any Linux temperature measurement tools? – hyperslug Aug 20 '09 at 17:41
I added a very good tool to the bottom of my answer. There is another package I used in the past, Hardware Monitor, but it's more obtrusive then the GNOME Sensors Applet. – Breakthrough Aug 20 '09 at 17:47

Computer companies such as DELL typically test their systems installed in a chassis in an oven at temperatures higher than 100 deg F. I, personally, wouldn't be too worried (also I live in Texas, so we have had 100+ deg F weather for the past 3 or 4 months...with AC, of course). :-)

The utilities mentioned here are great. Keep an eye on your system and if you start to see it throttling because it is getting too hot, then consider adding/adjusting the cooling. If the system is throttling, an inexpensive, temporary and easy step is to just buy a box fan or two from the closest department store and set it to blow air onto the system.

Now, home-built/assembled systems have increased risk of having thermal issues unless the system builder knows what they are doing. If you have problems with these, you could just open up the chassis and blow the box fan directly onto the motherboard. Hot spots on a motherboard typically have a heatsink on them or the heat sinks into copper on the PC Board. The higher the LFM (velocity) of the air, the better.

Another thing to look for is to keep the chassis out in the open air. If you stuff it into a desk or cabinet, then you're creating your own oven where the system is the heating element.

share|improve this answer

For the record, I live in Australia, and my system copes with 45c days without AC - worst case my monitors shut-off because of heat.

share|improve this answer

It isn't the exterior temperature that matters, but the interior temperature. There is no rule of thumb saying that computers should be shut down after the air temperature gets to 100 degrees or 107 degrees or something. In my experience, if the computers get too hot, they will shut themselves down safely.

share|improve this answer

Humidity could be a concern too; both in terms of damaging electronics and inadvertently voiding the warranty.

share|improve this answer

You should be able to purchase a few temperature gauges that you can install via USB to the machine, you could get a USB PCI card to and a few thermal guages, there are also thermal solutions that will be able to kick on additional fans automatically or manually. Nvidia makes one for there GPU's. Asus boards also have the cool'n'quiet software that monitors temperatures.

Your best bet is to look on tigerdirect for a fan controller/software combo. You could also go with a liquid cooling solution instead. depends on load.

share|improve this answer
There are also hardware solutions - like this one from NZXT which are independent of the computer - meaning it will control temperature even if the system crashes (also means you don't have to find drivers). – salmonmoose Aug 21 '09 at 3:36

You must log in to answer this question.

protected by Journeyman Geek Jul 5 '12 at 9:23

Thank you for your interest in this question. Because it has attracted low-quality or spam answers that had to be removed, posting an answer now requires 10 reputation on this site (the association bonus does not count).

Would you like to answer one of these unanswered questions instead?

Not the answer you're looking for? Browse other questions tagged .