Core #3 is hotter than the rest

[Bob Scott 7,719]

7 minutes ago, joemiller said:

Yeah, and I had delid the processor,

I didn't see this mentioned in the original post.  A void or bubble in the TIM under the heat spreader is also a possibility if you've delidded the CPU.

[joemiller 269]

3 minutes ago, martin-w said:

 

Huh! The liquid metal is applied to the surface of the die. And the underneath of the IHS. 

Right, but what I'm wondering is if Intel would make this surface wider to cover the entire surface of the "waffer" wouldn't this help to transfer more heat to the IHS ?  And, you mentioned liquid metal. I have not been using Liquid metal, I'll try this.  

[joemiller 269]

5 minutes ago, w6kd said:

In my experience, when running a stress test, it's pretty rare to see more than 5 deg difference between cores under full load.

I shoot for 80 deg C or below--during the stress test.  With P3D running, temps stay in the 50s.

Also, the fan curve on my Corsair H100i cooler (on a secondary PC running a 7700K) was not anywhere near aggressive enough...could be the pump/fans are not ramping up fast enough if you're using the default "quiet" fan curve.  I use the coolant temp rather than CPU temp, and go from minimum fan (500rpm) at 25 deg C to full speed at 30 deg C coolant temp.

Regards

I keep the pump at max speed, and the radiator fans on "smart mode", meaning at lower temps, the fans  run around 25% speed. And,  at high temp, they the fans run at near 85%  

[joemiller 269]

7 minutes ago, w6kd said:

I didn't see this mentioned in the original post.  A void or bubble in the TIM under the heat spreader is also a possibility if you've delidded the CPU.

Forgive me Bob, what exactly does all this means, I'm illiterate to what just  mentioned. LoL. Please explain. 

[martin-w 3,234]

17 minutes ago, joemiller said:

Right, but what I'm wondering is if Intel would make this surface wider to cover the entire surface of the "waffer" wouldn't this help to transfer more heat to the IHS ?  And, you mentioned liquid metal. I have not been using Liquid metal, I'll try this.  

 

They cant make it wider Joe, the die size is the die size. 

If you have delidded and aren't using liquid metal, then your CPU temp wont have dropped by more than 5 degrees or so. Using Thermal Grizzly Conductonaut you should see closer to 15 degrees drop in temp. Also, many see that core temps even out after delidding.

Apply a thin layer of liquid metal to the die. Very small amount, and spread out. Also apply a thin layer to underneath of IHS. Then, either mount the CPU in the socket with IHS on top and allow the CPU retention system to hold it in place, or... use a SLIGHT smear of silicone in the corner of the IHS to hold it in place. Clamp gently together and allow to set before mounting in socket. For on top of the IHS I use Kryonuat. 

If you re glue the IHS, as I said, make sure it's only a slight smear, because if it's a thick layer you will negate the entire purpose of delidding. 

 

Re your overclock... stick to 5 GHz on all cores at the lowest voltage, plus  a very slight bump up in voltage as a safety margin. That's it. it's all you need. Aim for 86 degrees or less. we cant have you pushing it to 5.3 at crazy temp for a mere one frame per second or so. 

Watch videos on delidding and learn to do it properly. 

 

Edited December 12, 2018 by martin-w

[HiFlyer 9,358]

38 minutes ago, joemiller said:

Well, the average temperature is around 98 degrees. And, I manually overclock it.

😲

[Bob Scott 7,719]

7 minutes ago, joemiller said:

Forgive me Bob, what exactly does all this means, I'm illiterate to what just  mentioned. LoL. Please explain. 

If there's a bubble or a hole in the thermal interface material--in your case either under the heat spreader (because you delidded it and replaced the factory TIM there) or between the heat spreader and the Corsair water block--then you may have created a hot spot on the surface of the die that isn't being cooled properly.

Why run your fans at 85% when the chip is overheating?  I found that the "smart" mode wasn't so smart.  The fans should be set to hit 100% long, long before the CPU gets to nearly 100 deg.

It's possible that the H100i AIO cooler just can't keep up with the heat output of a 5.3GHz hexacore CPU.  But if you're not running the cooler all-out, how would you be able to tell?

Regards

[joemiller 269]

21 minutes ago, w6kd said:

If there's a bubble or a hole in the thermal interface material--in your case either under the heat spreader (because you delidded it and replaced the factory TIM there) or between the heat spreader and the Corsair water block--then you may have created a hot spot on the surface of the die that isn't being cooled properly.

Why run your fans at 85% when the chip is overheating?  I found that the "smart" mode wasn't so smart.  The fans should be set to hit 100% long, long before the CPU gets to nearly 100 deg.

It's possible that the H100i AIO cooler just can't keep up with the heat output of a 5.3GHz hexacore CPU.  But if you're not running the cooler all-out, how would you be able to tell?

Regards

Thank you for the explanation. All great advises. So, I'll get a new CPU (same- 8086k) , and a new Liquid cooler.

--- And, I'm thinking about lapping the Die,  the IHS ,and the  heatsink. .. Good idea, or bad idea? 

[joemiller 269]

37 minutes ago, HiFlyer said:

😲

Yes too hot.

--- WONDERING HERE... I wish there would be some kind of liquid cooling kit that would "refrigerate" the liquid... make the liquid cold then pump it to the heatsink and radiator, or anything similar. Anyone knows if anything like this is out there? 

[martin-w 3,234]

50 minutes ago, joemiller said:

Thank you for the explanation. All great advises. So, I'll get a new CPU (same- 8086k) , and a new Liquid cooler.

--- And, I'm thinking about lapping the Die,  the IHS ,and the  heatsink. .. Good idea, or bad idea? 

 

Why are you getting a new CPU and cooler. There's no definitive evidence that there's anything wrong with your CPU. Or your cooler. Overclock to 5 GHz and leave it at that. Test temps then.

Lapping isn't worth it.

[Bob Scott 7,719]

1 hour ago, joemiller said:

Yes too hot.

--- WONDERING HERE... I wish there would be some kind of liquid cooling kit that would "refrigerate" the liquid... make the liquid cold then pump it to the heatsink and radiator, or anything similar. Anyone knows if anything like this is out there? 

Oh yeah, there are, and they're really expensive (well north of $1000).  And you also then have to worry that the below-ambient temp parts of the system may cause condensation to form and drip onto your expensive electronics.  Not a practical solution in most instances.

That said, a larger-capacity water cooling system can make a difference over the AIO units, particularly when the AIOs reach the point where they can no longer keep up.  The key factors are water flow, cooling (heat exchange) surface area, airflow, and ambient temperature.  A custom water loop can address the water flow issue (pump flow and/or plumbing size), surface area (larger/multiple radiators), and airflow (more fans, shrouds etc).  I use two 120x360mm (triple-fan) radiators and large half-inch ID hose...the system was initially designed for a hot-blooded i7-975 CPU and GTX480 back about 10 years ago, and the cooling loop has been used on four build generations since then and still serves well to this day.  Because of all that radiator (overkill in my current app, but hey, it was already there), I run the fans way down low so it's whisper-quiet.  I also keep the "cooling tower" (pump, reservoir, radiators and fans) in a separate enclosure, with the two hoses forming a drip loop before running into the case.  That keeps the possible sources of a leak inside the case to a bare minimum.  Right now I do not have a water-cooled GPU, so all that cooling serves just the overclocked 8086K.  If I hit a thermal wall on the CPU, it's because the heat can't get from inside the chip out to the IHS, not because the cooling loop can't dissipate all of it.

If you want to run at 5.3, I think something more substantial than an H100i should probably get a look.  They make a somewhat larger version, the H115i, which I think has a 140x280mm radiator instead of the 120x240 in the H100i, but I don't know that'd make enough of a difference for you.

Regards

[lownslo 592]

2 hours ago, joemiller said:

And, I'm thinking about lapping the Die,  the IHS ,and the  heatsink. .. Good idea, or bad idea? 

Trying to hand lap the components of a CPU could do more harm than good.  Given how these products are assembled, delidding is a good idea (especially if you have a hexaCPU that has a temp difference across all cores of more than 5-7 degrees), but simply cleaning-up the components and then careful application of a liquid metal between the die and IHS is quite sufficient.

Good luck,

Greg

[martin-w 3,234]

Lapping CPU IHS and the die results in very little reduction in temp, often just 2 degrees or so.

I have heard of some IHS's though that are not flat, more convex. In which case lapping can make a bigger difference. You know if your IHS is convex if you place it face down and try to spin it. If it spins like a top its not flat. If it has resistance to spinning due to friction its flat. 

[joemiller 269]

32 minutes ago, martin-w said:

Lapping CPU IHS and the die results in very little reduction in temp, often just 2 degrees or so.

I have heard of some IHS's though that are not flat, more convex. In which case lapping can make a bigger difference. You know if your IHS is convex if you place it face down and try to spin it. If it spins like a top its not flat. If it has resistance to spinning due to friction its flat. 

Good idea. Well, I'm going to order some liquid metal. This may be a foolish question or may deserve a knock in the forehead...but, is it possible to remove the IHS and sit the heat sink on the opened CPU...on the die itself. Or will it kill the die, dead.  Hahaha. (No I won't try it, let me know first) 

 

[martin-w 3,234]

9 minutes ago, joemiller said:

 is it possible to remove the IHS and sit the heat sink on the opened CPU...on the die itself.

 

 

Yes, direct die contact. Some CPU's have direct die mounting frames available for them. Again though, the reduction in temperature isn't huge. Usually just a few degrees, so not worth the effort and risk. Die can be cracked. It's only super serious competitive overclockers looking for one or two degrees drop in temp that bother. 

Earlier you mentioned chilling the water. Yes, chillers are available. see Robs topic.