Massive Jump in CPU Temperature

Discussion in 'OT Technology' started by deusexaethera, Mar 27, 2006.

  1. deusexaethera

    deusexaethera OT Supporter

    Joined:
    Jan 27, 2005
    Messages:
    19,712
    Likes Received:
    0
    I recently upgraded my father's music recording computer from a 2.66GHz P4 to a 3.06GHz P4. Both chips have a 533MHz bus speed. The clock speed increased by 400MHz, and the idle temperature increased by 10°C! What's worse, the CPU now reaches 77°C under heavy load, and it NEVER used to go above 60°C under ANY load before the upgrade!

    I'm using a heatsink with a huge copper base, silver thermal paste, and a 92cm fan to cool it with. The power supply is P4-compatible and puts out 450W. It powers the motherboard and CPU, two hard drives, two CD drives, a mouse and keyboard, and a couple of USB hardlock keys.

    Is there a justifiable reason for this huge increase in temperature? Are the 3.06GHz P4's known to run really friggin hot or something?
     
  2. P07r0457

    P07r0457 New Member

    Joined:
    Sep 20, 2004
    Messages:
    28,491
    Likes Received:
    0
    Location:
    Southern Oregon
    possible mis-application of thermal paste?
     
  3. deusexaethera

    deusexaethera OT Supporter

    Joined:
    Jan 27, 2005
    Messages:
    19,712
    Likes Received:
    0
    I'm always very careful with thermal paste. I squirt some in the middle of the CPU die, and I smear it into a thin layer over the whole die using my finger. Then I do the same to the middle of the heatsink where the CPU will touch it.

    Shit's too expensive to use wrong.

    EDIT: Besides, I touched the base of the heatsink when the computer was on and it was plenty hot. If I messed up the heatsink installation and it's not conducting heat properly, then the CPU itself must be hot as a motherfucker.
     
  4. P07r0457

    P07r0457 New Member

    Joined:
    Sep 20, 2004
    Messages:
    28,491
    Likes Received:
    0
    Location:
    Southern Oregon
    well, you definitely applied heatsink paste incorrectly. Not saying that's the cause, 100%, but you did it way wrong.

    First of all, NEVER use your finger. Oils from your skin == bad. Secondly, Only apply to ONE surface, never both.

    Two applications will act as an insulator.

    I suggest you clean it up with q-tips and rubbing alcohol, and re-apply properly.
     
  5. mace

    mace i don't read

    Joined:
    Jun 6, 2004
    Messages:
    12,567
    Likes Received:
    0
    Location:
    Texas
    make sure you don't use the 70% rubbing alcohol as it has impurities in it. get at least 90%
     
  6. Peyomp

    Peyomp New Member

    Joined:
    Jan 11, 2002
    Messages:
    14,017
    Likes Received:
    0
    More megahertz = more temp?
     
  7. Shibboleth

    Shibboleth teh mad Plato skillz

    Joined:
    Sep 29, 2004
    Messages:
    1,769
    Likes Received:
    0
    cpu was the only thing that changed?

    also how about more info, like cpu and MB model number, or at least cpu core type and mb chipset
     
  8. Peyomp

    Peyomp New Member

    Joined:
    Jan 11, 2002
    Messages:
    14,017
    Likes Received:
    0
    Shouldn't a processor with a higher megahertz run hotter?
     
  9. Shibboleth

    Shibboleth teh mad Plato skillz

    Joined:
    Sep 29, 2004
    Messages:
    1,769
    Likes Received:
    0
    I guess, but the two aren't directly related. It's the power consumption. But I think jollyogre is right, and the problem will be solved with a new application of thermal paste
     
  10. Achance007

    Achance007 Active Member

    Joined:
    Oct 12, 2000
    Messages:
    16,128
    Likes Received:
    4
    Location:
    New Castle, DE
    yeah use that amount and spread it around with a razor blade.
     
  11. monolith

    monolith My soul grows ever weary and the end is ever near. OT Supporter

    Joined:
    May 9, 2004
    Messages:
    32,539
    Likes Received:
    237
    Location:
    Southern California
    :werd:

    A buddy of mine spreads it with a credit card.

    :ugh:

    Just put a small dab in the middle of the CPU and clamp the HS to it. That's it.
     
  12. EvilSS

    EvilSS New Member

    Joined:
    Jun 11, 2003
    Messages:
    5,104
    Likes Received:
    0
    Location:
    STL
    Could be a core difference too. Different cores, even in the same processor family, can have vastly different heat output.

    Most likely though Jolly is right and there is too much paste.
     
  13. P07r0457

    P07r0457 New Member

    Joined:
    Sep 20, 2004
    Messages:
    28,491
    Likes Received:
    0
    Location:
    Southern Oregon
    Incorrect! Many thermal paste manufacturers recommend only leaving a dot in the middle, but this is after much research and experience in how stupid their customers are! If you do it *properly*, nothing will beat a nice, evenly spread out, thin layer of thermal compound.
     
  14. deusexaethera

    deusexaethera OT Supporter

    Joined:
    Jan 27, 2005
    Messages:
    19,712
    Likes Received:
    0
    Funny you should mention it, when I used to do it your way I got higher temperature readings on my other computers. I came up with the way I do it now through trial and error. Modern heatsinks provide more than enough pressure to squeeze out excess thermal paste, and I always wash my hands immediately before applying the paste. I know the oil = bad rule.
     
  15. deusexaethera

    deusexaethera OT Supporter

    Joined:
    Jan 27, 2005
    Messages:
    19,712
    Likes Received:
    0
    Yeah, obviously, but 10C for 400MHz? That seems like a lot.
     
  16. P07r0457

    P07r0457 New Member

    Joined:
    Sep 20, 2004
    Messages:
    28,491
    Likes Received:
    0
    Location:
    Southern Oregon
    refer to explination as to why many thermal paste manufacturers recommend the dot...

    However, if you look at the really high-performance stuff, installed by the pros, they all spread it out. Simple physics is that, even though the pressure of the heatsink is enough to spread a dab sufficiently to cool effectively, it will still be an uneven application, with more in the center than on the ends... And the heatspreader still will not have 100% contact.

    I'm not arguing that your way isn't easier -- it definitely is. But let me tell you that it makes a few degrees difference to take the time to do it properly.
     
  17. deusexaethera

    deusexaethera OT Supporter

    Joined:
    Jan 27, 2005
    Messages:
    19,712
    Likes Received:
    0
    As I said, I DO spread it out. And as I said, I wash my hands immediately before I apply the paste so I won't get any skin oil into the paste. I'm not saying that I'm perfect, and that there's no way I could have screwed up the application, but is there anything else that could cause the heat problem?

    Technically it isn't even a problem, just an annoyance. It hasn't affected the performance of the computer at all. Though I am curious to know if any other 3.06GHz P4 users (there is only one model of chip with this exact speed) have noticed high temperatures.
     
  18. Supernaut

    Supernaut New Member

    Joined:
    May 20, 2003
    Messages:
    8,047
    Likes Received:
    0
    Prescott core? Those are known to run hot, not necessarily 77 degrees though. My Prescott 3.0 idled at 46 degrees water cooled. That same kit, with a DangerDen 939 waterblock instead of 478, keeps my AMD 3800X2 idling at 26 degrees.
     
  19. mace

    mace i don't read

    Joined:
    Jun 6, 2004
    Messages:
    12,567
    Likes Received:
    0
    Location:
    Texas
    My AS5 works so good I don't even use any case fans :mamoru: Only because my case fans are loud and I can't stand the noise, but my temps are good... even when playing a game.

    I did the dot method as the instructions recommend.

     
  20. Peyomp

    Peyomp New Member

    Joined:
    Jan 11, 2002
    Messages:
    14,017
    Likes Received:
    0
    I think you guys are overalayzing this.
     
  21. deusexaethera

    deusexaethera OT Supporter

    Joined:
    Jan 27, 2005
    Messages:
    19,712
    Likes Received:
    0
    It's my thread and I say we're not overanalysing. Applying heat goop is practically an art form.

    - - -

    Anyway, here's what was wrong. The layer of heat goop WAS too thick, but not for the obvious reason. It turns out, upon re-examination of the CPU, that the edge of the heat spreader had a lip on it that was holding the heatink away from touching the center of the chip properly. So I sanded it flat with some automotive emory paper and polished it with my trusty Dremel.

    The idle temperature has dropped from 56C to 45C, and I'm fine with that, considering that the computer has to be so quiet that the heatsink fan is practically for-decoration-only. Interestingly, the one program that made the temperature spike to 78C still makes the temp. hit 72C, and the weird part is the CPU barely registers any usage at all. Who knows; those boutique digital recording tools are so finicky.
     
  22. deusexaethera

    deusexaethera OT Supporter

    Joined:
    Jan 27, 2005
    Messages:
    19,712
    Likes Received:
    0
    I'm not sure if it's a Prescott or Northwood core; I want to say it's Northwood, because it's not a socket 775 and the bus speed isn't 800MHz, but it IS hyperthreaded and I don't know if Northwood ever supported that.
     
  23. P07r0457

    P07r0457 New Member

    Joined:
    Sep 20, 2004
    Messages:
    28,491
    Likes Received:
    0
    Location:
    Southern Oregon
    Northwood certainly did have HT-enabled models.
     
  24. deusexaethera

    deusexaethera OT Supporter

    Joined:
    Jan 27, 2005
    Messages:
    19,712
    Likes Received:
    0
    Then it's probably a Northwood. The packing sticker on the box says 2003.

    EDIT: On a related note, it's nice that a three-year old CPU is still good enough to be current, for a change. Hasn't been that way since the 80's. Of course, the 1.5GB of Rambus certainly helps the speed; DDR may claim to be as fast as Rambus nowadays, but I've never seen a computer load applications as fast as my dad's machine can.
     
  25. P07r0457

    P07r0457 New Member

    Joined:
    Sep 20, 2004
    Messages:
    28,491
    Likes Received:
    0
    Location:
    Southern Oregon
    sorry, but a 533MHz FSB RDRAM machine just isn't gonna compete with a 1066MHz FSB machine with DDR2.
     

Share This Page