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

Intel rumoured to restore soldered heatspreaders in Haswell-E

Discussion in 'Overclocking & Cooling' started by natr0n, Jul 29, 2014.

  1. natr0n

    natr0n

    Joined:
    Jan 29, 2012
    Messages:
    2,055 (1.95/day)
    Thanks Received:
    1,089
  2. a_ump

    a_ump

    Joined:
    Nov 21, 2007
    Messages:
    3,620 (1.40/day)
    Thanks Received:
    376
    Location:
    Smithfield, WV
    this will be great. at least for me i've never bothered to try de-lidding a cpu for better thermals.
     
  3. hat

    hat Maximum Overclocker

    Joined:
    Nov 20, 2006
    Messages:
    16,991 (5.76/day)
    Thanks Received:
    2,085
    Location:
    Ohio
    Let's hope so... the way they've been doing it hasn't been any good for anybody.
     
    Crunching for Team TPU 5 Million points folded for TPU
  4. FR@NK

    FR@NK

    Joined:
    Apr 30, 2006
    Messages:
    576 (0.18/day)
    Thanks Received:
    94
    All the previous -E versions have been soldered so this is nothing new.
     
  5. Mussels

    Mussels Moderprator Staff Member

    Joined:
    Oct 6, 2004
    Messages:
    42,552 (11.41/day)
    Thanks Received:
    9,824
    i havent delidded any CPU's in a long time, they should just sell an enthusiast version without the IHS to begin with, to cater to the extreme OC'ers
     
  6. Kursah

    Kursah

    Joined:
    Oct 15, 2006
    Messages:
    8,028 (2.69/day)
    Thanks Received:
    1,800
    Location:
    Missoula, MT, USA
    +1

    And solder at least the K series since they're unlocked and all and OC-ing is obvious.
     
  7. Mussels

    Mussels Moderprator Staff Member

    Joined:
    Oct 6, 2004
    Messages:
    42,552 (11.41/day)
    Thanks Received:
    9,824

    regular chips: paste
    K chips: soldered
    fancypants stuff like they anniversary edition: no or easily removed IHS (screw it on or something)
     
    Kursah says thanks.
  8. Octopuss

    Octopuss

    Joined:
    Aug 13, 2009
    Messages:
    986 (0.50/day)
    Thanks Received:
    120
    Location:
    Czech republic
    I have to say I kind of enjoyed delidding my 3770K. Even if they use some solder or whatever, it would still be another layer of metal and stuff between the die and cooler.
     
    Sasqui says thanks.
  9. RCoon

    RCoon Gaming Moderator Staff Member

    Joined:
    Apr 19, 2012
    Messages:
    7,969 (8.16/day)
    Thanks Received:
    4,086
    Location:
    Gypsyland, UK
    The real news in that image is that there's 12 cores on that die, and only 8 of them are activated.
     
    Ahhzz, xvi, Shambles1980 and 1 other person say thanks.
  10. LAN_deRf_HA

    LAN_deRf_HA

    Joined:
    Apr 4, 2008
    Messages:
    4,563 (1.86/day)
    Thanks Received:
    954
    Thought we figured out the problem is the gap, not the interface? Maybe the solder makes the gap irrelevant because it's just that much better than TIM?
     
  11. lilhasselhoffer

    lilhasselhoffer

    Joined:
    Apr 2, 2011
    Messages:
    1,680 (1.24/day)
    Thanks Received:
    1,040
    Location:
    East Coast, USA
    It seems like there's still some misunderstanding about thermal transfer.

    Let's look at thermal conductivity figures, measured in Watts per meter per degrees Kelvin (W/(m*K))
    Copper 390
    Air 0.024
    Thermal Paste (low) 0.5
    Thermal Paste (high) 10

    The conclusion then is that a brick on copper a meter high will provide the same heat transfer resistance as a high quality thermal paste brick 2.56 cm high. Assuming the thermal paste is of low quality, you're looking at a millimeter of paste. The astute observer will also point out that the conductivity of air means that a hand full of micrometer of air provide the same heat transfer as that meter of copper. We'll address this shortly.

    These figures make a few assumptions. First, they represent simple conductive heating, along a single axis. Next, they assume a mechanical bond between surfaces. Finally, the assumption is a uniform source of thermal energy.

    We use thermal paste between heat spreaders and heat sinks because that second assumption breaks down. The difference between the surfaces mean that you've effectively only got heat transfer at a few points, and thus heat transfer is hindered. Even a relatively poor conductor of heat will dramatically boost the surface area for heat transfer, so thermal paste is a good idea. A better idea would be to lap both the heat spreader and heat sink, then melt a metal between them to form a complete mechanical bond (solder them). As heat sinks vary greatly, this isn't a practical. Even with a known die and heat spreader, the process of soldering is much more expensive than a thermal paste. Intel saved money by having a heat paste, but never decreased processor price. They are now going to charge more for heat paste using processors, because it's a "premium" addition.


    Sorry, but none of this is news. It's Intel saving money, not passing on the savings, then charging more for what they used to do by default. I consider it an insult. Saying that soldering should demand a premium cost point is like saying that a brake pedal should be an add-on in a car. You could technically make the machine brake at all times the gas pedal is released, but that would produce a car that is far less effective for a minor decrease in component cost.
     
    Last edited: Jul 29, 2014
    RCoon says thanks.
  12. Sasqui

    Sasqui

    Joined:
    Dec 6, 2005
    Messages:
    7,800 (2.36/day)
    Thanks Received:
    1,523
    Location:
    Manchester, NH
    I'm with you there, it's like gaming the system ;) The difference between stock TIM and delid/new TIM was 10c-20c.

    I found that the difference between delidding, and putting new TIM on vs. going naked die was just about null. But I enjoyed the nakedness none the less.
     
  13. Arvutistuudio

    Arvutistuudio New Member

    Joined:
    Jul 29, 2014
    Messages:
    3 (0.02/day)
    Thanks Received:
    0
    It is really nice! But will it blend?
     
  14. Aquinus

    Aquinus Resident Wat-man

    Joined:
    Jan 28, 2012
    Messages:
    6,857 (6.48/day)
    Thanks Received:
    2,464
    Location:
    Concord, NH
    What are we talking about here? SB-E and IVB-E both had soldered IHS', why would Haswell-E be any different? It would be news is Broadwell will have a soldered IHS or if Haswell-E didn't, but this was expected as far as I'm concerned.
     
    lilhasselhoffer and TRWOV say thanks.
  15. a_ump

    a_ump

    Joined:
    Nov 21, 2007
    Messages:
    3,620 (1.40/day)
    Thanks Received:
    376
    Location:
    Smithfield, WV
    Hey that's kind of what my local Food Lion did with their nesquik! used to buy tons of bottles for resale for like $1.25ea, then they jumped the price to a whopping $1.69ea or 3/$5. Then a week later they stated "Sale on Nesquik!!!" at $1.39ea.

    These companies change something just long enough for us to accept it as the norm, then go back(or close in my case) to their previous superior offer at a cost to us relative to the "new" norm. Trixy basterds
     
  16. lilhasselhoffer

    lilhasselhoffer

    Joined:
    Apr 2, 2011
    Messages:
    1,680 (1.24/day)
    Thanks Received:
    1,040
    Location:
    East Coast, USA
    Nesquick is an interesting point. If you want to have some fun start tracking the price of luxury goods (TVs, phones, etc...) in October. Wait until November, just before Black Friday. You'd be surprised how many items ratchet up their pricing, then go on sale for the price from October.


    I'm not sure about other responders, but I was responding to the thermal paste situation in general. It's no surprise that Intel is doing it where they are doing it. Enthusiast processors, whether unique or just binned versions of their standard CPUs, are getting soldered lids as a matter of course. We are paying for this because the processor is marketed to enthusiasts, while the "mainstream" gets cheaper paste without reaping the cost savings.

    I agree 100% that the enthusiast series will be soldered for the foreseeable future, as the thermal loading demands it. My apologies for offering what seems like a stupid response in retrospect.
     
  17. theoneandonlymrk

    theoneandonlymrk

    Joined:
    Mar 10, 2010
    Messages:
    3,414 (1.96/day)
    Thanks Received:
    573
    Location:
    Manchester uk
    How are they calling it soldered when its clearly an epoxy type of Tim.
    Thats not soldering. .
     
    10 Million points folded for TPU
  18. Shambles1980

    Joined:
    May 3, 2014
    Messages:
    540 (2.33/day)
    Thanks Received:
    102
    could not agree more. i think all "k" versions should come with no IHS. and only need the metal part for the socket retention, with the die itself exposed
     
  19. Tallencor

    Tallencor

    Joined:
    Nov 25, 2013
    Messages:
    565 (1.45/day)
    Thanks Received:
    654
    Location:
    N.S. Can.
    So should I delid my 3770k? hardly looks easy. I am not upgrading for years yet and it would be nice to see lower than 85 @ 4.3.
     
    Crunching for Team TPU
  20. Shambles1980

    Joined:
    May 3, 2014
    Messages:
    540 (2.33/day)
    Thanks Received:
    102
    85C ??
    pretty sure a 3770k's tjmax is lower than that, (or im thinking tcase, but pretty sure its a bit hotter than id be willing to run at there)
     
    Last edited: Jul 30, 2014
  21. LAN_deRf_HA

    LAN_deRf_HA

    Joined:
    Apr 4, 2008
    Messages:
    4,563 (1.86/day)
    Thanks Received:
    954
    A lot of people stress up to 90.
     
  22. MxPhenom 216

    MxPhenom 216 Corsair Fanboy

    Joined:
    Aug 31, 2010
    Messages:
    10,210 (6.49/day)
    Thanks Received:
    2,383
    Location:
    Seattle, WA
    they will, and have always been running solder on their enthusiast platform. (lga2011) Nothing is going to change. Its only their mainstream platform (lga1150) that went to TIM.
     
  23. Octopuss

    Octopuss

    Joined:
    Aug 13, 2009
    Messages:
    986 (0.50/day)
    Thanks Received:
    120
    Location:
    Czech republic
    Definitely. I am at 70°C running at 4.4GHz. And that's with Hyper 212 EVO, which is hardly a huge performance cooler.
     
    Tallencor says thanks.
  24. MxPhenom 216

    MxPhenom 216 Corsair Fanboy

    Joined:
    Aug 31, 2010
    Messages:
    10,210 (6.49/day)
    Thanks Received:
    2,383
    Location:
    Seattle, WA
    Its really easy, just use hammer and vice method. Takes all of 5 minutes.

     
    Tallencor says thanks.
  25. Tallencor

    Tallencor

    Joined:
    Nov 25, 2013
    Messages:
    565 (1.45/day)
    Thanks Received:
    654
    Location:
    N.S. Can.
    Cool. Thanks. I'm on it. I wont hijack so maybe I will start a new thread with my results.
     
    MxPhenom 216 says thanks.
    Crunching for Team TPU

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

Share This Page