Skylake CPU's "extra" susceptible to damage??

[throwa]

I've been following the story over at Tomshardware about Skylake CPU's allegedly being damaged by big/bulky heavy aftermarket coolers.

I was just wondering, is this all sensationalism or is there some truth to it? Have you guys noticed any damage on your Skylake chip?

Specifically I was curious about big/bulky coolers such as Cooler Master Hyper Evo and Noctua coolers.

Thanks for clearing this up, I was considering upgrading to Skylake, but saw the Tomshardware story and instantly got "cold feet". I'm holding off for now until this is either confirmed or denied.

 

[master_shake_]

apparently some people can't handle the application of a heatsink.

they just tighten the crap out of the hold down screws and deforms them

 

[videogames101]

the pcb is thinner, yes

don't literally sheer the mounting bolts and you'll be fine

 

[mysticjbyrd]

I thought this story couldn't be replicated?

 

[LTC8K6]

All specs are the same as Haswell. Skylake socket and CPU pcb strength is the same as Haswell. There is no reason for any cold feet.

 

[throwa]

apparently some people can't handle the application of a heatsink.

they just tighten the crap out of the hold down screws and deforms them

I see.

So i take it this is mostly a beginner/unskilled/n00b internet complaint then, seems to make sense.

Maybe people are just mis-reading the instruction manuals, almost all installation manuals say "secure screws TIGHTLY" or have language to that effect..... maybe some beginner people over-do it and think "really REALLY tight".

 

[bystander36]

There is more to it than just over tightening, there is also transportation problems. If you drop the PC, the weight of the big coolers can cause damage.

 

[ehume]

Some heatsinks like Noctua's have springs that apply the specified pressure, along with tensioning screws with positive stops, where you can't overtighten them. On another heatsink, you can go too far and the instructions say "don't overtighten." How tight is too tight? Now Intel makes a chip where you really can overtighten, and guess what?

 

[Zodiark1593]

Basically the usual "use some discretion" applies when it comes to oversized heatsinks. Considering video card sizes as well. I'd have figured most gamers would take some care in making sure nothing will get knocked around too much in transport. Oh well.

 

[frozentundra123456]

There is already a thread about this. motherboard issue

 

[moonbogg]

Yes. These are considered throwaway CPU's.

 

[VirtualLarry]

All specs are the same as Haswell. Skylake socket and CPU pcb strength in the same as Haswell. There is no reason for any cold feet.

I don't think that's true. SKL desktop CPUs have a thinner substrate. The spec may not have changed, but the Haswell substrate strength may have been under-specified. Simply given their physical characteristics, unless Intel has radically changed the material used for the substrate, then SKL's is definitely thinner, and thus, weaker, in actual practice.

 

[NTMBK]

The real reason for the Skylake CPU shortage, they're too delicate and keep arriving at shops broken, like a bag of factory reject biscuits :awe:

 

[LTC8K6]

I don't think that's true. SKL desktop CPUs have a thinner substrate. The spec may not have changed, but the Haswell substrate strength may have been under-specified. Simply given their physical characteristics, unless Intel has radically changed the material used for the substrate, then SKL's is definitely thinner, and thus, weaker, in actual practice.

Intel has already released a statement saying that while the pcb is thinner, it is the same strength. It's covered in the other thread.

The idea that it has to be weaker because it's thinner, is false.

 

[Erenhardt]

Intel has already released a statement saying that while the pcb is thinner, it is the same strength. It's covered in the other thread.

The idea that it has to be weaker because it's thinner, is false.

No. Both are rated at 50lb of static force. That doesn't mean they have the same strength. One can be 100lb, the other 50, and their answer would still be true. They never said both have the same critical force.

The thiner it is, the lesser force perpendicular to cross-section surface is. With bending, one side of cross-section is stretched, one is pressured - tension distribution in simple bending. To have a static momentum that will prevent 'overhang' from bending you need enough force on large enough arm (eccentricity) to compensate the momentum from the reaction force from the socket.

So, thiner pcb ont only deacreses the area of tension, but also a distance from the neutral axis to net force from tension.

There is a unit that describes resistance to bending for different cross-section shapes. For rectangles it is Wx=(b*h^3)/12
Notice the high - tightness is in power of 3.

I would really like to see their numbers, or tests.

 

[LTC8K6]

All covered in the original thread...

Not sure why it's still being discussed, really.

http://forums.anandtech.com/showthread.php?t=2456465

 

[LTC8K6]

http://hexus.net/media/uploaded/2015/12/1e9a6224-119c-4a19-a058-3f1497e25d63.jpg

If you do that to your CPU socket, it's not Intel's fault.

 

[Erenhardt]

How did you rule out the bent CPU pcb?

CPU cooler is in contact with motherboard only in the mounting spots, which are outside of socket.
Any damage to the socket is caused by the CPU - be it directly or indirectly.

 

[ShintaiDK]

If you get socket damage, it doesn't matter what CPU you used.

People have bent motherboards, CPUs, and sockets. Chipped dies. Cracked boards, CPUs and sockets etc before due to over tension.

 

[LTC8K6]

How did you rule out the bent CPU pcb?

CPU cooler is in contact with motherboard only in the mounting spots, which are outside of socket.
Any damage to the socket is caused by the CPU - be it directly or indirectly.

Remind me not to ask you for computer build advice.

And for the last time, we already have a thread on the subject where all of this has already been discussed.

 

[aigomorla]

If you get socket damage, it doesn't matter what CPU you used.

torque can also do that...

if u have a heavy heat sink where the weight is at the end, and it torques down, over time it will bend the socket.

Now i just need to see the underside of the board to see if the retension backplate is bent as well to rule anything out.

But technically it shouldn't happen like that, unless something is seriously wrong.

Lastly, heat sinks are no longer flat anymore.
They have a slight bow to apply pressure in the middle to warp the IHS and allow a better mount with IHS + Die.

This has been seen many many times on waterblocks, and infact now all waterblocks have a slight bow.

Any damage to the socket is caused by the CPU - be it directly or indirectly.

this is exactly why we need to see the underside of the board.
need to see if the back retension plate is warped, or bent in any regards.

If the plate is warped, the heat sink mount pressure can be blamed.
If the plate is still fine, then the cpu is most definitely at fault.
It could not withstand the weight of the heat sink in a torque setting, hence the cpu itself started to bow along with the heat sink, which caused socket damage as listed.

This would totally be a manufacturer goof on intel seeing how the cpu itself warped under the weight of the heat sink.

People have bent motherboards, CPUs, and sockets. Chipped dies. Cracked boards, CPUs and sockets etc before due to over tension.

did u even see the damage to the socket?

Again... the rear plate prevents any of that sort from happening.
It forces the cpu to be sandwitched and should not bend the socket like that.

I have placed many many test hours on retension systems with i used to review waterblocks for the community.
Here is an example of extreme overtightening... yet you can clearly see the socket + cpu itself is kept flat..

This is mostly because the mounting holes are outside the socket + cpu area, and the board will rise at the pivot point of the mounting holes, and not the cpu itself.

With a proper backplate, which 95% of intel boards have now a days due to the latch system being built on them, it should of prevented the socket from being damaged like that, unless the said cpu itself became warped under the mounting pressure.

 

[LTC8K6]

Lastly, heat sinks are no longer flat anymore.
They have a slight bow to apply pressure in the middle to warp the IHS and allow a better mount with IHS + Die.

Who specified this non-flat heat sink? Are these specs listed anywhere?

 

[frozentundra123456]

All covered in the original thread...

Not sure why it's still being discussed, really.

http://forums.anandtech.com/showthread.php?t=2456465

Because some just want to beat to death anything that can be twisted to make intel look bad.

 

[aigomorla]

Who specified this non-flat heat sink? Are these specs listed anywhere?

it became standard when they realized it could improve performance by a couple of C's.

 

[LTC8K6]

it became standard when they realized it could improve performance by a couple of C's.

Says who?

Most heatsinks are incapable of warping the IHS, since they just attach with push pins or spring loaded screws.

Where is this spec about curved heat sinks and warped IHSs?