7700k delid with CLU - should I redo it?

[Czachol]

Hi everyone,
This is my first post here, but I have been reading the forum for a long time.

I have delidded my 7700k (with Rockit 88-super easy) due to unacceptable temps even after changing to kraken x61 AIO. Overall improvement I gained is 17C on load (Prime 95 v26.6, small fft, stock voltage and freq). While I am quite happy with the result, I am not sure whether my application of CLU is prone to causing long term issues. I am concerned if air bubbles may have formed. I used a miniscule amount of it, but the spread is far from mirror like.

Can you guys give me your opinion on whether I should redo the procedure or leave it like this?

I am attaching some pictures of the process.

 

[aigomorla]

leave it as it is.

I dont see anything wrong with the application.
As u clamp down the IHS again, the bubbles ur worried about will be pushed off to the side so it wont impact much.

 

[Czachol]

Thx aigomorla, that's good to hear. One more thing: I just realized I didn't protect the row of tiny contacts near the corner of the die with varnish or otherwise. Can this be an issue?

 

[DrMrLordX]

Maybe I shouldn't comment since my delid/relid experience was with a Kaveri, but I will chime in that I actually used a large amount of CLU between the die and IHS since my initial attempt at a relid was a terrific failure. It did not press out beyond the die onto any critical element of the CPU PCB that I could tell. The only downside was that I got some spatter when I removed the heatsink and the IHS bonded itself to the HSF (due to copper/copper contact and CLU, hah woops). Anyway it appears as though you used less than I.

CLU and other metal TIMs (Conductonaut, Phobya Liquid Metal) do not seem to retain air bubbles to any significant extent. They even out very quickly under pressure.

 

[Czachol]

Thanks for your input. Well, the principle is basically the same with AMD CPUs I guess. Are saying that more CLU is better? What did you do wrong the first time? Sorry i didn't get it, most certainly because I'm not a native English speaker heh.

 

[DrMrLordX]

I had a lot of experience with CLU between the base of an HSF and the CPU IHS where I would paint on a very, very thin amount. I tried to do the same with the silicon die of my A10-7700k, leading to horrendous thermal contact between the IHS and die after relidding the CPU. Putting on a larger quantity - maybe 2-3 grains of rice worth - fixed the problem nicely.

 

[otinane]

I'm still failing to understand the faith people put into the king of cpus after reading this thread.

 

[Valantar]

I'm still failing to understand the faith people put into the king of cpus after reading this thread.

I think quite a few people have realized by now that a 7700K is only worth buying if you plan to delid (or run it at stock and don't mind high load temps).

 

[LTC8K6]

I'm still failing to understand the faith people put into the king of cpus after reading this thread.

People probably think 80-85C is unacceptable, when it's really not since Tj is 100C for the chip. It will run forever at 85C.
If it's not thermal throttling, the temps are okay.

 

[Czachol]

Ok guys, things went slightly off topic, I guess to each his own when it comes to choosing a brand, ppl have their reasons. I was unlucky enough to get a really hot-running chip, so any overclock was out of question. Am I pissed at Intel? - yes, but what can I do now, and delid was the only option for me. I asked for your opinion about whether my application of CLU could cause problems, like it "drying" and corroding the die over time or oozing/splattering to the nearby row of contacts (which I stupidly left uncoated). Would appreciate it if more ppl helped me here.

I had a lot of experience with CLU between the base of an HSF and the CPU IHS where I would paint on a very, very thin amount. I tried to do the same with the silicon die of my A10-7700k, leading to horrendous thermal contact between the IHS and die after relidding the CPU. Putting on a larger quantity - maybe 2-3 grains of rice worth - fixed the problem nicely.

Wow, Maybe Kaveri has bigger distance die-IHS, but I'm pretty sure that putting 3-rice-grain worth of anything there would result in flooding the entire space under the IHS. Glad that worked in your case though. Also, thanks for the info about liquid metals not retaining bubbles. Didn't know that, but if that's really the case, I guess I am safe.

 

[MrTeal]

People probably think 80-85C is unacceptable, when it's really not since Tj is 100C for the chip. It will run forever at 85C.
If it's not thermal throttling, the temps are okay.

Higher temps can impact stability though, so you may not be able to get the same OC with a chip at 85C as you can with a chip at 60C.
A difference of 20C is pretty huge, people spend a lot of money on custom water cooling loops that can't give a 20C improvement over a $20 Evo 212.

 

[LTC8K6]

Higher temps can impact stability though, so you may not be able to get the same OC with a chip at 85C as you can with a chip at 60C.
A difference of 20C is pretty huge, people spend a lot of money on custom water cooling loops that can't give a 20C improvement over a $20 Evo 212.

I guess. We also have people worried about Prime 95 temps, which they shouldn't be...

 

[DrMrLordX]

Wow, Maybe Kaveri has bigger distance die-IHS, but I'm pretty sure that putting 3-rice-grain worth of anything there would result in flooding the entire space under the IHS. Glad that worked in your case though. Also, thanks for the info about liquid metals not retaining bubbles. Didn't know that, but if that's really the case, I guess I am safe.

I think you will find that many users experience . . . shall we say, shockingly large gaps between the die and IHS. There was a 3770k delid thread on this forum that showed the same phenomenon. I thought of it when doing my delid, and the hunch paid off. Here's the thread:

https://forums.anandtech.com/thread...-temperatures-drop-by-20°c-at-4-7ghz.2261855/

Very good read. I do recommend it.

 

[Czachol]

I think you will find that many users experience . . . shall we say, shockingly large gaps between the die and IHS. There was a 3770k delid thread on this forum that showed the same phenomenon. I thought of it when doing my delid, and the hunch paid off. Here's the thread:

https://forums.anandtech.com/threads/delidded-my-i7-3770k-loaded-temperatures-drop-by-20°c-at-4-7ghz.2261855/

Very good read. I do recommend it.

I am familiar with that monster thread. A true gem of empirical approach to the subject of TIM. As far as I remember, the IHS of his 3770k rested directly on the die after removing excess glue, with a gap now appearing between IHS and substrate instead.

 

[DrMrLordX]

I am familiar with that monster thread. A true gem of empirical approach to the subject of TIM. As far as I remember, the IHS of his 3770k rested directly on the die after removing excess glue, with a gap now appearing between IHS and substrate instead.

Well, point being, if you look at his experience with CLU, you will find that it took him a lot of it to get good results using it on the die. I had the same experience.

 

[aigomorla]

I'm still failing to understand the faith people put into the king of cpus after reading this thread.

Overclocking is a hobby.
Why do people put useless K&N Air filters inside their cars when all it does is just create extra noise?
Because that is also a hobby... and well hobbies makes us do funny things somethings...

I think quite a few people have realized by now that a 7700K is only worth buying if you plan to delid (or run it at stock and don't mind high load temps).

This is SO True, which is why i dont recommend watercooling as much anymore.
You really see no net benifit from the poor TIM/IHS contact u see on LGA115x vs a LGA2011.
Your just best off air cooling it, claming the sink down with a lot of pressure, and calling it a day without the extra maintance a LCS system brings, unless ur on LGA2011.

People probably think 80-85C is unacceptable, when it's really not since Tj is 100C for the chip. It will run forever at 85C.
If it's not thermal throttling, the temps are okay.

*Cringe*
The last time i have had CPU load temps excessive of 70C even was back in 2006 on a prescott...
I like to keep things cooler.
IC's like to be kept cool, the cooler the better, as they work more efficiently and have less loss.

Well, point being, if you look at his experience with CLU, you will find that it took him a lot of it to get good results using it on the die. I had the same experience.

Except he cheated!
he used that cool kit to delid his cpu, while im sure u and i spent a finger painful experience with razor blades prying the sucker off.

I really wanted to buy one of those deliding kits, but they were all sold out and out of stock when i tried to purchase it.

 

[Czachol]

Heh, that's why I just waited patiently a couple of months for the delid tool to be restocked The thought of razors anywhere near the cpu.. brrr

Here's sth you guys may find interesting: Yesterday I decided to stress test it once again, but this time I left Prime95 running longer. After 15 minutes, I noticed an interesting behaviour: the cores 0 and 1 started shifting randomly between 60C and approx. 52C, most of the time hovering about the lower of the values. Then, the stabilized again at 59C or so, and the cores 2 and 3 began to exhibit the same behaviour, within similar range of values. My thought is the CLU is settling in or sth. Well, I shut the PC down and tightened the waterblock a bit more, then let it cool down overnight. In the morning, I ran Prime95 once more and the same thing happened. I am confused. Does this mean the CLU does not make good contact in some places between die and IHS? That I may be able to gain another 8C by putting more CLU there?

 

[LTC8K6]

*Cringe*
The last time i have had CPU load temps excessive of 70C even was back in 2006 on a prescott...
I like to keep things cooler.
IC's like to be kept cool, the cooler the better, as they work more efficiently and have less loss.

The CPU is not staying at those 85C temps. It hits them once in a while under very high loads.
99.9% of the time the temp is much lower.
Unless you keep playing Burn In or Prime 95...

Intel warrants the cpu for 3 years, so that 100C junction temp spec must be somewhat conservative.
Tjmax is the shutdown temp, and I believe it is considerably higher than the 100C throttle temp?

It's mostly high voltage that kills a CPU, rather than high temps, isn't it?

After all the CPU really can't reach damaging temps because it will throttle first, and then shut down if necessary.

 

[aigomorla]

It's mostly high voltage that kills a CPU, rather than high temps, isn't it?

After all the CPU really can't reach damaging temps because it will throttle first, and then shut down if necessary.

you correct, voltage is a faster cpu killer then heat.

however heat has to play with overclocking due to throttling.
The cooler you can keep your cpu the better you can clock it.

every 10C u knock off heat, you effectively double the life of said cpu, which translates to more voltage it can handle.

Also less heat means less voltage leakage, which translates to more efficient overclocking, so the heat a cpu puts out can not be overlooked, especially on lga115x.

I really hated intel's decision on using that gray tim under the IHS.
I mean really, does it save them that much money?
To me it seemed more like a stop gap wall from LGA2011 vs LGA115x where u see intel dressed as gandolf saying "YOU SHALL NOT PASS" unless we brunt force attack it by deliding.

 

[DrMrLordX]

Except he cheated!
he used that cool kit to delid his cpu, while im sure u and i spent a finger painful experience with razor blades prying the sucker off.

I really wanted to buy one of those deliding kits, but they were all sold out and out of stock when i tried to purchase it.

Those kits didn't even exist when I did my delid. I was sweating bullets when I sliced into that black epoxy crap, but eventually I got the hang of it (I used an Xacto which isn't much different than using a razor, I just had a stick attached to it). Chipped the hell out of the blade tilting it up towards the IHS, heh. Fortunately it worked, and it bought me 200 MHz max OC on my Kaveri.

 

[Czachol]

Heh, that's why I just waited patiently a couple of months for the delid tool to be restocked The thought of razors anywhere near the cpu.. brrr

Here's sth you guys may find interesting: Yesterday I decided to stress test it once again, but this time I left Prime95 running longer. After 15 minutes, I noticed an interesting behaviour: the cores 0 and 1 started shifting randomly between 60C and approx. 52C, most of the time hovering about the lower of the values. Then, the stabilized again at 59C or so, and the cores 2 and 3 began to exhibit the same behaviour, within similar range of values. My thought is the CLU is settling in or sth. Well, I shut the PC down and tightened the waterblock a bit more, then let it cool down overnight. In the morning, I ran Prime95 once more and the same thing happened. I am confused. Does this mean the CLU does not make good contact in some places between die and IHS? That I may be able to gain another 8C by putting more CLU there?

bump. Whatcha think? Could this be because I tightened the clamping screw during relidding too much, and upon de-clamping the relidded CPU, the CLU was somehow disturbed?

 

[DrMrLordX]

It's possible that you're having contact problems between the die and IHS. I know I had problems, but mine were just bad temps across the board until I just slapped on more CLU.

 

[Valantar]

I have to ask: is a 17-degree improvement really bad enough to warrant the question "did I mess this up"? Doesn't that temp drop speak for itself?

 

[RichUK]

Your application of CLU is absolutely fine. The air bubbles are unavoidable when applying the CLU and when you clamp/compress the IHS back on it will remove any residual air bubbles.

As long as you don't apply too much CLU, you don't have to worry about it seeping out onto the exposed contacts. I never covered the contacts and my 7700k has been running flawlessly at 5.1ghz for over 5 months with no degradation in temps.

 

[Czachol]

Guys, you have been incredibly helpful. I guess I will leave it be. Now I only wonder if it is possible to reinstall/change waterblock atop the CPU without the need to redo the delidding procedure or maybe somehow the CLU will get disturbed and loose its tight bond? I have a regular TIM between IHS and waterblock.