delidded my 8700k, amazing temp drop

Apone70 · 18 Oct 2018 at 16:19

Just delidded my 8700k and used Conductonaut between the die and IHS, then used Kryonaut between IHS and heat sink (Corsair 100i V2) My temps were around 83 using Aida64 CPU stress and now they are 55 which is a huge drop. My CPU does need more volts than others I've seen to stay stable to 5ghz and i think my motherboard is pants which doesn't help (Gigabyte Ultra Gaming 2).
Anyone thinking of doing it just do it. Its really easy using the der8auer delid tool and worth the huge drops in temps

BluD · 19 Oct 2018 at 17:42

Its nice aint it

, my last two chips before my current one ran really hot before i delidded too.

Hopefully it will soon be a thing of the past now Intel are soldering their mainstream chips again.

toxicredcat · 19 Oct 2018 at 17:55

Glad to hear that it made such a difference as I plan on doing mine in the next week or so.

Ordered a copper IHS to replace the stock one as well as reviews show that also helps a tiny bit.

Journey · 19 Oct 2018 at 17:56

BluD said:
Hopefully it will soon be a thing of the past now Intel are soldering their mainstream chips again.

You would have though that right, but no, they should have left them with paste on for easier delidding it would seem.

Apone70 · 19 Oct 2018 at 18:41

Yeah im really happy with the results, i was bench marking earlier and had all my case fans at max. I idle temp was 25. I cant get 5.1ghz stable even at 1.4 volts. 5ghz is solid at 1.38 but when i test it with Cinebench its only using 1.25

BluD · 20 Oct 2018 at 10:08

Yeah if your at 5ghz and 1.38v then i wouldnt bother trying for 5.1ghz, it would probably require 1.45v to get it stable, it would probably be fine for some benchmarking but ya wouldnt wanna run high volts all the time.

Apone70 · 20 Oct 2018 at 13:16

Kieran said:
Nice result. I delidded mine on Thursday and dropped 15-20C. Previously I was running 5ghz core/-2avx/4.6ghz cache @ 1.28v (LLC 6 Asus board) Temps stopped me bumping voltage higher. Now I'm @ 1.35v at 5.2ghz core/-2avx/4.9ghz cache at the same or lower temps.

Thats decent, i think my board is holding me back. Might have to treat myself for xmas, the wife keeps asking what i want. Might go the Z390 route

Apone70 · 20 Oct 2018 at 13:47

what i don't get though is if i run blender (BMW) then my vcore is 1.26 all the way through but it i still have to set the volts to 1.38 in the bios for it to be stable.
I get 1607 in cinebench and again its shows 1.26v while i perform the test. Think its just the rubbish Ultra gaming 2 mobo?

Bowie92 · 21 Oct 2018 at 15:34

Apone70 said:
what i don't get though is if i run blender (BMW) then my vcore is 1.26 all the way through but it i still have to set the volts to 1.38 in the bios for it to be stable.
I get 1607 in cinebench and again its shows 1.26v while i perform the test. Think its just the rubbish Ultra gaming 2 mobo?

Have you manually set voltage or is it ok automatic? If automatic play around with manual and try to find the lowest stable voltage. If manual see if you can change LLC (load line calibration) which affects how the motherboard responds under load. This is usually different depending on manufacturer so have a Google and find what the options are are for your mobo. Ideally you want the voltage to remain the same under load.

BluD · 21 Oct 2018 at 17:14

Apone70 said:
what i don't get though is if i run blender (BMW) then my vcore is 1.26 all the way through but it i still have to set the volts to 1.38 in the bios for it to be stable.
I get 1607 in cinebench and again its shows 1.26v while i perform the test. Think its just the rubbish Ultra gaming 2 mobo?

yeah its normal, use LLC to add extravoltage under load to help prevent it dropping so much like Bowie92 says.

Apone70 · 21 Oct 2018 at 19:13

Yeah the voltage is manually set. It's stable even though the voltage drops during stressing but I don't get why I can't get it stable at that voltage (1.26) in the bios instead of having to go upto 1.38

BluD · 21 Oct 2018 at 20:44

Because its normal for your voltage to drop lower than what you set it in the bios when stressing the CPU, if you dont want it to do so then you need more LLC, find LLC in your voltage settings and increase it, don't forget to lower your vcore also though. Try high LLC first, if thats not enough then try turbo. I use high on mine.

Apone70 · 21 Oct 2018 at 21:22

As long as it's normal then that's fine. I saw a certain site slating my motherboard after I had already brought it saying it's awful for overclocking and the VRMs get crazy hot. I haven't seen the vrms go over 75 degrees. As long as the drop in volts while stressing is normal I'm happy

Distracted · 21 Oct 2018 at 21:25

Awesome result. Shame you had to do it (intel :mad:

) but i bet it feels good!

Apone70 · 21 Oct 2018 at 21:26

I'm a bit out of the loop. I built this system 6 months ago after I quit PC gaming. Back on my Pentium 4 1.6ghz it was much simpler.... thin copper wire round a couple of pins to get loads more volts and overclock to crazy speeds lol

Apone70 · 21 Oct 2018 at 21:28

Distracted said:
Awesome result. Shame you had to do it (intel ) but i bet it feels good!

There doesn't seem much of an improvement if any with the soldered Intel chips to be honest. Thicker die for the soldering process and high temps again

Smffy · 23 Oct 2018 at 22:18

toxicredcat said:
Glad to hear that it made such a difference as I plan on doing mine in the next week or so.

Ordered a copper IHS to replace the stock one as well as reviews show that also helps a tiny bit.

If you sand the stock one back you will find it’s copper also just lap the Nickel off the inside

Distracted · 23 Oct 2018 at 22:56

Smffy said:
If you sand the stock one back you will find it’s copper also just lap the Nickel off the inside

This is true.

I think the copper replacement ihs they are talking about actually has a larger surface area. If it is the same one gn tested then it reduced temps by a few degrees over the stock nickle-plated copper ihs. That's probably down to the increased surface area.
Whatever the reason, that doesn't seem much improvement in isolation, but it all adds up.

toxicredcat · 24 Oct 2018 at 22:15

Distracted said:
This is true.

I think the copper replacement ihs they are talking about actually has a larger surface area. If it is the same one gn tested then it reduced temps by a few degrees over the stock nickle-plated copper ihs. That's probably down to the increased surface area.
Whatever the reason, that doesn't seem much improvement in isolation, but it all adds up.

Just fitted the copper IHS after attacking the CPU with a Stanley blade (too impatient to wait for the delid tool XD). The copper one does have a greater surface area in contact with the CPU heatsink which is bound to help somewhat. Its dropped my temps down 15°C on average with it now peaking around 60°C under load.

Also fitted a NZXT X42 AIO to my 1080ti which is managing to keep it around 50°C. PC is now much cooler and a lot quieter at long last!