Mac Pro 4,1 CPU Upgrade - something's wrong...

Discussion in 'Mac Pro' started by MentalVizion, Mar 13, 2017.

  1. MentalVizion macrumors regular

    MentalVizion

    Joined:
    Oct 30, 2013
    Location:
    Austria
    #1
    Hey,

    So I recently upgraded my E5520's to X5675's. I did delid the CPU's myself without any major problems. They also got recognized by the system at the very first try and are generally working flawless. However, once the CPU's get under load, my fan speeds increase quite drastically. Running geekbench for example, boosts the CPUA and CPUB heatsink fans to around 3000RPM. When idling the speeds seem quite normal to me - maybe CPUA and CPUB are slightly to high with around 1100RPM. These are my idle stats right now:

    PCI - 800 (forced to 800 because of the annoying bug when starting up the mac)
    PS - 690
    EXHAUST - 600
    INTAKE - 600
    BOOSTA - 1110
    BOOSTB - 1110

    I am using Macs Fan Control btw. These seem normal to me for being idle.

    BUT: For CPUA the red light between heatsink and RAM banks starts flashing randomly - only once the CPU is under load though.

    After reading the Mac Pro 2009 technical guide, this hints to overheating issues if you ask me. Increasing fan speeds also point towards that direction.

    At first, this red LED also started flashing when I booted up the Mac. However, tightening the heatsink screws sooner or later eliminated that red light when starting up. But it is still there when the CPU is somewhat stressed. And it is only happening to CPUA - CPUB is completely fine.

    So this led me to tighten the heatsink screws as much as possible. I am now at a point, where it's pretty much impossible to further increase the tightness on them, without applying some major force - which I am scared of because of the socket pins underneath.

    A list of stuff I already did:

    Swap CPU's around - I'm pretty confident, that both CPU's are not faulty.
    Check the socket for bent pins - no bent pins
    Do a SMC reset - already did plenty
    Do a PRAM reset - already did plenty
    Check if the heatsink connector is properly seated - already checked, but since Macs Fan Control spits out a value for the heatsink RPM, I suppose it must have a proper connection?
    Re-applied thermal paste - already did it 2 or 3 times
    Re-seated the CPU - already did it 2 or 3 times - particularly scared of this, as it's possible for the CPU to be stuck to the heatsink and all of a sudden fall down onto the socket pins bending them...

    I am seriously out of ideas...
    Don't wanna risk tightening the heatsink screws even more to guarantee a proper contact with the CPU in fear of bending the socket pins. On the other hand, CPUA needed quite some tightening to even recognize all of the RAM, so maybe I am just puss*ing to much regarding tightness. CAN I actually bend the pins under the CPU if it is delidded?!
     
  2. raymanster macrumors 6502

    raymanster

    Joined:
    Feb 13, 2008
    Location:
    UK
    #2
    I had a similar issue with my x5677's just a month ago. On load, boosta and boostb would go very high along with intake and exhaust.

    I re-installed the original CPUs to make sure everything was ok, and it all ran fine for a few days.

    Therefore I re-applied thermal paste on the X5677's. I followed the Apple tech guide regarding screwing down the heat sinks, and then was careful to try and count number of turns i tightened down the heatsinks. This seemed to work and fan no longer go crazy under load.

    IMHO so unless you caked on loads of thermal paste it's probably because your heat sinks are not tightened down enough or just not evenly enough.
     
  3. MentalVizion, Mar 13, 2017
    Last edited: Mar 13, 2017

    MentalVizion thread starter macrumors regular

    MentalVizion

    Joined:
    Oct 30, 2013
    Location:
    Austria
    #3
    I reapplied some new thermal paste again - this time making sure to not use TOO much. Also REALLY tightened the heatsink - same result :(

    Also switched CPU's again. CPU A - which was the one "not working" before, is working perfectly fine in socket B and I get no red light flashes whatsoever. This has to have something to do with either socket A or heatsink A. But I have no clue in the slightest anymore what could cause this problem. No bent pins, heatsink is seated and not damaged either. SIGH.
     
  4. ActionableMango macrumors 604

    ActionableMango

    Joined:
    Sep 21, 2010
    #4
    What are the temps?
     
  5. MentalVizion, Mar 13, 2017
    Last edited: Mar 13, 2017

    MentalVizion thread starter macrumors regular

    MentalVizion

    Joined:
    Oct 30, 2013
    Location:
    Austria
    #5
    I have now put the E5520's back in and sadly ran out of thermal paste because of the constant switching. So can't tell you for certain, bud when idle they seemed pretty normal to me. When running geekbench they'd sometimes peak at around 80-90 (celsius) with the fans spinning up and the red light flashing on the board.
     

Share This Page