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

For fear that this question is too specific, I'm posing the general question first and including details of my own situation below.

General Question

If I know that my system is overheating (to the extent that it's actually shut itself down a few times), how can I determine what's causing it to overheat? What are the most likely causes/solutions (e.g., CPU heatsink dysfunctional, thermal paste not applied correctly, internal fan damaged, etc.)?

Specific Question

So, I'm pretty certain my CPU is running way too hot. According to HWMonitor, both of my Athlon 64 X2 5400 cores are around 80C. I understand this is way too high.

I THINK the issue may be my heatsink, so I am ready to order another from Newegg. However, I don't want to go ahead with that if there are other likely causes of this overheating. For example, how can I be sure the fans inside my system are working properly?

My motherboard is an ASUS M3A78-EM. When I go into setup when starting my computer, and navigate to the Power section, I see the following under Fan Speeds:

CPU     3000 RPM [approx.]
Chassis 1300 RPM [approx.]
Power   N/A

Should I be concerned about that "N/A" next to "Power"? Do I need another fan?

Also, in HWMonitor, what does the "Fans PWM" section indicate? There are three entries, and they all read 0% across the board. Here's a screenshot, just to make it clear:

alt text

Clearly, I am rather helpless here. Where do I start? How can I fix this problem?

share|improve this question
up vote 4 down vote accepted

Can you verify the voltages using a different tool, perhaps Everest or SpeedFan ?

Those +5v and -12V rails are low, I want to confirm that this is because of HWMonitor readings being incorrect. I'd start with examining if the thermal paste has dried up. If so, you should re-apply the thermal paste.

share|improve this answer
SpeedFan confirms those voltages. What should they be? Could re-applying the thermal paste address that issue? – Dan Tao Sep 6 '10 at 22:46
@Dan Tao I can't imagine that those +5v and -12v rails are providing +3.3 & -6v, those should be as close to the reference ( ie, +5v & -12v) respectively. It is conceivable that your PSU has gone bad, providing low voltages. Low voltages == harder the CPU has to work to process == higher temps. See if the BIOS reports the same values, if so I'd recommend changing your PSU ASAP. – Sathya Sep 6 '10 at 22:58
@Sathya: OK, thanks for the input. So since posting the question the +5V reading has come up to about 4.95 V and is staying pretty steady there. The -12V reading is different between HWMonitor (around -6 V) and SpeedFan (around -8 V), but both seem to be pretty bad, if I'm understanding you correctly. What I am planning to do is buy some new thermal paste for the CPU temp, and, if I can confirm these bad voltages from my BIOS when I restart my PC, I will replace the PSU as well. Does this sound like a good strategy to you? – Dan Tao Sep 6 '10 at 23:03
@Dan The fluctuations indicate that the PSU is not able to cope up. Do replace the the thermal paste, and keep an eye on those voltages. If they vary wildly ( as you just said) then yes, replace the PSU. That seems good. – Sathya Sep 6 '10 at 23:15
@Andreja I undervolt my laptop processor, so I know a little about undervolting. When you undervolt, you reduce the voltage provided to the processor by a small amount. But here software indicates that the voltage coming from the rails is low. The two situations are not the same. @Dan Tao: Can you confirm the voltages by having a look at what the BIOS says ? – Sathya Sep 7 '10 at 13:47

Your first step should be to actually check how many fans you have and to what is each fan connected. The names for fans are just that: names.
Usually on Asus motherboards CPU fan connector is close to CPU socket, so CPU fan is connected to it. Other connectors have names too, but their names do not have to match their functions. You can plug any fan into any connector and it will show up under that connector in diagnostic programs. Unless you actually plugged the power supply fan into the power supply fan socket on motherboard, PSU fan reading has no relation to what PSU fan is doing.

The PWM comes from Pulse Width Modulation and is a way of controlling fan speed. I wouldn't be concerned with zeros there. Check how many pins the fan connectors have and how many wires the fans have. PWM connectors should have 4 pins. If fan cables have less, then the aren't using PWM.

Also your heatsink could be badly installed. Try reinstalling it first.

I myself would be more concerned with voltage readings. the 3.41V at +5 line is dangerously low and can be a symptom of bad PSU. Check what voltages in your BIOS say, because sometimes diagnostic software can report bad readings.

Also that 128C temperature looks really bad, but on the other hand it could be some phantom reading.

share|improve this answer
It seems the voltage readings are probably accurate (same measurements from HWMonitor and SpeedFan). What is the solution to a low voltage? Buying a new PSU? – Dan Tao Sep 6 '10 at 22:52
@Dan Tao First make sure you check the readings in BIOS as well! I've seen cases where Asus motherboards gave really strange readings to monitoring utilities. The thing is that if you +5 voltage is really 3.41V, your motherboard shouldn't have allowed your computer to boot. During boot voltages are tested and if they are out of ± 5% for +5V computer should just wait with blank screen until they stabilize. Also, if you can, verify voltages with a voltmeter. Take a look here: voltage between red and black wires should be around 5V. – AndrejaKo Sep 7 '10 at 8:40
@Dan Tao I'm not saying that it's not PSU, because when overloaded PSUs will shut down, but we need to be sure first. Why change the PSU if all you need are couple of fans. – AndrejaKo Sep 7 '10 at 8:41

A great way to check thermal issues on a system is to use a laser temperature sensor and a thermal camera.

share|improve this answer

You must log in to answer this question.

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