CPU lapping

[Intervention]

I want to improve my cpu cooling but i dont want to use liquid metal either. is lapping a CPU a good idea? I have the same CPU as this guy:

Lapping an AMD R7 2700. Is it worth it?

 

[Vya Domus]

Probably a major waste of time.

 

[Intervention]

Probably a major waste of time.

Why do you think that? Too much work or is it risky?

 

[Vya Domus]

Why do you think that? Too much work or is it risky?

I guess it's risky because you might bend the pins (if it's an AMD chip). But that's not really it, people have tried this over and over and the results have varied from absolutely no change to maybe a couple of degrees. Question is, are you really that thermally limited such that this is the last option you have ?

 

[Intervention]

I guess it's risky because you might bend the pins (if it's an AMD chip). But that's not really it, people have tried this over and over and the results have varied from absolutely no change to maybe a couple of degrees. Question is, are you really that thermally limited such that this is the last option you have ?

I have a good AIO, the Corsair H100x, and using TG Kryonaut TIM. There is plenty of ventilation also. I am running the AMD R7 2700 at 1.425v. Clocked at 4.2Ghz all cores. Temperatures are in the low 80's. I saw a video by Der8uer also where he lapped a cpu and gain about -5c! Thats major in terms of cooling for overclocking.

 

[Vya Domus]

If you thinks it's worthwhile go ahead.

A 5c difference wouldn't mean anything to me if I had no problems anyway such as thermal throttling or anything like that. And it's really unlikely that 5c will get you a higher overclock.

 

[ShiBDiB]

It sounds like your mind is already made up? Why even make this thread

 

[HUSKIE]

It's soldered Cpu.

Why do you think that? Too much work or is it risky?

Try it and you will find how it goes.

 

[dorsetknob]

I'm leaving this site after this comment. I think this post really shows where TPU stands. Stupid libs.

Welcome Back
you won't gain any appreciable drop in temp

 

[eidairaman1]

Are the ihs soldered on for AMD?

By the way use Thermalgrizzly Kryonaut, Geil GC Extreme Cooler Master Gel Maker, Arctic Cooling MX 4 or Arctic Silver 5

 

[newtekie1]

Are the ihs soldered on for AMD?

The CPUs are soldered the APUs use paste.

 

[TheoneandonlyMrK]

I have a good AIO, the Corsair H100x, and using TG Kryonaut TIM. There is plenty of ventilation also. I am running the AMD R7 2700 at 1.425v. Clocked at 4.2Ghz all cores. Temperatures are in the low 80's. I saw a video by Der8uer also where he lapped a cpu and gain about -5c! Thats major in terms of cooling for overclocking.

I got about that -5°c lapping an fx8350 but its IHS was quite concaved , i couldn't guess on the gains with Ryzen though since monoblocks are made for stock IHS i didn't lap this one.

 

[infrared]

I still haven't got my pc back together yet, but here's my lapped 1800X. The heatspreader was fairly concave so under heavy load it usually around 75c.. Not a big deal but considering that's on decent watercooling I'd prefer it to be cooler.

I thought about ways to protect the pins but in the end I just used 3 fingers spread across all the pins with light contact and didn't bend any.

 

[MrGenius]

That right there is how it's done!^^^

Anyhow...don't spend the time to do a ultra-fine lap job and then use Kryonaut. That shit will RUIN all your hard work. I don't know for sure what's in it(I HIGHLY suspect it to be nano-diamonds). But whatever it is, it might as well be beach sand. It will pit and scratch your surfaces SEVERELY.

FYI: Thermalright just released some TIMs that have better W/mK than Kryonaut. I haven't tried them yet. Hopefully they aren't nano-diamond filled too. That shit's just not safe for mating surfaces. None of it. Every nano-diamond paste I've used caused damage to one degree or another. Kryonaut being the worst...BY FAR. So it either has nano-diamonds in it...or something even worse.

TFX 14.3 W/mK
TF8 13.4 w/mK
TF6 12.5 w/mK(same as Kryonaut)

 

[TheoneandonlyMrK]

I still haven't got my pc back together yet, but here's my lapped 1800X. The heatspreader was fairly concave so under heavy load it usually around 75c.. Not a big deal but considering that's on decent watercooling I'd prefer it to be cooler.

I thought about ways to protect the pins but in the end I just used 3 fingers spread across all the pins with light contact and didn't bend any.


View attachment 121007

View attachment 121008

I got a foam pad with a cpu once ,the perfect size for most cpus , quite firm foam i push the cpu into it.

 

[infrared]

That right there is how it's done!^^^

Anyhow...don't spend the time to do a ultra-fine lap job and then use Kryonaut. That shit will RUIN all your hard work. I don't know for sure what's in it(I HIGHLY suspect it to be nano-diamonds). But whatever it is, it might as well be beach sand. It will pit and scratch your surfaces SEVERELY.

Ahh I wondered why my waterblock had some mild pitting in it after a couple of years, I was blaming myself for getting some contaminants on it or something. Oh well, still my favourite paste.

Edit - will check out those other pastes tho, ty

 

[MrGenius]

Thermalright also has some liquid metal that's got better W/mK than Conductonaut too. But I can't find their product page for it(I did order some on Newegg a few days ago though).

Silver King Liquid Metal 79 W/mK

Also picked up a tube of DEBROGLIE liquid metal recently that's >128 W/mK. Haven't tried it yet though. Saving it for a special occasion.

Yes...I am a MAJOR TIM snob.

 

[FreedomEclipse]

Welcome Back
you won't gain any appreciable drop in temp

Don't be so quick to palm him off

 

[Susquehannock]

I guess it's risky because you might bend the pins (if it's an AMD chip). But that's not really it, people have tried this over and over and the results have varied from absolutely no change to maybe a couple of degrees. Question is, are you really that thermally limited such that this is the last option you have ?

^^ This exactly. Is your system running hot and throttling? Or are you trying to lower temps a few degrees in hopes of increasing longevity? Only you can determine whether the time and effort is worth it.

Varied improvements (if at all) after lapping are almost certainly caused by differences in flatness from CPU to CPU. No two will be exactly same. Below is my Xeon W3680 as example. Heatspreader was far from flat. Starting with wet 400 grit paper on a mirror, high spots in shape of a capital letter "I" became immediately apparent and was obvious I would be there all day. So switched to 120 grit to speed things up. Finished with wet 2,000 grit. When you get into surfaces like this TIM particle size becomes more an issue. Are they small enough to drop into micro imperfections in the mirror-like surface? Or will they sit on top and prevent the metal surfaces from touching much as they could? Which is what we want in an ideal situation - direct metal to metal contact.

Now the question is ... did this improve my temps? Hard to say since small drop perceived was well within margin of error. Switching to Kryonaut already dropped temps a few degrees over the AS5, and several degrees lower than standard white paste, even before lapping.

As for diamonds. I don't think Kryonaut employs them otherwise it would be a big marketing point. More likely they use Aluminum Oxide (aka - alumina) like most other ceramic non-metallic or silicone based TIMs. Excellent thermal conductivity, inexpensive, and readily available since significant portion of Earth's crust (read: 'bauxite') is comprised of it. And by the way. Aluminum Oxide is a very common abrasive and probably what is used on the sandpaper you will be lapping with. So no wonder it scars CPU surfaces.

 

[Bones]

I normally just use Ceramique 5 for everyday stuff and it works, sometimes I'll breakout the Noctua NT-H1 and use it instead.

In terms of an everyday build there isn't enough difference between TIM's to worry about it alot as long as it's a good quality TIM and applied correctly.

Kryonaut for example is geared more for extreme OC'ing and speaking of the diamonds or whatever else extreme OC'ers aren't looking for longevity of a CPU, they are looking for MHz so longevity isn't a concern.
That's why it's made the way it is.

Works for everyday stuff but (For me) too expensive to justify using it for everyday stuff when there are cheaper alternatives out there that would do about as well anyway in an everyday scenario, not to mention the bad side effects of it's use as mentioned above.
If you want to use Kryonaut you certainly can but personally I don't.

MX4 is a good TIM to use and not too expensive, have some of that here too and it's the newer version with carbon. https://www.newegg.com/Product/Product.aspx?Item=9SIAGXK92Y6326
Or this: https://www.newegg.com/Product/Product.aspx?Item=9SIAGTU91U5187

EDIT:
The actual stuff I got earlier: https://www.newegg.com/Product/Product.aspx?Item=9SIA4YU92X0379

 

[delshay]

If you really want to lap a pinned CPU, then I strongly recommend getting a CPU socket. Either buy one or cut one from a dead motherboard.
This way you have taken away any risk of bending or breaking any pins. This is what I do & is standard when I lap any pinned CPU.

 

[delshay]

Here's a bit of a shocker.

Iv been doing a new CPU modification never seen before (to be uploaded to the GHETTO THREAD)), but this is not the reason why I'm posting in this old thread.

Every time I change the Liquid Metal Compound I use a brillo pad to scrub the surface of the CPU core. With the new modification in place I'm thinking I have broken the the CPU as there is a strange reading to ground from the heatsink. I was then thinking part of the modification is shorting to the IHS..

After a long check I found the CPU core, the part where all of us put our thermal paste has become conductive, ie it's out putting 21.2mv. I did not know this part can become conductive. I check three other CPUs, it's not conductive,. So it looks like if you lap the core it will become conductive. at some point, at lease it does here on a single CPU.

 

[Durvelle27]

With these type of questions


My answer is always it's only worth as much as you say its worth

The risk of damaging the CPU is much higher than the gains you may or may not achieve

 

[FreedomEclipse]

If youre talking about scrubbing the bare cpu die with a brillo pad - i think thats where you messed up.

Im not that smart but im sure that CPUs usually come with some sort of protective coating/layer similar to varnish that protects the die to a certain extent. By using the brillo pad you have worn away this protective layer and the LM has been absorbed by the CPU itself and become conductive.

The pad on a brillo sponge is quite a hard and abrasive, normally used to scrub chunks of left over food off plates and burnt chunks off cookware. Some cotton swaps or soft tissue and isopropyl alcohol would have worked just fine on the CPU

 

[Valantar]

Here's a bit of a shocker.

Iv been doing a new CPU modification never seen before (to be uploaded to the GHETTO THREAD)), but this is not the reason why I'm posting in this old thread.

Every time I change the Liquid Metal Compound I use a brillo pad to scrub the surface of the CPU core. With the new modification in place I'm thinking I have broken the the CPU as there is a strange reading to ground from the heatsink. I was then thinking part of the modification is shorting to the IHS..

After a long check I found the CPU core, the part where all of us put our thermal paste has become conductive, ie it's out putting 21.2mv. I did not know this part can become conductive. I check three other CPUs, it's not conductive,. So it looks like if you lap the core it will become conductive. at some point, at lease it does here on a single CPU.

What do you mean by "CPU core"? The IHS? The die? If it's the IHS, it's copper, so that sounds like you've gotten something conductive between it, some voltage rail, and it might be shorting to ground through the heatsink. If you mean the die itself, then yes, sanding that down will eventually expose the conductive innards of the CPU.