WYP
News Guru
Japanese Publication PC Watch has delidded Skylake, revealing a tiny CPU die and improving thermals dramatically.
Read more on Delidding Intel Skylake.
Read more on Delidding Intel Skylake.
Skylake delidding, does it improve thermals?
- Thread starter WYP
- Start date
Thelosouvlakia
Classy-Fried-Chicken
WOW
those temps with the coolaboratory stuff!!!!!
However I've been wondering, why doesn't intel use a better TIM on the die? Ok silicon lottery is silicon lottery but Intel can improve the TIM so everyone gets lower temps, they could partner up with coolaboratory and use that stuff for the i7s
Unless they plan doing something similar with the 1150 Haswels and Devils Canyon....
those temps with the coolaboratory stuff!!!!!
However I've been wondering, why doesn't intel use a better TIM on the die? Ok silicon lottery is silicon lottery but Intel can improve the TIM so everyone gets lower temps, they could partner up with coolaboratory and use that stuff for the i7s
Unless they plan doing something similar with the 1150 Haswels and Devils Canyon....
WYP
News Guru
I imagine it is to keep costs down, though it also forces high end overclockers to break their warranties in a way intel can obviously spot.
the thing is that that changing the way they do thing doesn't help Intel much, after all they are literally the only horse in the high end market, so why compete with yourself.
A bit cynical, but true.
Dicehunter
Resident Newb
If they spent maybe 20 cents more on each CPU's TIM, Thermals wouldn't a be a problem.
Hell they could even add that 20 cents on to the end cost of the CPU.
Hell they could even add that 20 cents on to the end cost of the CPU.
SuB
The Webmaster
20 cents in a 'this is what it costs to do' end does not equal 20 cents at the customer buying the chip end.
There is an awful lot of logistics you have to take into account and that will all multiply out to quite a significant bit of cost to the end-user eventually. Nothing is cheap when it gets into the final user's hands.
Surprised they didn't try ultra as well as pro
20 degrees is pretty ridiculous though, that paste might as well be flipping mayonaise at this point
There is an awful lot of logistics you have to take into account and that will all multiply out to quite a significant bit of cost to the end-user eventually. Nothing is cheap when it gets into the final user's hands.
Surprised they didn't try ultra as well as pro
20 degrees is pretty ridiculous though, that paste might as well be flipping mayonaise at this point
Kleptobot
Member
It's not just the thermal compound that was altered, the material that secured the IHS to the pcb was removed. I remember that with ivy bridge the gap between the IHS and the silicone was relatively large and removing that black material reduced the gap. So yes I agree that the thermal compound used does make a difference I but don't believe that this test shows the full picture.
The difference between the stock Intel tim and the PK-3 could quit possibly be attributed to this change. And I would hardly consider cool laboratory products cheap. Especially considering it would most likely require retooling some production machinery
The difference between the stock Intel tim and the PK-3 could quit possibly be attributed to this change. And I would hardly consider cool laboratory products cheap. Especially considering it would most likely require retooling some production machinery
Last edited:
Dicehunter
Resident Newb
I've worked for enough companies throughout Europe *Not theoretically like a lot of people but in actual practice* to know this type of thing would cost Intel pennies per CPU to fix/implement and wouldn't cost the end user anything more other than maybe a few dollars because they would just be replacing the TIM they use now with a slightly better one which yields much better results.
I can only think of 2 reasons why they haven't implemented this yet.
1. They have a contract with the people who make their TIM so can't do anything yet.
2. They simply don't want to or care.
I can only think of 2 reasons why they haven't implemented this yet.
1. They have a contract with the people who make their TIM so can't do anything yet.
2. They simply don't want to or care.
I refuse to believe this is because money. The TIM they use must be advantageous for the manufacturing process or longevity in some way. Or perhaps artificially hindering overclocking is actually beneficial for them. My point is, it's clearly been engineered to be as good as it is for a reason that is not financial.
JR
JR
Kleptobot
Member
The thing is the stock tim performs perfectly well at stock speeds. I see no reason why Intel would spend the engineering teams time researching, validating and implementing such a change. Then the added costs of retooling for something that gains them little net profit. Sure if competition was more fierce and they were looking for any possible advantage then we might see stuff like this but otherwise it will always be an afterthought
Kleptobot
Member
The point is that they don't need an excuse. This is the same as saying that they have no excuse for not setting the stock clock speed to 4.6GHz because you can overclock them that high.
The most logical reasoning that Intel doesn't use the CLLP or ultra is that it's a conductive element. Beyond that, as someone who has used the pro version for an ivy bridge delid, I can personally say that it's an utter pain to use. The ultra I've heard is more "paste-like" though from what I understand. Regardless, what everyone seems to be missing is that the PK-3 is actually pretty close to the stock intel tim. Anyone remember when switching out the Ivy-bridge internal past yielded near 20 degree benefits? This test seems to show how close the stock intel TIM has come to aftermarket TIM compared to previous iterations. I really don't think comparing any nonconductive thermal paste to cool laboratory stuff is really fair. It consistently is the go to TIM for absolute best performance. The Test seems a bit misleading to me.
The other thing that isn't mentioned, and I may have missed it, is that it doesn't say if the CLLP and PK-3 were used both between the die/IHS and IHS/heatsink. I believe most people use the CL stuff on the die and normal paste on the heatsink. The liquid pro doesn't clean off of the IHS or copper well without an abrasive.
The other thing that isn't mentioned, and I may have missed it, is that it doesn't say if the CLLP and PK-3 were used both between the die/IHS and IHS/heatsink. I believe most people use the CL stuff on the die and normal paste on the heatsink. The liquid pro doesn't clean off of the IHS or copper well without an abrasive.
Last edited:
Thelosouvlakia
Classy-Fried-Chicken
DarthDudo
New member
Yea I was just about to mention Tolerances, the 3570K I killed was pretty bad you could clearly see one side of the IHS to be mm higher than the other also the TIM(Not the Glue/Epoxy/Crap) seems to be similar to CL Pro I definitely think the big gains are from the removal of the glue hence better contact with IHS.
Off Topic: Local hardware store wanted 45euro for a 4" Bench vice, ridiculous!
Agost
New member
The TIM used by Intel isn't bad, the problem is still in the distance between the die and the IHS
The CLP gives better results mainly because it creates a very thin film, while normale pastes are more difficult to spread evenly, resulting in a thicker layer
In fact the PK3 only yelds a 4°C improvement. And it's probably because the layer is thinner than the stock one, derived from a blob of thermal paste spread only by the IHS while it gets assembled
The CLP gives better results mainly because it creates a very thin film, while normale pastes are more difficult to spread evenly, resulting in a thicker layer
In fact the PK3 only yelds a 4°C improvement. And it's probably because the layer is thinner than the stock one, derived from a blob of thermal paste spread only by the IHS while it gets assembled
AgustinXtreme
New member
Excuse me guys but i think we are all wrong about the real causes of this annoying overheating problem on the last Intel CPU generations, it's not rellated to thermal interface, take a look at this:
I found these graphics on anandtech forum, some guy thougth the reason could be other and let me say i did a similar experience trying to put some glue between the IHS and PCB and reached the same conclusion, themperatures went worse directly proportional to the gap between them despite the TIM used.
I found these graphics on anandtech forum, some guy thougth the reason could be other and let me say i did a similar experience trying to put some glue between the IHS and PCB and reached the same conclusion, themperatures went worse directly proportional to the gap between them despite the TIM used.
