3770K to lap or not to lap

[Redoitall]

Ok so I have the following question. I already delidded my 3770k, temps with H-100 are ok I think they stay below 70 c if I don't go beyond 4.3 to 4.5 I have reseated block several Xs. Keep in mind one of the cores is 10 Celcius cooler than the other 3. I am wondering if I would benefit from lapping the IHS or by simply putting Liquid Metal Pro on the die to the IHS and Artic S 5 or prolimatech on the IHS to HSF alias H-100. Is the surface of the IHS the culprit of the inconsistency in temps, or is it the lack of soldering? Or is it the 22nanometers too small to handle even temps? I am no expert at all I am just asking because it is all over and most people are having trouble keeping all cores within 5 Celcius . Oh one more question, is H 100 really ok or should I go to noctua I have been somewhat disappointed with Corsair, not that is terrible but it seems a little latent when it comes to cooling.

 

[Liquid_Static]

Lapping has provided some individuals on this forum with tangible benefits and has not helped others in the least. It really depends on how flat the IHS is to begin with.

 

[Idontcare]

You would benefit from lapping both the IHS as well as the H100.

My H100 was extremely warped and concave. Took a long time to get it flat.

That said, 70C at 4.5GHz is what I get too.

 

[UaVaj]

since you already delid. toss the IHS. unless there is a good reason to retain the IHS (is there even one?). direct on die for maximum cooling.

as for lapping. definitely do lap. have yet to find a (consumer grade production) cpu IHS nor HSF that is true flat out of the box.

 

[BrightCandle]

Waterblocks are always concave. The waterblocks (presumably prebuilts as well but I haven't seen any tests) get better performance for doing it as they achieve higher pressure where the grand majority of the heat is and the flex in the motherboard and CPU mean the contact in other areas is decent. When you lap it away performance is reduced.

Lapping the IHS on the other hand can be quite beneficial because they tend to be quite a rough surface and the best thermal grease is terrible in comparison to genuine metal on metal contact.

 

[Idontcare]

Waterblocks are always concave. The waterblocks (presumably prebuilts as well but I haven't seen any tests) get better performance for doing it as they achieve higher pressure where the grand majority of the heat is and the flex in the motherboard and CPU mean the contact in other areas is decent. When you lap it away performance is reduced.

Lapping the IHS on the other hand can be quite beneficial because they tend to be quite a rough surface and the best thermal grease is terrible in comparison to genuine metal on metal contact.

I think you mean convex? Concave means it is a bowl which results in a thicker pool of TIM in the center of the block instead of the edges.

If your waterblock is concave, which my H100's was, then you most certainly want to lap it so it is flat which will reduce the thickness of the TIM needed between the two surfaces.

Convex is better than flat only if the radius of curvature is ridiculous huge and the stiffness (deformability) of the block is quite small. Otherwise you end up with great contact only at one point and everywhere else you end up with an air gap that must be filled in with TIM.

Case in point is the AMD FX-8350 and stock HSF that I just lapped. They were both convex, ideal if you believe the waterblock mantra. But lo and behold lapping them and taking those convex surfaces to flat resulted in a 9°C reduction in temperatures.

 

[CurrentlyPissed]

Unless you delid, lapping is pretty much moot. Seen no difference. IDC also has a post on this if you search. I have seen some people actually lap the underside as well. I wonder if IDC can post up something about that.

Also shocked no one asked. What voltage are you running?

I use a custom wc setup so my resaults are obviously going to be different. But I did a prime run 5.1@1.4 for around 65-70c. Around 1.45v and I start seeing consistent prime runs in the 70s. Granted 3dMark, and gaming sits around 60c.

 

[UaVaj]

Case in point is the AMD FX-8350 and stock HSF that I just lapped. They were both convex, ideal if you believe the waterblock mantra. But lo and behold lapping them and taking those convex surfaces to flat resulted in a 9°C reduction in temperatures.

x2. lapped both pentium d and stock hsf. gain ~12C. from the red zone to a comfortable 59C. with prime95.

 

[Redoitall]

Unless you delid, lapping is pretty much moot. Seen no difference. IDC also has a post on this if you search. I have seen some people actually lap the underside as well. I wonder if IDC can post up something about that.

Also shocked no one asked. What voltage are you running?

I use a custom wc setup so my resaults are obviously going to be different. But I did a prime run 5.1@1.4 for around 65-70c. Around 1.45v and I start seeing consistent prime runs in the 70s. Granted 3dMark, and gaming sits around 60c.

Hi thanks for the replies.
I am running 4.5 @ 1.20 4.6 @ 1.25 4.7 @ 1.30 4.8 @ 3.75 . all these voltages are stable Running prime 95 but Even with Delid at 4.8 things get very hot , upper 90's and one of the cores at 100 which at that point it throttles back. In the other hand one of the cores always stays below 90. That brings me to think that there is some kind of uneveness that I just do not understand. When I had My I7 920 CO it never had that problem. To tell you the truth these has been so educational about themal conditions under the hood of a cpu. It is amazing the amount of people such as IDONTCARE and some other guy GResky, and many others have done all sort of diffenrent tests with very mixed results. I think Ivy is a moody processor that yields differet results to everyone that runs in her path

 

[Idontcare]

Hi thanks for the replies.
I am running 4.5 @ 1.20 4.6 @ 1.25 4.7 @ 1.30 4.8 @ 3.75 . all these voltages are stable Running prime 95 but Even with Delid at 4.8 things get very hot , upper 90's and one of the cores at 100 which at that point it throttles back. In the other hand one of the cores always stays below 90. That brings me to think that there is some kind of uneveness that I just do not understand. When I had My I7 920 CO it never had that problem. To tell you the truth these has been so educational about themal conditions under the hood of a cpu. It is amazing the amount of people such as IDONTCARE and some other guy GResky, and many others have done all sort of diffenrent tests with very mixed results. I think Ivy is a moody processor that yields differet results to everyone that runs in her path

The core that sits next to the iGPU will always run cooler than the other cores because that core has a huge area of underutilized silicon sitting next to it acting as a region of "gray silicon" (not quite dark silicon, but same concept) that helps to lower temperature by increasing the effective surface area that is dissipating heat.

Since the IB die is not symmetric, there is no way to have symmetric operating temperatures across the cores unless by some fluke coincidence the thermal conductivity just so happens to mirror match the asymmetry in the core layout...very unlikely to happen.

Your reported results do not surprise me.

 

[Redoitall]

The core that sits next to the iGPU will always run cooler than the other cores because that core has a huge area of underutilized silicon sitting next to it acting as a region of "gray silicon" (not quite dark silicon, but same concept) that helps to lower temperature by increasing the effective surface area that is dissipating heat.

Since the IB die is not symmetric, there is no way to have symmetric operating temperatures across the cores unless by some fluke coincidence the thermal conductivity just so happens to mirror match the asymmetry in the core layout...very unlikely to happen.

Your reported results do not surprise me.

That is a great explanation thanks a lot. So from what I see on the graph lapping will only make things cooler and maybe slightly even . But to obtain perfect symmetry on each core would be impossible. I see even sandy suffers from the same ailment which still is a phenomenal processor.This is by far the best forum I have ever been to I thank you very much I learned more in three days than in a month of investigating through other sources . I will start sanding IHS Tomorrow. And applying Liquid metal to die afterwards , will report before and after results :thumbsup::thumbsup: