Delidding 7700K

[tronder]

Hi All! 

 

I have a watercooled loop and not getting good temps on my i7 7700k and was wondering if you guys thought if it was worth delidding to get better temps. 

There's nothing wrong with my loop as my gtx 1080 is sitting at 32 degrees idle and dont think goes above 50 when under load.

Tossing up wether the temps would be worth it considering you void warranty... Any ideas?

[TrigrH]

i would relid:

 

[19_blackie_73]

remember that you'll not only voiding warranty but also take the risk loosing $400. Delidding can damage the die itself and therefore it will not work any longer.

Can you risk loosing $400: then try it if you're after max oc

Are you screwed if you loose $400 or don't want to squeeze every last single bit of performance and oc out of it: let it be

 

EDIT: what are your temps on cpu?

[JoostinOnline]

4 hours ago, tronder said:

Hi All! 

 

I have a watercooled loop and not getting good temps on my i7 7700k and was wondering if you guys thought if it was worth delidding to get better temps. 

There's nothing wrong with my loop as my gtx 1080 is sitting at 32 degrees idle and dont think goes above 50 when under load.

Tossing up wether the temps would be worth it considering you void warranty... Any ideas?

What temps are you getting on the CPU?

 

[TrigrH]

1 minute ago, JoostinOnline said:

Relidding defeats the whole purpose.

no it doesn't, Intel's TIM is garbage and when replaced with something decent you can reduce temps by 15-25 C.

[JoostinOnline]

2 minutes ago, TrigrH said:

no it doesn't, Intel's TIM is garbage and when replaced with something decent you can reduce temps by 15-25 C.

From what I've seen, more like 5C.

[tronder]

Well atm with no overclock idle is lie 35 degrees and under cinebench load is about 70 degrees but with a 5GHZ OC at 1.35V its like 40-45 idle and 85-95 under load...

44 minutes ago, 19_blackie_73 said:

-SNIP-

39 minutes ago, JoostinOnline said:

-SNIP-

 

[JoostinOnline]

3 minutes ago, tronder said:

Well atm with no overclock idle is lie 35 degrees and under cinebench load is about 70 degrees but with a 5GHZ OC at 1.35V its like 40-45 idle and 85-95 under load...

Lol I'm surprised you were even able to OC it to 5GHz.  You are getting expected temps, not bad ones.

[19_blackie_73]

1 hour ago, tronder said:

Well atm with no overclock idle is lie 35 degrees and under cinebench load is about 70 degrees but with a 5GHZ OC at 1.35V its like 40-45 idle and 85-95 under load...

well first off, congrats that you achieved 5GHz. That temps are for a custom loop very high tbh. Though you really have to consider whether you can buy in worst case a new 7700k because something went wrong. What I've heard, 10-20°C less should be possible. I really recommend to use some sort of delid tool that is proven to work and this would minimalize risk of damaging anything. And maybe try first figure out if you messed up by waterblock installation and if theres enough flow through your cpu block (don't know your loop layout  ) So maybe reseat the cpu block and reapply thermal paste (check if the old one is distributed well and only a thin layer) before voiding any warranty.

 

BTW what parts are you using for block, rads etc?

[Lays]

2 hours ago, JoostinOnline said:

What temps are you getting on the CPU?

Relidding defeats the whole purpose.

no it doesn't lol

[JoostinOnline]

22 minutes ago, Lays said:

no it doesn't lol

The point is to keep it as cool as possible by getting rid of the IHS, not drop the temps 5C.

[Lays]

8 minutes ago, JoostinOnline said:

The point is to keep it as cool as possible by getting rid of the IHS, not drop the temps 5C.

Temps don't drop 5c by delidding, when done properly, it's 15-25c. and getting rid of the IHS does next to nothing for temps, bare die vs IHS + CLU is only a few C difference, and the giant risks associated with bare-die are not worth.

 

You must be extremely uninformed if you think people that delid for temp decreases ditch the IHS and go straight for bare die.

[Lays]

3 hours ago, tronder said:

Hi All! 

 

I have a watercooled loop and not getting good temps on my i7 7700k and was wondering if you guys thought if it was worth delidding to get better temps. 

There's nothing wrong with my loop as my gtx 1080 is sitting at 32 degrees idle and dont think goes above 50 when under load.

Tossing up wether the temps would be worth it considering you void warranty... Any ideas?

Delidding doesn't void the warranty, you can re-seal your IHS before you send it back in, and they won't know. I know a few people that have done it with no problems.

[XenosTech]

3 hours ago, JoostinOnline said:

What temps are you getting on the CPU?

Relidding defeats the whole purpose.

Lol must be new to the interwebz.... A delid and relid done correctly does a lot for temps

[JoostinOnline]

4 minutes ago, XenosTech said:

Lol must be new to the interwebz.... A delid and relid done correctly does a lot for temps

Maybe I've only seen it done incorrectly.  I'm not above admitting when I'm wrong.

[MageTank]

4 hours ago, JoostinOnline said:

Maybe I've only seen it done incorrectly.  I'm not above admitting when I'm wrong.

I don't really blame you. LTT has mislead the masses with their god awful delid video, and people keep spreading that misinformation because of it. What people fail to realize is, he is still human and is prone to making mistakes. It's just sad that he refuses to admit when he makes those mistakes, or to actually use a proper testing methodology and re-visit the videos that gets called out.

 

I just delidded my cousins 6600k yesterday evening. At 4.5ghz 1.3v, he was hitting 89C peak under 48k FFT Prime95 (AVX2) before the delid. After the delid + CLU under the exact same stress test, his max peak temperature was 64C. I was then able to bring the vCore down to 1.23v, which then resulted in his max temperature being 55C. Now, it's unlikely that the delid aided in me being able to lower the volts, mostly because when I originally overclocked it, we did it quick and dirty just so he could play games sooner. My point is, at the exact same voltages, we saw a 25C difference with the only change being the delid + CLU. 

 

He saved before and after pictures on his PC, so if you would like to see them, i'll have him text those to me. 

[JoostinOnline]

3 hours ago, MageTank said:

I don't really blame you. LTT has mislead the masses with their god awful delid video, and people keep spreading that misinformation because of it. What people fail to realize is, he is still human and is prone to making mistakes. It's just sad that he refuses to admit when he makes those mistakes, or to actually use a proper testing methodology and re-visit the videos that gets called out.

 

I just delidded my cousins 6600k yesterday evening. At 4.5ghz 1.3v, he was hitting 89C peak under 48k FFT Prime95 (AVX2) before the delid. After the delid + CLU under the exact same stress test, his max peak temperature was 64C. I was then able to bring the vCore down to 1.23v, which then resulted in his max temperature being 55C. Now, it's unlikely that the delid aided in me being able to lower the volts, mostly because when I originally overclocked it, we did it quick and dirty just so he could play games sooner. My point is, at the exact same voltages, we saw a 25C difference with the only change being the delid + CLU. 

 

He saved before and after pictures on his PC, so if you would like to see them, i'll have him text those to me. 

I'll take your word for it. However, could you fill me in on what Linus did wrong? I thought it was just putting the IHS back on (based on YouTube comments). 

 

I'm also like 99% sure I've seen it say elsewhere that reapplying the IHS only saved a few degrees. I figured that's why laptops almost never have them (I mostly fix laptops in my work). 

[MageTank]

2 hours ago, JoostinOnline said:

I'll take your word for it. However, could you fill me in on what Linus did wrong? I thought it was just putting the IHS back on (based on YouTube comments). 

 

I'm also like 99% sure I've seen it say elsewhere that reapplying the IHS only saved a few degrees. I figured that's why laptops almost never have them (I mostly fix laptops in my work). 

He did quite a few things wrong, but the most important that I take issue with, was his paste of choice (CM Maker Gel, known to not handle thermal shock very well), his refusal to spread the paste (IHS doesn't put enough mounting pressure on the die to evenly spread what little paste he put on the die), and his failure to acknowledge the impact the delid had on his per-core temperatures. He was looking at a 13C difference between his coolest and hottest core pre-delid, and only a 5-6C difference between hottest/coolest afterwards. That is primary why we delid, when per-core variance is intolerable (10C+ difference). He was also factually wrong, blaming the paste (Dow Corning 2688) when it's not the fault of the paste, but the automated glue process. The glue itself is thick, and expands overtime, creating pockets in between the die and IHS. If you've ever delidded a chip, and noticed a bare spot on your die completely devoid of paste, you will understand why that specific core was getting hot. 

 

He completely downplayed the significance of delidding, and did not properly educate the masses as to why it's done, and the full impact it has. Is delidding for everyone? No, not by any means. Not everyone overclocks, and not everyone ends up with a lemon of a chip that needs to be delidded in order to operate normally. However, some people lose the lottery hard (my friend Darl did) and others simply want their CPU's to run cooler to prolong their lifespan or get more thermal headroom for better overclocks. 

 

As for laptops: Laptops often don't need them. They go bare-die because it allows for a thinner design. Remember, if you have an IHS on your CPU, you need a retention mechanism on top of the IHS to keep it socketed and to help distribute the weight of the heatsink. Laptops like the Sager NP9873 have socketed 6700k's or 7700k's and still have the IHS and retention mechanism on them (and is way thicker than your standard laptop). As for thermal savings, you see a difference of maybe 3-5C at best, going naked vs keeping the IHS on, but the risk is far greater when going that route. It's simply not worth it. That being said, you can totally put CLU on the bare die of a laptop, as long as it's heatsink isn't aluminum. Gallium turns aluminum into chalk, lol. 

 

If you need the full rundown of what Linus did wrong with that delidding video, a ton of us wrote about it in that video's forum post. Feel free to give it a read:

 

[JoostinOnline]

3 minutes ago, MageTank said:

He did quite a few things wrong, but the most important that I take issue with, was his paste of choice (CM Maker Gel, known to not handle thermal shock very well), his refusal to spread the paste (IHS doesn't put enough mounting pressure on the die to evenly spread what little paste he put on the die), and his failure to acknowledge the impact the delid had on his per-core temperatures. He was looking at a 13C difference between his coolest and hottest core pre-delid, and only a 5-6C difference between hottest/coolest afterwards. That is primary why we delid, when per-core variance is intolerable (10C+ difference). He was also factually wrong, blaming the paste (Dow Corning 2688) when it's not the fault of the paste, but the automated glue process. The glue itself is thick, and expands overtime, creating pockets in between the die and IHS. If you've ever delidded a chip, and noticed a bare spot on your die completely devoid of paste, you will understand why that specific core was getting hot. 

 

He completely downplayed the significance of delidding, and did not properly educate the masses as to why it's done, and the full impact it has. Is delidding for everyone? No, not by any means. Not everyone overclocks, and not everyone ends up with a lemon of a chip that needs to be delidded in order to operate normally. However, some people lose the lottery hard (my friend Darl did) and others simply want their CPU's to run cooler to prolong their lifespan or get more thermal headroom for better overclocks. 

 

As for laptops: Laptops often don't need them. They go bare-die because it allows for a thinner design. Remember, if you have an IHS on your CPU, you need a retention mechanism on top of the IHS to keep it socketed and to help distribute the weight of the heatsink. Laptops like the Sager NP9873 have socketed 6700k's or 7700k's and still have the IHS and retention mechanism on them (and is way thicker than your standard laptop). As for thermal savings, you see a difference of maybe 3-5C at best, going naked vs keeping the IHS on, but the risk is far greater when going that route. It's simply not worth it. That being said, you can totally put CLU on the bare die of a laptop, as long as it's heatsink isn't aluminum. Gallium turns aluminum into chalk, lol. 

 

If you need the full rundown of what Linus did wrong with that delidding video, a ton of us wrote about it in that video's forum post. Feel free to give it a read:

 

Wow, thanks for that info!

 

I knew very little about delidding when I first saw the video, but I could tell immediately that he was doing it as nothing more than a cautionary tale.  It reminded me of a government anti-pot commercial.

 

I wasn't the least bit surprised when people in the comments were saying it wasn't effective because he put the IHS back on, and when I looked up guides they all said to leave the IHS off.

[MageTank]

1 minute ago, JoostinOnline said:

Wow, thanks for that info!

 

I knew very little about delidding when I first saw the video, but I could tell immediately that he was doing it as nothing more than a cautionary tale.  It reminded me of a government anti-pot commercial.

 

I wasn't the least bit surprised when people in the comments were saying it wasn't effective because he put the IHS back on, and when I looked up guides they all said to leave the IHS off.

As much as I hate tooting my own horn, I have a thread about my first time delidding, if you'd like to see my results. It was a while ago, but still applicable: 

I even show screenshots of the CPU and what my CLU looks like on the bare die and IHS. I'll be re-doing my CLU within the next month (not because I have to, as temperatures are exactly the same now as the first day I put it on) because I'll be sanding the under-side of my IHS down to the copper. I'll also document what it's like to clean the CLU off, and if it leaves any residue or stains on the CPU. 

[tronder]

On 2/13/2017 at 8:39 PM, 19_blackie_73 said:

SNIP

 

BTW what parts are you using for block, rads etc?

Ive got a EK D5 Revo combo pump 200mm RES i think => Hardline to 1080 with EK waterblock and backplate => 240 bitspower thin rad => CPU EK evo supremacy (and i did observe the in and out labels for ports) => bitspower 360 thick rad => RES
 

i already tried reseating the CPU block after taking 1 hour to drain the bloody loop, so dont think its anything to do with the loop itself unless the bitspower Rads are crap?

[19_blackie_73]

3 hours ago, tronder said:

Ive got a EK D5 Revo combo pump 200mm RES i think => Hardline to 1080 with EK waterblock and backplate => 240 bitspower thin rad => CPU EK evo supremacy (and i did observe the in and out labels for ports) => bitspower 360 thick rad => RES
 

i already tried reseating the CPU block after taking 1 hour to drain the bloody loop, so dont think its anything to do with the loop itself unless the bitspower Rads are crap?

I don't think the rads are crap, and you have enough rad space to cool. Then probably the only chance to bring down temps will be to delid. Though be aware of the risks

[tronder]

6 hours ago, 19_blackie_73 said:

I don't think the rads are crap, and you have enough rad space to cool. Then probably the only chance to bring down temps will be to delid. Though be aware of the risks

slightly dissapointed with the way intel handled the heat transfer. Makes no point in having a custom loop if you limited by the thermal transfer of the IHS. And now rumours that theyre gonna release at 7700k sucessor already to combat AMD ryzen....

[sl06bhytmar]

Just get tool that doesn't cost a fortune and begin delidding.
https://www.entersetup.com/

 

Delidding improves temps 10-20C degrees on AIDA64/Prime/IBT/Linx when using LIQUID METAL between die and IHS.

 

Disclaimer: There is always a chance to destroy your CPU when delidding, usually user errors but sometimes glue just won't give up and so much pressure builds up that IHS just gives out with louder bang and hits die also. Tools should prevent this from happening but heck if it is possible to do in anyway then it will happen [Murphy's law]

[ToastyBear]

A month ago I watched the Linus video and thought "Wow delidding is dumb"

Then I bought a 7700k and was like "holy shit this CPU runs hot" 

 

Its not cooling because im on full water with a 480 and 280 rad. 2 seconds into prime95 and my CPU flashes to 85C. That is an immediate jump while my water temp was still around 20c room temp. Clearly heat wasn't getting transferred efficiently from the CPU to the water. Doesn't mean the IHS is shit but lets do some science and find out. Sign me up for a delidding tool and liquid metal.

 

After the delid I noticed significant temperature differences. At idle its nothing but at load its up to 15c. 2 seconds into prime95 flashes my temps to 70c instead of 85c. 15c difference. Now I have to say I did a shit job of applying liquid metal. I was sure i need to do it again but was too lazy anyway. Still got the temps.