Investigating Reports Of Intel Skylake CPUs Damaged By CPU Coolers

Status
Not open for further replies.


its not the heatspreader that is bending, its the substrate that is affected.
It's also not Intel's responsibility if a company doesn't adhere to Intel's specifications when designing its mounting mechanisms.

Thus far it seems to be isolated to Scythe coolers but Intel is doing its due diligence to varify that.
 


It looks like the issue is one of two cases:

1. The system is moved around heavily and dropped (such as in shipping) or

2. The heatsink mounting causes more pressure to be put on than the CPU can handle. This is not due to there not being enough aluminum but rather than the CPU itself is smaller and thinner due to newer technologies.

This issue probably will not affect 99% of the people.
 
It's skylake, you don't need a mega cooler. Also, motherboards are thinner and flimsier as a result of needing to deal with less watts.
 


It is probably due to the 14nm process more than anything. This is an inevitable issue, the process node gets smaller not only in mm squared but also in the thickness and layers needed in the silicon substrate.

I remember not too long ago Intel was looking into ways to help mitigate the loss of cores due to just wear from how small the process nodes were getting.
 
lol. i get the xeon 1270 v1 for 90us! i will never get that new shine cpu. for now i have a 3770k 3470 and nice xeon 1270 why intel high price for a good garbage ?

What a genius... With that kind of thinking it's a miracle we're not still all using commodore 64s...
 
Built a few skylake systems for customers and luckily all have been with stock intel supplied coolers so far.

My new laptop has a skylake i5 but of course that's not a problem, although when I replace the thermal paste I'll now be extra careful.
 
Key words in Intel's response: "same STATIC load."

During shipping/moving, the CPU may experience DYNAMIC mechanical loads greatly exceeding static load: if you bump the case while moving it, the CPUs sees the full static load plus all the dynamic loads from the sudden acceleration or deceleration magnified by leverage.
 
My father was a quality engineer (Oversees the quality of the product). He told me no recall that has ever happened was unseen by them. They decided it was too small of a problem and its a good possibility that nobody would notice. I doubt it was never heard of until a few days ago as they have many people to oversee their quality not to mention test the CPUs.
 
I have a 6600K and a Corsiar H100i GTX and have found that after a couple days of troubleshooting that removing this CPU cooler eliminated a very strange set of seemingly impossible to identify causes for repeated crashes. Replaced MB, CPU, Ram, and HD before removing the most benign (at least so I thought) component, the CPU cooler. Temps were great, <45° C at full load using Prime 95 so I thought no way it's the cooler...
 
There might be only 5c worth of alu on a cpu cooler, but many weigh well over 1kg (nearly 3lbs for a dual fan cooler).
Add a decent length which gives it great leverage, and all you need is a few G's to put a massive 'twisting' force onto the CPU and socket.
And if the cooler is bolted down firmly to begin with it will take even less to exceed the 50lbs limit.
 
In terms of moving pc's with a large air cooler installed it's been good practice for some time now to remove the cooler while in transport. System builders have been encouraged to ship their towers with the cpu cooler separate and have the end user install the cooler at its final destination. That's not new.

Obviously larger coolers, especially taller tower style with further height/distance from the socket and any decent amount of weight (aka 212 evo types and larger) can exert much more dynamic stress than they do static pressure/load against the chip/socket. Having slightly thinner pcb's doesn't help the issue but in all honesty it's generally not a good idea to ship with a larger cooler installed.

If someone overtightens any cooler the clamping force can be too high but generally that's why aio coolers ship better installed than air coolers. The water block rides relatively low on the cpu with little weight to flex anything beneath it.
 
"did some independent testing and determined that the substrate on Skylake CPU’s is thinner than that of previous generations."
Why did they need an independend testing for that?! Intel themselves told us that.
 
Friend of mine few weeks ago assembled skylake system with i5, and he had problem with strangely bent cpu socket pins, though he didn't inspect cpu itself. Maybe it's his fault, but I know guy. He is very careful and its not his first system. But most important is that it was with boxed intel cpu cooler.
 
It isn't too surprising that it was missed in the initial QC. Intel would test their chips with their own stock coolers, possibly others, but such a thing is hard to predict. It is most like the bulkier / heavier heatsinks which cause this issue as they cause quite a bit of pressure due to weight. Add on a few bumps and the end result is a damaged CPU. Good to know that they're aware of it now, Intel can either thicken the top of their IHS a bit, or cooling vendors could modify their mount.
 


Probably not since, as you said, there is very little hanging weight which is the biggest issue when the system is being moved.

Thing is that having the end user install the heatsink after shipment has always been a good idea, or even removing the heatsink during long transportation due to the weight of some of these air coolers.
 

Net weight is not the problem since the CPU and socket are designed for 50lbF and even the heaviest heatsinks are under 3lb. The problem is the combination of mounting force applied by the screws, leverage and dynamic loads when things move.
 
Well this is scary. I installed an NH-D14 in a friend's Skylake build, now I'm worried it'll break, and he'll kill me.
 
Status
Not open for further replies.