1. Welcome to TechPowerUp Forums, Guest! Please check out our forum guidelines for info related to our community.

My 4770k delid video (IHS removal)

Discussion in 'Overclocking & Cooling' started by boulard83, Jun 30, 2013.

  1. Ketxxx

    Ketxxx Heedless Psychic

    Joined:
    Mar 4, 2006
    Messages:
    11,510 (2.98/day)
    Thanks Received:
    570
    Location:
    Kingdom of gods
    1. Just because something carries the word "official" it doesn't make it true. I wasn't (and still aren't) talking about which platform gives you the most performance, I covered that. I am talking about which platforms is the most technologically advanced, again thats Z77 and Z87, not skt 2011.

    2. Intel still sell EE CPUs with a massive 600-1000 premium so your point there is moot. The K series just allowed intel to bat the ball closer to what AMD do, but my point here, and always has been (probably my fault for not clarifying better I get wrapped up in details too much for my own good sometimes) that the K series at a £44 premium is still too much when intel don't even offer a enthusiast friendly warranty with the K series CPUs, by which I mean your warranty isn't void if you OC a K series CPU. Last I read even with K CPUs if you OC them intel say your warranty is void.
     
    10 Year Member at TPU
  2. newtekie1

    newtekie1 Semi-Retired Folder

    Joined:
    Nov 22, 2005
    Messages:
    23,024 (5.81/day)
    Thanks Received:
    8,981
    1. See my post, you're wrong.

    2. Sure, for a small cost it could run cooler, a small cost across millions of products that adds up to a huge cost, that has no affect on the life or intended function of the product.

    3. And yet I haven't had a single processor with TIM cause me problems, and neither have millions of other people. Odd...

    1. Just because it is more technologically advanced does mean it provides the best performance. Enthusiasts are after performance, overclockers are after performance, Socket 2011 is for enthusiasts socket 115X is for mainstream users.

    2. I don't see how the fact that Intel still selling extreme edition processor make the point that an unlocked multiplier very recently carried a $7-800 price premium moot. And the warranty doesn't matter, again an real enthusiast knows that. We didn't have warranties for overclocking back then, even with the $1000 unlocked Extreme processors. No enthusiast would be bitching about a warranty... "OMG, the warranty should cover overclocking!" - Said no enthusiast ever.
     
    Last edited: Jun 30, 2013
    The Von Matrices says thanks.
    10 Year Member at TPU Crunching for Team TPU
  3. The Von Matrices

    The Von Matrices

    Joined:
    Dec 16, 2010
    Messages:
    1,455 (0.69/day)
    Thanks Received:
    529
    I really don't think this is a manufacturing a defect; this seems to be user error. The chances of having the same person with multiple failures (with proper usage) is extremely slim or otherwise there would be many more posts of this topic.
     
    Last edited: Jun 30, 2013
    Naito says thanks.
  4. tigger

    tigger I'm the only one

    Joined:
    Mar 20, 2006
    Messages:
    10,211 (2.66/day)
    Thanks Received:
    1,423
    I would never de-lid a cpu I couldn't afford to replace, I guess that is the difference twixt a enthusiast and a user. The enthusiast will do it knowing they will just buy another one if they break it, were as a normal bod wouldn't dream of it.

    Believe me If I was loaded I would just buy 2 cpu's and feck about with the first one, knowing full well if I bust it, I can just pop the other one in.

    Does anyone know how many cpu's them ln2 nutters break attempting their records?

    Lets face it, a cpu used under normal running guidelines/usage will probably last for years, we all have a 5/10yr old cpu lying around that still works fine.
     
    10 Year Member at TPU
  5. The Von Matrices

    The Von Matrices

    Joined:
    Dec 16, 2010
    Messages:
    1,455 (0.69/day)
    Thanks Received:
    529
    You make a good point. He didn't ever define his overclocks that killed the CPUs. Maybe he's running them at >1.5V. That would definitely cause quick failures.
     
  6. Ketxxx

    Ketxxx Heedless Psychic

    Joined:
    Mar 4, 2006
    Messages:
    11,510 (2.98/day)
    Thanks Received:
    570
    Location:
    Kingdom of gods
    I have run all of my CPUs in a well ventilated NZXT Apollo using either a Coolermaster Hyper 212 Plus or a CoolIT ECO unit. Cooling has not been the issue here at all - purely related to the TIM used under the IHS. As for my friends, they don't even OC just buy decent grade gear and I've still had to end up popping the IHS on their CPUs at a period of anywhere between 1 - 3 years. Thats not coincidence.
     
    10 Year Member at TPU
  7. Ketxxx

    Ketxxx Heedless Psychic

    Joined:
    Mar 4, 2006
    Messages:
    11,510 (2.98/day)
    Thanks Received:
    570
    Location:
    Kingdom of gods
    If by "he" you mean me, my personal (and only actual death of a CPU) was my 2500k. The voltages run through that were never extreme, about 1.38v for 4.5GHz. Its all good speculating, but you could of just asked :p
     
    10 Year Member at TPU
  8. tigger

    tigger I'm the only one

    Joined:
    Mar 20, 2006
    Messages:
    10,211 (2.66/day)
    Thanks Received:
    1,423
    I've run a e6300 at 1.57v on air and it survived to tell the tale. :D
     
    10 Year Member at TPU
  9. The Von Matrices

    The Von Matrices

    Joined:
    Dec 16, 2010
    Messages:
    1,455 (0.69/day)
    Thanks Received:
    529
    Are you sure that you didn't just need to replace the TIM between the processor and the heatsink? That would reduce temperatures and cheap thermal paste has been known to dry out. Thermal paste under the IHS on the other hand... I've never heard of anyone needing to replace it before your posts. And the Phenom II you delidded - I thought that all Phenom II's were soldered.
     
  10. The Von Matrices

    The Von Matrices

    Joined:
    Dec 16, 2010
    Messages:
    1,455 (0.69/day)
    Thanks Received:
    529
    That's a 65nm processor though; the 32nm/22nm processors have been known to have durability problems running 24/7 at 1.4V at higher temperatures.
     
  11. Ketxxx

    Ketxxx Heedless Psychic

    Joined:
    Mar 4, 2006
    Messages:
    11,510 (2.98/day)
    Thanks Received:
    570
    Location:
    Kingdom of gods
    That 2500k had AS Ceramique sitting between it and the cooler, so yes, it was good TIM. I dunno, I pop IHS' for fun, serve as technician and direct line to Asrock if anyone gets issues with their asrock mainboard, make custom VRM heatsinks and a multitude of other things and you thought I wouldn't use / know good TIMs from bad? I'll let you off as I haven't been around here until recently for some years :p As for the Phenom, I dunno if they were supposed to be soldered or not but mine wasn't I didn't even bother checking I saw the symptoms and instantly knew it must be TIM between the core and IHS, unsurprisingly it was. Sad truth is intel / AMD don't even use good TIM, if they did temps would be quite a bit better thats why so many people record such huge temp drops when popping IHS'. 10c or more temp drop doesn't happen just because of a 0.006mm gap made by the IHS sealant.
     
    Last edited: Jun 30, 2013
    10 Year Member at TPU
  12. The Von Matrices

    The Von Matrices

    Joined:
    Dec 16, 2010
    Messages:
    1,455 (0.69/day)
    Thanks Received:
    529
    Please do not read my posts as if I am insulting you. I am solely curious as to why you are having bad experiences with multiple processors while there are no other cases on these forums attributing processor failure to the presence of paste versus solder between the CPU and IHS.

    From a scientific point of view I would like to know your failure rate - how many processors do you test that versus the number of failures. Also, I am curious about other confounding variables. For example, if you maintained at least one other common component in the system (presumably the power supply) then that could be causing all the failures. Or maybe the failures even can come down to static electricity if you live in an especially dry place.

    I'm also there are no comparison of the same die soldered versus using TIM paste having different failure rates. Maybe this failure rate is due to the design of the core rather than the TIM. I don't doubt that solder is better for reliability than TIM paste, but I do doubt that the difference is statistically significant.

    Also, the reduction in TIM thickness by removing the sealant during delidding is much greater than the 0.006mm you stated. It's more like 0.09mm, which is significant. See http://forums.anandtech.com/showpost.php?p=34053183&postcount=570.

    Edit: Also see http://forums.anandtech.com/showpost.php?p=34059601&postcount=579 showing that the IHS itself doesn't provide a large thermal barrier; the thickness of the TIM is the largest variable in processor temperature.
     
    Last edited: Jun 30, 2013
  13. Naito

    Naito

    Joined:
    Oct 10, 2009
    Messages:
    688 (0.27/day)
    Thanks Received:
    352
    Try older :p I booted up my 1988 Amstrad PC1640 HD 20, yesterday. Still works fine! :cool: Ah! Seeing that GEM GUI brought back memories. My very first PC.
     
    tigger says thanks.
  14. boulard83

    Joined:
    Jan 28, 2009
    Messages:
    408 (0.15/day)
    Thanks Received:
    65
    Well my thread have gone to another place ! I like it ;)

    I also think that K series proc should be soldered ... They are sold as "overclocking" porcessor after all !
     
  15. newtekie1

    newtekie1 Semi-Retired Folder

    Joined:
    Nov 22, 2005
    Messages:
    23,024 (5.81/day)
    Thanks Received:
    8,981
    And they overclock just fine, not great, but good enough.
     
    10 Year Member at TPU Crunching for Team TPU
  16. 15th Warlock

    15th Warlock

    Joined:
    Aug 16, 2004
    Messages:
    3,002 (0.68/day)
    Thanks Received:
    1,580
    Location:
    Visalia, CA
    Isn't this a little extreme? I still have an original Pentium 166MHz OC'd to 200MHz that's running in my father's office, that little fella has been running for a little over 15 years without a hitch.

    In all my years building PCs I've never seen a single CPU fail (and I began over 20 years ago), and I started OCing my procs since that Pentium I referred to above.

    Now, I'm not arguing that Intel is being cheap here by using bad TIM for their latest procs, cause any enthusiast can see that, however, saying that they purposely design their procs to fail in an X amount of time, forcing ppl to break their warranties by de-liding their processors is bordering on paranoia.

    Just think about that statement for a moment, out of millions and millions of processors Intel sells a year, how many do you think will get to be OC'd to over 5GHz? 10%? 1%? I bet you it's not even close to 0.0001%

    Now you're saying that all those millions of processors are going to fail in 5 years, give or take? If that were the case, Intel would be facing a lawsuit of biblical proportions down the road, on a global scale, mind you.

    Do you honestly think Intel caters to us enthusiasts? Think again, that is obviously not the case. Are they being greedy by saving a few cents in TIM? You bet they are, but they are not stupid either... unfortunately that's how capitalism works, and all corporations are out to make a profit, and if Intel can save $0.03 on every chip sold, and alienate a few enthusiasts by doing so, of course the choice is obvious.

    Truth is enthusiasts, in Intel's eyes are a dying breed, and Broadwell may not even be sold as a stand alone processor anymore, and unless AMD don't step up to the plate, our days are counted indeed :(
     
    Last edited: Jun 30, 2013
    10 Year Member at TPU
  17. MxPhenom 216

    MxPhenom 216 Corsair Fanboy

    Joined:
    Aug 31, 2010
    Messages:
    11,794 (5.32/day)
    Thanks Received:
    3,533
    Location:
    Seattle, WA
    Sounds like neither of you guys truly know why they used TIM on Ivy and Haswell.
     
  18. radrok

    radrok

    Joined:
    Oct 26, 2011
    Messages:
    3,099 (1.72/day)
    Thanks Received:
    852
    Well its not like there are many reasonable explanations.

    Either they use solder or TIM based on the chip's TDP or Ivy and Haswell die is too small to apply a fluxless solder :toast:
     
  19. boulard83

    Joined:
    Jan 28, 2009
    Messages:
    408 (0.15/day)
    Thanks Received:
    65
    +1, the Die is so small, ivy/haswell dont have that much space to dissipate heat.
     
  20. D007

    D007

    Joined:
    Mar 7, 2007
    Messages:
    3,736 (1.07/day)
    Thanks Received:
    662
    Location:
    Pompano beach, Florida
    Yea, a bit silly to start going all conspiracy theory, about something that makes total sense..
    so after u take the lid off, you put new tim on and put the lid back on?
     
  21. tigger

    tigger I'm the only one

    Joined:
    Mar 20, 2006
    Messages:
    10,211 (2.66/day)
    Thanks Received:
    1,423
    you put the chip in the socket with the ihs on top and close the gate, I think some people use some sort of gunk to stick it back on.
     
    10 Year Member at TPU
  22. erocker

    erocker Super Moderator Staff Member

    Joined:
    Jul 19, 2006
    Messages:
    41,500 (11.15/day)
    Thanks Received:
    16,884
    Most of the temperature issues are a result of too much silicon "glue" that keeps the IHS attached to the PCB, not the TIM itself, though higher quality TIM will yield a bit better result.
     
    boulard83 says thanks.
    10 Year Member at TPU
  23. boulard83

    Joined:
    Jan 28, 2009
    Messages:
    408 (0.15/day)
    Thanks Received:
    65
    +1

    You need to remova ALL the black glue, this is usually the main issue.

    I will do further testing with a variety of paste under the IHS later.
     
  24. niko084

    niko084

    Joined:
    Dec 5, 2006
    Messages:
    7,637 (2.13/day)
    Thanks Received:
    729
    Nice video, we got lazy and didn't go through the whole process to make one.

    We used the liquid pro TIM, stuff is kinda cool, little bit of a pain to work with.
    I'm still not sure we have a good contact and are still considering lapping the cpu.

    Out temp difference full load 1.25v 4.2ghz had a 16c difference.

    This is on a 360 rad 30dpi, heatkiller block, d5 pump and a big dual bay res. I'm under the impression there is no heatsoak in this system beyond the pump heating up.
     
  25. boulard83

    Joined:
    Jan 28, 2009
    Messages:
    408 (0.15/day)
    Thanks Received:
    65
    Thanks for sharing your experience.

    I will also take a look at the IHS and lap it if needed, i have all the sandpaper need at home. I also borowed my GF nail vernish for the CPU ;)
     

Currently Active Users Viewing This Thread: 1 (0 members and 1 guest)