Is the Skylake "bending" issue, really an issue?

[Noirgheos]

3 minutes ago, typographie said:

The warning about coolers over 700 grams is to remove the cooler when shipping the system, which I'd recommend anyone to do. Probably even with coolers under 700 g. That has nothing to do with Skylake specifically, it's so that the cooler doesn't pull on the motherboard when the PC moves during travel.

 

I don't think the weight of the cooler itself is a factor in whether Skylake bends or not. In an upright PC case, the weight of the cooler is directed toward the floor, not into the CPU die. It's mounting pressure that matters, and here's what Noctua says about that:

 

"Our SecuFirm2™ mounting systems undergo comprehensive compatibility tests before being approved for new platforms. We have neither detected any problem with regard to the Intel LGA1151 platform (“Skylake”) during these tests nor have we received any reports from our customers, sales partners and system integrators that would indicate any possible issue. Our SecuFirm2™ mounting systems (with the exception of some compact L-type models) rely on coil springs to create the required contact pressure, which gives a certain flexibility both with regards to tolerances in stackup height and shocks or other forces. Compared to spring-less mounting systems, which exclusively rely on the bending of the mounting brackets in order to create contact pressure, this allows to reduce the mechanical stress on CPU, socket and motherboard and thus helps to prevent possible damage from excessive forces. However, as it is not possible to reliably calculate or control the forces that act upon a system during transport (e.g. in shipping), we generally recommend, due to safety reasons, to take off coolers with a total weight of more than 700g (incl. fan)."

 

The only mention of weight is at the end, where they reiterate that they recommend coolers over 700 g are removed before shipping. And again, that's not specific to Skylake.

So I'm basically overreacting? My U12S should be fine as long as I don't move my PC around too much?

[KWelz]

43 minutes ago, don_svetlio said:

From a user at PCgamer

I find it highly unlikely this is from a cooler. 

The edges of the processor are covered by the retention bracket.  I can think of no way that a cooler mounted on the chip can cause this.   An over tightened cooler could cause it to bend if there is enough weight there but it would be a bend across the entire chip not just the edge like in the photo.  

 

 

As for the OP.  I have a Noctua DH15s on a 6600K and have had no issues.  

[Noirgheos]

1 minute ago, KWelz said:

I find it highly unlikely this is from a cooler. 

The edges of the processor are covered by the retention bracket.  I can think of no way that a cooler mounted on the chip can cause this.   An over tightened cooler could cause it to bend if there is enough weight there but it would be a bend across the entire chip not just the edge like in the photo.  

 

 

As for the OP.  I have a Noctua DH15s on a 6600K and have had no issues.  

How long have you had it on?

[typographie]

Just now, Noirgheos said:

So I'm basically overreacting? My U12S should be fine as long as I don't move my PC around too much?

I don't know, I can't say because there just isn't any information. But that alone suggests to me that it's just not happening much. The ones who did report it seem to be ones who used an electric screwdriver to over-tighten the mounting hardware. I remember hearing about this back near Skylake's release, and then never again until I saw this thread. I would not expect problems if you hand-tighten and follow the directions.

 

Noctua doesn't seem too concerned. They advertise Skylake compatibility without even specifying model names, and they're the ones users will go to if Noctua coolers kill their CPUs.

[Noirgheos]

1 minute ago, typographie said:

I don't know, I can't say because there just isn't any information. But that alone suggests to me that it's just not happening much. The ones who did report it seem to be ones who used an electric screwdriver to over-tighten the mounting hardware. I remember hearing about this back near Skylake's release, and then never again until I saw this thread. I would not expect problems if you hand-tighten and follow the directions.

 

Noctua doesn't seem too concerned. They advertise Skylake compatibility without even specifying model names, and they're the ones users will go to if Noctua coolers kill their CPUs.

I use a normal screwdriver and tighten until the spring screws won't allow me to tighten further, and my screwdriver just slips off.

[KWelz]

4 minutes ago, Noirgheos said:

How long have you had it on?

Only a couple months.  It is my Hackintosh So while less powerful than my primary gaming computer it actually used more often.   

[Noirgheos]

Just now, KWelz said:

Only a couple months.  It is my Hackintosh So while less powerful than my primary gaming computer it actually used more often.   

Alright my worries are pretty much alleviated. Also, correct me if I'm wrong, isn't the bending caused by the mounting pressure rather than the weight of the cooler?

[Scitesh]

1 hour ago, Noirgheos said:

Well I'm waiting for people who use heavier coolers with Skylake to give me their experience. I have already out the U12S on a 6600K no issues for a few months, but I just want to make sure I'm not treading a thin line...

 

It also seems the article that reported it in the first place used a drill to secure the cooler, and every other site just copied and pasted that article to begin with.

if you dont have the cooler go with an AIO (i used 212x with 6700k later upgraded to h100iGTX) 

[KWelz]

Just now, Noirgheos said:

Alright my worries are pretty much alleviated. Also, correct me if I'm wrong, isn't the bending caused by the mounting pressure rather than the weight of the cooler?

Physics says that is the only way it can happen.   The stress points for the mounting of the cooler are on the board.  In the case of most after market coolers you have the stress spread out due to the use of specialized mounting hardware such as a backplate.   The only pressure on the CPU is directly downward.   Now good coolers are going to have a slightly convex curve to it.  SO if you really over-tighten the cooler and it gets bounced around a lot, then it is possible it could cause some bending.   However this bending would originate from the center of the CPU.  And it would have to transfer first through the lid and then to the processor itself.  

 

So yes it can happen.  But only in a specific set of circumstances and due in part to mistakes by the builder.    

[Noirgheos]

7 minutes ago, KWelz said:

Physics says that is the only way it can happen.   The stress points for the mounting of the cooler are on the board.  In the case of most after market coolers you have the stress spread out due to the use of specialized mounting hardware such as a backplate.   The only pressure on the CPU is directly downward.   Now good coolers are going to have a slightly convex curve to it.  SO if you really over-tighten the cooler and it gets bounced around a lot, then it is possible it could cause some bending.   However this bending would originate from the center of the CPU.  And it would have to transfer first through the lid and then to the processor itself.  

 

So yes it can happen.  But only in a specific set of circumstances and due in part to mistakes by the builder.    

I'll pretty much only be moving my PC around on its desk, maybe laying it down flat on its side once in a while but if I do, I'll probably hold the cooler up a little just so nothing happens, even if I don't have to. Gently laying it forward on its side shouldn't do anything right? I mean it is technically going downward on the CPU, but if your cooler doesn't do that, why should mine, which weighs 3/4? I mean Noctua means shipping by transport in their statement, so unless I throw my PC around like in shipping, I don't see why I should have any issues just moving it around the house.

[KWelz]

2 minutes ago, Noirgheos said:

I'll pretty much only be moving my PC around on its desk, maybe laying it down flat on its side once in a while but if I do, I'll probably hold the cooler up a little just so nothing happens, even if I don't have to. Gently laying it forward on its side shouldn't do anything right? I mean it is technically going downward on the CPU, but if your cooler doesn't do that, why should mine, which weighs 3/4? I mean Noctua means shipping by transport in their statement, so unless I throw my PC around like in shipping, I don't see why I should have any issues just moving it around the house.

When you are talking about transporting the case we mean transporting it.   Car, freight, etc.  Moving it around your house isn't going to be a problem.  

[Noirgheos]

2 minutes ago, KWelz said:

When you are talking about transporting the case we mean transporting it.   Car, freight, etc.  Moving it around your house isn't going to be a problem.  

Alright I think I've had enough re-assurance now. Will leave my U12S on. Thank you. Just hope I don't start BSOD'ing or my PC stops turning on in a few years

[don_svetlio]

1 minute ago, Noirgheos said:

Alright I think I've had enough re-assurance now. Will leave my U12S on. Thank you. Just hope I don't start BSOD'ing or my PC stops turning on in a few years

If it works for a few weeks you'll be fine,

[Noirgheos]

3 minutes ago, don_svetlio said:

If it works for a few weeks you'll be fine,

Had it on my old 6500 for a week before I upgraded. PCB looks pretty straight.

[don_svetlio]

Just now, Noirgheos said:

Had it on my old 6500 for a week before I upgraded. PCB looks pretty straight.

I'd give it a 2-3 week testing period personally

[Daveeede]

What I find funny about this whole fiasco is that if someone in civil engineering (a field that has literally nothing to do with processors) would have calculated the forces on the substrate, they could've avoided the problem. The calculation is identical to determining the thickness of plate that a column sits on. 

 

Kind of ironic.

[KWelz]

1 minute ago, Daveeede said:

What I find funny about this whole fiasco is that if someone in civil engineering (a field that has literally nothing to do with processors) would have calculated the forces on the substrate, they could've avoided the problem. The calculation is identical to determining the thickness of plate that a column sits on. 

 

Kind of ironic.

Not really.  The two are in no way related.   The CPU is not load or stress bearing.  In the case of a Heat sink that stress is taken up by the logic board.  The CPU is then sandwiched between two plates with space left for the cooler to make contact.   So the only pressure on the board is from that point of contact.  It doesn't actually support the weight.  

 

[Noirgheos]

2 minutes ago, KWelz said:

Not really.  The two are in no way related.   The CPU is not load or stress bearing.  In the case of a Heat sink that stress is taken up by the logic board.  The CPU is then sandwiched between two plates with space left for the cooler to make contact.   So the only pressure on the board is from that point of contact.  It doesn't actually support the weight.  

 

Yep with SecuriFirm 2 most of the pressure and weight is put on the brackets right?

[Daveeede]

Just now, KWelz said:

Not really.  The two are in no way related.   The CPU is not load or stress bearing.  In the case of a Heat sink that stress is taken up by the logic board.  The CPU is then sandwiched between two plates with space left for the cooler to make contact.   So the only pressure on the board is from that point of contact.  It doesn't actually support the weight.  

 

But isn't the bending issue caused by the pressure between the heatsink and the cpu heatspreader? I understand that the weight of the cooler is supported by the motherboard, but there is still pressure applied to the processor. In a building, the baseplate is technically "sandwiched" between the column and the foundation. That doesn't mean it sees no stress.

[Daveeede]

5 minutes ago, KWelz said:

Not really.  The two are in no way related.   The CPU is not load or stress bearing.  In the case of a Heat sink that stress is taken up by the logic board.  The CPU is then sandwiched between two plates with space left for the cooler to make contact.   So the only pressure on the board is from that point of contact.  It doesn't actually support the weight.  

 

Unless you're saying there's no pressure at all to the heatspreader, in which case why again are these things bending?

[KWelz]

3 minutes ago, Noirgheos said:

Yep with SecuriFirm 2 most of the pressure and weight is put on the brackets right?

with this method of mounting you have 4 screws going through the board and being secured to a backplate.  This backplate distributes the force across a larger area.  In addition it uses 4 plastic barrels that the screws pass through that help distribute additional force on the front of the board.  The Cooler is then attached at 2 points These points also have springs that keep too much additional pressure from being applied.  

1 minute ago, Daveeede said:

But isn't the bending issue caused by the pressure between the heatsink and the cpu heatspreader? I understand that the weight of the cooler is supported by the motherboard, but there is still pressure applied to the processor. In a building, the baseplate is technically "sandwiched" between the column and the foundation. That doesn't mean it sees no stress.

You are correct that there is some stress.  However the CPU is no supporting the cooler.  And this is the important difference.   

 

The only way to flex the CPU is to overtighten the cooler and even then only in some extreme cases.  

[KWelz]

2 minutes ago, Daveeede said:

Unless you're saying there's no pressure at all to the heatspreader, in which case why again are these things bending?

Because people are not following specs.  A heat spreader should be firmly attached.  Not torqued down by a gorilla.  When you overtorque the screws on a mount on a heat spreader that has a convex shape then it could bend the chip.  But this is only one factor involved in the issue.  

[Daveeede]

Just now, KWelz said:

The only way to flex the CPU is to overtighten the cooler and even then only in some extreme cases.  

This is what I'm talking about. As you overtighten the cooler, this causes higher stresses on the processor, causing the substrate to fail and "bend" along the edges. 

 

To say that putting a cooler over 700g will bend your cpu is crap, because most of that weight (if not all) is supported by the motherboard. But to fix that cooler to the motherboard without it falling, may require you to really tighten the screws. This overtightening is the issue, not the actual weight of the cpu cooler.

[Noirgheos]

1 minute ago, KWelz said:

Because people are not following specs.  A heat spreader should be firmly attached.  Not torqued down by a gorilla.  When you overtorque the screws on a mount on a heat spreader that has a convex shape then it could bend the chip.  But this is only one factor involved in the issue.  

Also, a lot of people reporting it were using drills to tighten coolers, which I doubt helped.

[Noirgheos]

Just now, Daveeede said:

This is what I'm talking about. As you overtighten the cooler, this causes higher stresses on the processor, causing the substrate to fail and "bend" along the edges. 

 

To say that putting a cooler over 700g will bend your cpu is crap, because most of that weight (if not all) is supported by the motherboard. But to fix that cooler to the motherboard without it falling, may require you to really tighten the screws. This overtightening is the issue, not the actual weight of the cpu cooler.

Well his didn't bend under a 900+g cooler, and mine weighs less than 700g without the fan, and I sure as hell won't be installing with the fan on. I'll be putting it on after.