Hot CPU faulty?

[jeromek]

Hi,

I'm having a lot of trouble installing a second Xeon E5410 into my Dell T7400. After installing the CPU several times using Artic Silver 5 (following the instructions to the letter), I'm still seeing CPU temps of 85 degrees under high CPU load (20 degrees higher than the other CPU and ~15 degrees over the processor's operational limit). I don't think the thermal interface is the problem, because the heat sink on the second CPU (both are original Dell items) is noticeably hotter than the first. I haven't measured the difference yet, but after forcing lots of air across the two heatsinks, low temperatures are maintained under high CPU load. This leads me to think that the CPU/socket may be faulty. I'm considering swapping the CPUs around in the sockets to see how this affects temps, but I'd like to get a bit of feedback from the pros before taking this on.

Has anyone seen something like this before? Any thoughts are much appreciated!

 

[ulillillia]

A common case that leads to this is that the heatsink and/or fan is not seated properly. Try reseating the heatsink and/or fan before attempting to swap CPU locations. If that doesn't work, check to make sure you have enough (but not too much) thermal grease (the Arctic Silver 5).

 

[jeromek]

I've reseated the CPU again, being _very_ careful to ensure that everything is done properly. I'm fairly sure that the heatsink is on correctly and the thermal interface is good. There is no heatsink fan on these Dell models: just a single large fan in front of both CPUs blowing air across them.

What's really bothering me (and it's what seems to prove that the thermal interface is good) is that the heatsink on the new CPU is significantly _hotter_ than the old CPU. Doesn't this mean that the heatsink is doing its job, and it's basically just not able to dissipate all of the heat that the CPU is generating? Even at idle, when the temperatures are perfectly acceptable, there is a noticeable difference between the temperature of the two heatsinks. When under load, the new heatsink heats up very quickly and the reported CPU temperatures steadily climb up to 80+ degrees. Unfortunately, I can't be more precise about exactly what the surface temperature difference between the two heatsinks is: I don't have access to a thermometer that will measure surface temperature.

Do you think it's feasible that the CPU/socket is generating all this extra heat? Or have I got it all wrong with respect to the thermal interface? Thanks for the answer, and any further comments are much appreciated.

 

[ulillillia]

Not the CPU, the heatsink and/or fan. If your computer doesn't have a CPU fan, I would recommend getting one. You just need to make sure that the fan you get will both fit within the limited space of your case, and that it's designed for your motherboard's socket type (e.g. LGA1155).

Another possibility is that you may be overclocking the newer CPU.

Beyond this, I can't help much - someone else might be able to. Knowing what your first processor is would help some. Given the first sentence in the original post, it's supposedly a Xeon E5410 as well.

 

[jeromek]

Well, it turned out to be a combination of two factors: a duff heatsink and the wrong screws/springs. I bought the heatsink on ebay (Dell part FD841) and it arrived without the screws required. Being a Dell part, these weren't standard screws or anything, so I had to find something similar. The closest match I found was some bolts with the same thread pitch but which were quite a lot longer than the Dell screws. I got some springs and same washers, and figured it would be near enough.

After much messing around trying this and that, I came to the conclusion that it must be the heatsink at fault, as I'd swapped around the processors in the sockets and had no problems. I ordered another heatsink yesterday (and made sure it had the screws included this time!), and it arrived this morning. To put my mind at rest, I tried the heatsink I was having problems with using the correct screws/springs, and there was an immediate improvement. However, the temperatures were still climbing to unacceptable levels, albeit at a slower rate. A quick swap for the new heatsink (I'm an expert at swapping these heatsinks in and out now...) with the correct bolts, and the change was immediate. Under consistent high-stress load, both CPUs show a maximum of 70 degrees, which I would expect to come down a few degrees as the Artic silver goes through its curing process.

It's curious that the heatsink turned out to be defective, as it looks fine. There's no apparent difference between it and the heatsink that's working perfectly now. I'm just happy this bloody ordeal is finally over!

Thanks for your help ulillillia; hopefully this story might spare somebody else some pain in the future.

 

[jeromek]

Best answer selected by jeromek.

 

[Mousemonkey]

This topic has been closed by Mousemonkey