I have a Mac Pro "Eight Core" 2.26 (2009/Nehalem) that I was running with dual X5690 (3.46GHz/6-core/12MB/130W) processors for the last couple years (yes, I've got the 2010 firmware hack on it). Been running fantastic, but I needed to do swap some hardware around and so ended up needing to take these processors out and put some older processors (that were from another actual 2010 Mac Pro; 2.66 GHz 6 Core Xeon X5650 x2).

After the transplant was completed I fired it up and immediately got a light on the right side of CPUA:


I held the power button in and immediately turned it off. This light then lit ("OVTMP CPUA"):

ovtmp cpua

I figured something must be up with the heatsink as it seemed as though the screws for the 2009 heatsinks ratcheted down tightly and immediately instead of having some give that would indicate that it was in contact with the CPU (lidded CPU btw). Made me figure that perhaps the heatsink was not pressing down the CPUA processor enough. I also recalled that the previous processor that was in here may have been SLIGHTLY taller.

I then took the heatsink back off and put about .5 to .75 mm of copper cut to size for the Xeon lid and used thermal paste again to make up for this possible spacing issue. I repeated this for CPUB. The heatsink seemed to ratchet down better this time as I could feel some pressure being applied and the torque didn't stop immediately as before and thus i could torque it down a bit.

After doing this, I unfortunately found that the same error lights were emitted - no change.

I also went ahead got some additional feedback from the DIAG LED button:

diag led results

I'm not really sure what to do at this point. Should I add thicker copper to make the heatsinks apply greater pressure to these Xeon processors?

I believe Intel has pretty good heat protection for these processors, so I don't believe the processor would be "burned out" at this point, right?

Further, would resetting the motherboard by removing the battery and / or pressing the reset button beside the battery (above the 1st PCIe slot / behind the GPU) help at all?

Any other ideas aside from buying a new processor?

  • I thought the 'real' Apple CPUs were all un-lidded on these 4,1s - there's reams of info out there on how to de-lid the [non-Apple] X5690s. Might that mean that at some time in the past, someone has changed the functional gap some other way? [I have X5690s in mine, but I bought it like that so I've never taken it apart to see.] – Tetsujin May 20 '19 at 6:50
  • Red OVTEMP can indicate either an overheat or missing CPU - as a first test, I'd swap the processors round & see if it follows. I presume you already have the Service Source manual? If not, you can get it from tim.id.au/laptops/apple/macpro/macpro_early2009.pdf – Tetsujin May 20 '19 at 7:01
  • Regarding un-lidded vs lidded, the system was running lidded processors before - but originally came with un-lidded processes, yes. There are a variety of options available for upgrade on eBay and elsewhere, example, notice how they're selling them both ways and advertising for 2009 on up: cl.ly/eaa8c5d95f2a – ylluminate May 20 '19 at 15:10
  • @tetsujin so as per your first test recommendation, I did swap CPUA & CPUB, but getting the same failure for CPUA - no changes. This leads me to think something is wrong with the socket interface in some fashion... but how is the big question as there seems to be no noticeable differentiation or problem... – ylluminate May 20 '19 at 23:49

You must log in to answer this question.

Browse other questions tagged .