Slashdot Mirror

← Back to Stories (view on slashdot.org)

Intel Skylake CPUs Are Warping Under Mounting Pressure From Third-Party Coolers (hothardware.com)

Posted by Soulskill on from the try-cooling-it-with-a-sky-lake-instead dept.

MojoKid writes: It's been discovered that some third-party heat sinks can physically damage Intel's new Skylake CPUs, along with the pins in the accompanying motherboard socket. The problem has prompted at least one cooler maker to change the design of its Socket 1151 heat sinks and it wouldn't be surprising if others soon followed suit. The apparent issue is the substrate Intel used for its Skylake chips. A close-up shot of a Skylake CPU sitting side-by-side with a Broadwell processor (Google translation of German original) shows that the substrate is noticeably thinner on Skylake, and thus prone to bending from the force that some third-party heat sinks exert. Intel has addressed the issue by saying, “The design specifications and guidelines for the 6th Gen Intel Core processor using the LGA 1151 socket are unchanged from previous generations and are available for partners and 3rd party manufacturers. Intel can’t comment on 3rd party designs or their adherence to the recommended design specifications. For questions about a specific cooling product we must defer to the manufacturer.”

17 of 115 comments (clear)

  1. Well, stop requiring such high pressures by kriston · · Score: 2, Interesting

    Well, maybe Intel should stop requiring such high pressures on the heat sink/heat spreader interface. Surely there's a more efficient way to handle cooling. This idiocy started with the Pentium 4 and needs to stop.

    --

    Kriston

    1. Re:Well, stop requiring such high pressures by Anonymous Coward · · Score: 3, Insightful

      Intel still sells stock coolers for all of their CPU's, they just don't ship with them. Intel coolers are more than adequate unless you are going crazy with overclocking. Intel coolers don't bend Intel CPU's.

      Not sure why you are blaming Intel here.

    2. Re:Well, stop requiring such high pressures by PitaBred · · Score: 3, Informative

      They don't. The pressure rating for Skylake is the same as Broadwell, 50psi. Which ensures a good contact. What is likely happening is that some aftermarket coolers used more pressure because they're huge, heavy hunks of metal and needed more force to keep from lifting away, and those worked with the stronger substrate that Broadwell and earlier had, but not with Skylake.

      --
      My blog. Good stuff (when I remember to update it). Read it.
    3. Re:Well, stop requiring such high pressures by Tailhook · · Score: 3, Insightful

      Intel still sells stock coolers for all of their CPU's, they just don't ship with them.

      Huh? The last two I personally purchased (Ivy Bridge i5-3570K and Skylake i7-6700) both provided a heatsink+fan.

      Intel coolers are more than adequate unless you are going crazy with overclocking. Intel coolers don't bend Intel CPU's.

      Both 100% true. The fans on the Intel coolers are quiet under common loads (gaming included) and the temperatures are in spec at all times. The only time I've ever heard the CPU get loud enough to matter was running prime95 to deliberately load the devices.

      I don't know what "high pressures" kriston is talking about. The Intel cooler, particularly on the Skylake, didn't require high pressure to install. Honestly the Skylake case kind of confused me because so little pressure was needed.

      This is about overclockers buying out-of-spec aftermarket stuff. Self inflicted.

      --
      Maw! Fire up the karma burner!
    4. Re:Well, stop requiring such high pressures by GerryGilmore · · Score: 3, Informative

      Actually, non-overclocked Skylakes do ship with a stock cooler. The first Skylake processor released was the 4790K (K=overclocked) and does not ship with its own cooler. Non-Ks do, though.

    5. Re:Well, stop requiring such high pressures by bluefoxlucid · · Score: 4, Informative

      Likely they're intentionally trying to deform the surface to fit, and to compress and thin the thermal paste.

      You want a thin, fully-contacted surface to maximize thermal transfer. Intel likely made the substrate thinner to increase thermal transfer: 1 inch of highly-thermally-conductive silver is a better insulator than 1 micrometer of highly-thermally-conductive silver (hell, a mile of silver wire is a better electrical resistor than a layer of silver foil stuffed between two terminals). Silicone substrate isn't particularly thermally conductive, and thick silicone substrate is even less so.

      Because the CPU surface isn't perfectly flat but *is* flexible, pressure will help conform it to the (also not perfectly flat) heat sink surface. This squeezes some of the thermal paste away, and compresses out voids to increase contact surface area. If you assume more is better, you'll naturally conclude that crushing the CPU into dust will give you better contact.

      If Intel puts out a specification saying 50psi is rated contact pressure for correct heat sink function, you might decide to put 90psi on it. It works. You ship that, Intel releases a CPU that can handle 70psi instead of 140psi, they don't bother telling anyone because the CPU's spec is still 50psi, and shit starts breaking.

      Amusingly, Intel could have glibly put every cooling system manufacture on a list of warranty-terminating equipment ages ago. They could have said, "Hey, we tested all these EXXXTREME COOLING jet fans and they dump 120psi onto our chips like the Hulk putting our balls in a vice. You strap that to the CPU and it fails, we're not responsible." It's fair for Intel to claim that unknown third-party equipment can destroy their hardware; can they quality control third-party equipment? They could, and they could stamp their name on it. That's how motherboards are made, and the board manufacturer is still liable if their board is mis-manufactured (now, if Intel passes the design and the design is flawed, Intel's certification makes Intel liable--Intel is negligent here; if the manufacturing is not within tolerance to produce properly-working equipment and it pumps out flawed boards, it's the manufacturer's fault).

      These things happen. It sucks when they do.

      --
      Support my political activism on Patreon.
    6. Re:Well, stop requiring such high pressures by Anonymous Coward · · Score: 3, Informative

      It sounds a bit like you didn't install the Intel boxed cooler correctly...

    7. Re:Well, stop requiring such high pressures by Anonymous Coward · · Score: 2, Informative

      4790K is not Skylake, but Haswell refresh codename Devil's Canyon. It has a different thermal material between the core and the heat spreader.

    8. Re:Well, stop requiring such high pressures by UnknowingFool · · Score: 4, Informative

      I think the the "K" version of the 6700 is the overclocked version so it does not come with a cooler as most overclockers will buy a 3rd party one. The poster had the non "K" version which is not meant to be overclocked.

      --
      Well, there's spam egg sausage and spam, that's not got much spam in it.
    9. Re:Well, stop requiring such high pressures by sexconker · · Score: 2

      Exactly this. Intel stopped shipping coolers with the K variants a while back because they went straight into the trash.

  2. Warping by rossdee · · Score: 5, Funny

    Great, now all we need to do is build the rest of the starship

    1. Re:Warping by TeknoHog · · Score: 2

      Be sure to use enterprise-grade coolers.

      --
      Escher was the first MC and Giger invented the HR department.
  3. Comment removed by account_deleted · · Score: 3, Insightful

    Comment removed based on user account deletion

  4. It's Intel's fault by MetricT · · Score: 2, Interesting

    This would matter less if Intel would include a usable heatsink with their CPU's. I have worked in high performance computing for over a decade, so putting a heat sink on isn't exactly some exotic task to me, but I couldn't get either of my home OEM Haswell heatsinks to hold onto the motherboard, they would both pop off after the slightest bump. So I *had* to use third-party heatsinks.

    Intel should make backplates with threaded mounts mandatory, and should ensure that their OEM heatsink is capable of actually staying on the motherboard and keeping the CPU from thermally throttling during a Prime95 run. If the user needs a third-party heat sink due to overclocking or unusual case geometry, that's fine, but their OEM heatsink should work properly for 95% of users. But it doesn't.

    Doing the right thing at Intel's scale couldn't cost more than a dollar (a little extra aluminum/steel in the right spots), yet they mysteriously cheap out. Even AMD's stock heatsink is better.

    1. Re:It's Intel's fault by sexconker · · Score: 3, Insightful

      I don't remember which one Haswell was (2 generations ago?) but I fucking HATE the Intel mounting design.
      You have to line up 4 plastic pegs, and then push and twist (to lock) each of them.
      While the pegs have a convenient arrow on them to tell you which direction you need to twist to lock them, they DON'T have an indicator for what the starting position should be. In the unlocked state they're fairly loose and can rotate in the package. I believe the slightly flatter side of the top of the peg should be closest to the heatsink, but I've never truly figured out if that was the case.
      The pegs are spring loaded, and if you push them down too far before locking you'll get it stuck in the mounting hole and you'll have to twist and pull and tug to get it back out, upsetting any of the others you have already mounted. Once you do free it, you have to reset the thing to neutral, not just by rotating it to the starting position, but also by determining if the spring or peg are stuck, and you'll need to pull and twist and tug again to reset that.
      The things are so fucking finicky. Just use a real backplate with screws. Fuck.

    2. Re:It's Intel's fault by Kjella · · Score: 2

      You have to line up 4 plastic pegs, and then push and twist (to lock) each of them. While the pegs have a convenient arrow on them to tell you which direction you need to twist to lock them, they DON'T have an indicator for what the starting position should be.

      In an astonishing leap of logic, if turning the peg in the direction of the arrow leads to the final position then turning it the other way leads to the starting position. The peg only moves about 90 degrees and is flat on two sides so with the fan below you start like |_ and twist it counter-clockwise until it's like _| with the round side out. I guess it's time for the corollary to Murphy's law, even if there's no real way to do it wrong someone will find one.

      --
      Live today, because you never know what tomorrow brings
  5. Re: Will the cooler manufacturers pay? by sexconker · · Score: 5, Insightful

    Look, oh_my_shittynumbers is being an idiot again.

    The question isn't whether or not shit is thinner, it clearly is.

    The question is whether:

    A: Third party heatsinks are exceeding the limit (50 pounds) for mounting pressure.
    B: Intel fucked up and the 50 pound limit needs to be lowered for the thinner shit.
    C: Both A and B.

    My guess is A.
    The heatsink mounting pressure is way over limit for many third-party heatsinks, but the 50 pound limit was conservative for older Intel CPUs and they could handle a lot more. The newer CPUs can almost certainly handle the listed 50 pounds, and more, but not nearly as much as the older CPUs.

    Anyone who has installed a large third-party heatsink knows that the fucking motherboard will flex and bend and you'll put a ton of fucking pressure on it when installing it and trying to get it latched/screwed/etc. properly. After installation you look at it and think "Uh, is this really okay? Should I switch to a horizontal configuration?". But you'll leave it vertical anyway.