A customer has a 5-year-old Pentium D 2.8 GHz processor. It has been running fine until recently. Now, the procesor overheats and shuts down with only several minutes of use. We removed the heat sink and re-applied the thermal grease in case it wasn't making good contact. However, the original grease still looked ok and the behavior is no different after applying the new grease. If a processor was run at high, but not critical temperatures for a few years, can that weaken the processor to the point where it more easily overheats and perhaps cannot be cooled properly any longer? Thanks for any insight.
I can't comment from the standpoint of the silicon itself (I will leave that to the silicon experts), but a normal effect of time and constant operation is a degradation in the operation of the fans that cool the system. This consists of not only the wear and tear on the fans themselves, but the buildup of oily dirt and dust that can affect the fan's ability to spin at full capacity (not to mention block airflow). Remember that a component of the equation used to determine the airflow necessary from the processor's heatsink-fan unit is the temperature of the air being used (the hotter the inlet air, the less efficient it will be dissipating heat); it is thus just as important to have well-performing chassis fans as it is to have a well-performing processor fan. I would suggest that you look closely at whether any of the system's fans have stopped operating or are gummed up to the point where they cannot spin fast enough to provide sufficient airflow. I would also suggest that you (carefully!) vacuum or blow out the case regularly (once a month - and this goes for laptop owners too!). To blow it out, you can use compressed air or an old cannister vacuum with the hose attached to the vacuum's outlet instead of its inlet. Take your system outside to do this so that you do not put the dust back into the air around your system...
spearson, thank you for your reply. I should have mentioned that we already performed a thorough cleaning of the system, which appeared to have only minor dust accumulation. The fins of the heat sink were free of dust, but we removed it and blew it out anyway, since we wanted to reseat the heat sink and replace the thermal grease. All fans (processor, case, and power supply) are clean, spinning, and moving expected amounts of air. Temperature based fan speed control appears to be working fine. Within 5 minutes of startup, the fan speeds gradually increase from about 1,000 rpm to about 3,000 rpm as the processor temperature increases. Temperature alert software starts displaying alerts at 75 C and shortly after that the processor overheat alarm sounds and a few seconds later the system hangs. I'm sure the remaining issue is not one of proper airflow.
xB2Spiritx, thank you for your reply. Interesting idea. Why would the new, smallerr heat sink do a better job than the old, larger model? The construction of the two heat sinks appears similar in that they both use a copper core and aluminum fins, but the older model has more fin mass. Shouldn't the older model do a better job on the older, hotter running processors?
Well, the vapor chamber architecture of the newer heatsinks can dissipate heat more efficiently than the older solid plug designs, but you have to be very careful to choose a solution that can dissipate the necessary wattage. The Pentium D line has a maximum TDP of from 95-130 watts, which is higher than most of the newer processors. I would certainly recommend that you try a different heatsink-fan solution before giving up on the processor, but remember that you are talking about a very aged processor generation...
We finally did discover the issue in our case. Very careful measurement of the heat sink base showed it to be slightly warped about the center. You couldn't see it without placing the heat sink on a hard flat surface. Only one half of the heat sink was in contact with the surface at a time. Although the difference very, very small (thinner than a piece of paper), it was enough so that it had poor heat conduction on one side. Even filling the open side with thermal grease was not good enough. We replaced the heat sink and all was well. I doubt this happened during operation; it appears to be a manufacturing defect. Perhaps when the original pre-applied thermal paste dried out, that was enough to cause the issue.
ryung, I think you must cool the processor to get into BIOS. If it is shutting down that quickly, it must be very hot. Working around the shutoff will most likely endanger your processor.
Beware the factory thermal paste! I'm not saying this is the case with all of them but please thoroughly check it before use. 3 months ago had a new i3 system and started overheating last week. This was my first time doing this and took me awhile to figure out the problem but apparently the paste was partially dried out to being with I took the cooler off and my lines were still there it never squished down properly and dried out. Applied new paste and im back to 50 C.