I recently purchased a gaming computer. I have got the following components ::
Intel i7 7700K
Corsair H100i V2
Gigabyte GTX 1080 G1
Samsung 850 EVO
ISSUE :: The AIO is set as intake fan, with speed kept at stock settings. The package temp goes to max 77C, up to 4.6 GHz and 1.35+V max while playing GTA V for couple of hours. I set the fan speed of AIO to full, sounds horrible, but temps come down to 73-75C. Which is still extremely high.
First :: Do I need 1.35V for 4.6GHz ? I think not, let me know what are the things I need to do in BIOS to limit it to 1.25/1.28V to test the stability at 4.6 GHz.
Second :: Lets cut to the chase directly. EVERYONE is shouting on top of their voice that Intel went cheap on the thermal paste between the die and heat-spreader. Delidding shows up to 20C reduction in temps. What you guys are planning to do about it ? Its a brand new system I bought only 2 weeks back, should I return the CPU and Motherboard and get a AMD Ryzen system instead or are you guys going to do something about it ?
I want concrete answers, this processor is almost 25% costlier in India than it is in US. I've paid a lot for the best product and I didn't cheap out going for a 7600K. I want value for the money I spent or I am ready to take any action you suggest which includes consumer court and/or going vocal on social media. I've been searching the forums all day, and I don't see any solid response from Intel on replacing these cheap rubbery glue that is inside instead of a quality thermal compound. If delidding solves this "problem" but voids warranty, and I cant overclock this unlocked CPU with an AIO, then you guys really went "way too arrogant" on your business strategy.
I know I'm being very harsh on words, but its my hard earned money, and I had to spend a lot more here in India. I did all my soft research on this, and I can already see how much apathy Intel has about it.
Thank you for contacting Intel Communities.
This case is being reviewed and will be updated when new information is available.
I have also sent you a private message.
I have reduced the Voltage to 1.220V at 4.6 GHz, ran stress test for couple of mins, 75C -- AIO @ Full fan/pump speed.
Played GTA V, Temp max at 70C.
I'm seeing the actual voltage being around 1.27x but that's maybe the motherboard.
As you see, for 1.272V-1.278V, it is still high for 70C. Which means, I can go max up to say 1.35V at around 4.8 maybe with 80-85C with AIO.
Again, my question remains the same, so much expensive CPU, how come de-lidding the CPU with some metal thermal paste reduces the temps 20C on average ?
In that way, I could've gone up to 5.2 GHz with maybe around 75C !!!!! That is insane !!
I read all these earlier but I imagined Intel to be gentlemen enough to fix that on next batches, but it was never fixed. What is the fix for these temps ??
1.35 is definitely to much for 4.6Ghz
My Asus MOBO tries to set 1.365 just for the XMP profile (4,6Ghz)
But CPU work fine with 1.275 right now.
My previous CPU (before the replacement) worked with 1.25v.
Intel cannot replace their termal solution in your CPU.
So you may delid, but before, understand all the dangers it holds. Thinking about deliding myself, but deliding is a reliable way to throw away 400euro.
I would try to ask for a replacement.
As links said delidding is a risk practice and should not be done unless you fully understand the risks.
Also you were correct in the fact your voltages were too high. I am curious what setting you used to change the voltage. Did you use Manual, Adaptive, or Offset?
If you used adaptive remember that the limit you set is used only for when the chip is boosting. So if you have a chip that requires only 1.22 at 4.6, but have a VID requesting 1.25 at the non-boosted 4.2ghz it will mean that your cpu will actually use more power when not boosting at 4.2ghz, but luckily these chips try to boost and chance they can so this should be a minor issue.
If you want to reduce the range of voltages across the various "steps" you can do that with offset. Offset can be tricky as it does apply it to the whole range so while you might have enough power at max usages some of the lower states might be power starved leading to issues. That being said if you did decide to use offset you would need to find out what actual voltage your chip requires at 4.6ghz. Then see what it is requesting via the VID or by running it on auto and seeing the vcore. Once you have these two values you can determine how much voltage to offset it by... i would generally give myself a .05 cushion here for the lower clocks.
So if you can offset it by .35 I would only go .30 so that your lower frequency steps have less rick of being power starved.
Then again adaptive is still probably the best method for most people regardless of the few downsides to it. I am not trying to persuade you into using offset, but instead just giving you more information so you can make an educated decision.
Lastly, if your temps are under 80c I wouldn't worry too much about them. That still gives you a good bit of thermal headroom and these chips run hot regardless. My stock chip was pulling low 70's on a custom loop when I got it (25C ambient).