anyway we have to factor/count the "heat spreader" right ? or they put back after reapplied new TIM ? temp before delidding is + heat spreader then temp after delidding is without heat spreader right ? cmiiw (just like GPU) i am wondering why gpu never cased with heat spreader while cpu need to be cased with heat spreader i understand without heat spreader, its fragile, that can probably broke under sunlight bath(uv)? or other stuff... but it should can be solved with right packaging and for the article regarding liquid pro this article right ? http://pc.watch.impress.co.jp/docs/column/sebuncha/20150806_715335.html rough translate high risk - high return before delid, it tested on 1.325v 4.6Ghz , with prime95 28.5 small ffts, cpu temp is near 90C so even without delidding, this CPU is really great OC CPU but after seeing overwhelming different, there is value in risk in delidding put aside to what point it can be OC, the temp different (intel TIM vs liquidpro) can simply translate to around extra 100~200mhz clock rate in daily use the other thing that they mention about Cool Laboratory Liquid Pro why liquid pro make so different is that, 14nm die size as its better thermal conductivity, it transfer heat from small size die better than other TIM
Where are you seeing this? I had a 3770K and delidded with both Noctua NT-H1 and Liquid Ultra. The results were within 2-3c and most of the time simply just re-applying the Noctua stuff changed the temps within 4-5c +- each time. The vast majority of the difference was simply shaving down the adhesive and closing the gap. http://forums.anandtech.com/showpost.php?p=34053183&postcount=570
When I switched from Phoyba Extreme HeGrease on Die to Liquid Pro on Die I saw a 5-10c drop in temps. Not quite as high as Agent is saying but still a nice drop there.
http://pc.watch.impress.co.jp/docs/column/sebuncha/20150806_715335.html This is the original news. Their interpretation is that (1) The difference between Polimatch PK-3 and control (standard TIM) was small because Intel used next generation polymer TIM which was used for Devil's Canyon, to 6700K as well. (2) The difference between Cool Laboratory Liquid Pro (Thermal conductivity 82.0W/(m x K)) and PK-3 (Thermal conductivity 8.5W/(m x K)) was astonishingly great probably because the die size of Skylake (14nm) was very small. When the die area is small the thermal resistance increases. Therefore it requires high thermal conductive paste.
With my i5-4670k I saw a drop of 30'c under full load. My chip was approaching 90'c at 4.2ghz and after delidding, shaving the underside lip of the IHS that contacts the PCB so the core made contact with the IHS, and lapping the IHS surface that contacts the heatsink, and applying liquid metal to each surface... the hottest my CPU ever gets now on air cooling is 56'c @ 4.2ghz. The problem with my CPU was there was a gap between the core and IHS and that had a glob of thermal compound filling it. Maybe that was a defect, but it's not like you can return the chip after finding out. That's pretty substantial. You live in Canada. The rest of the civilized world doesn't have the luxury (if you want to call it that) of a natural sub-zero cooling system just outside our window. Being from Southern California, we have perhaps a handful of days per year that get below freezing. Most of the winter we average in the 50's. If I lived in where you do, i'd reverse flow my fans so they'd blow from rear to front and hook a duct from the window to the rear of my case. Either that or water cool my PC and keep the radiator outside the house. hehe. edit\ Maybe that's a bad idea. Condensation is a problem if water got chilled too much.
The problem with intel's design is a physical gap between the CPU core and underside of IHS. This is resolved by either completely removing the IHS or the safer method, shaving the lip that contacts the CPU PCB so the IHS sits lower, thus making proper contact. Once this is done, the only difference you'll find between various thermal compounds is what is shown on countless "thermal compound roundup" articles around the net. It's only a few degrees if you go with Coolaboritories liquid metal. I used the liquid metal regardless because I figured this was a "one-and-done" mod and wasn't planning on replacing dried out AS5 at regular intervals.
Yes, the gap is the problem. And it is NOT resolved by removing the adhesive and using something thinner. With my 4670k, even if the IHS sat completely flush against the PCB, the gap was substantial. You have to physically remove metal from the underside lip of the IHS so it sits lower and reduces the gap between the CPU core and IHS. This link shows that just removing adhesive isnt enough: http://forums.anandtech.com/showpost.php?p=34053183
I've been wanting to use my new vice for a couple of weeks now. All this jibber-jabber made me take my 4790k to the garage for the treatment. It's made a nice difference. Running 8 Prime inplace FFTs it hovers ~70C. 4.6ghz@ 1.26v Because the h80i is temp controlled, it's only a few degrees cooler but the fans are running at a lot lower RPM. If anybody has these Corsair fans, they'll know how nice that is!
Not in my experience, i have delidded 3 times.. A 3770k and a 4670k,4770k. My 3770k i lapped both the underside and the top and replaced with gelid gc extreme or PK-3 i forget, i dropped 7c at load. Nothing spectacular. Put the liquid pro and my total temp drop went to 28c. Same for the other chips, regular tim didnt drop much, liquid ultra dropped 20+c on all chips. I shaved off all the glue as well. TLDR dont read what you see on the internet as fact as my personal testing disproves that "fixing" the gap makes liquid ultra useless Check the delid thread on OCN and youll see many people getting MUCH better results with liquid ultra
Solution one is just going to fill your case with dust much quicker and solution two just lets damp air into your case causing corrosion to start making a foothold on your expensive PC components causing failure to happen much sooner the cool laboratories did make a considerable difference in temps when loaded so it just makes me wonder why Intel don't use superior tim under their IHS's god knows you pay enough for their CPU's