7700k bad chip?

[mrobscura]

specs
z170 gaming k6
7700k
h115i
r9 390
16g 2400 ddr4
evga 750w b2

Hi, I recently upgraded from a 6600k to a 7700k(yes, bios is updated) and this is actually the 2nd 7700k I've tried. The first needed to be undervolted just to keep temps in check stock, so I exchanged it. This second chip runs a bit cooler, but still hot so I upgraded my cooling to a h115i. Thing is max temps are still high. Idle is fine between 25-30c but under load while in the 60-70s most of the time, temps will randomly spike into the 80s, or even 90s if I try to oc and run occt.

And speaking of overclocking, I can run realbench and xtu for hours stable at 4.9ghz and around 1.30v with realbench maxing at 88c. I tried for 5ghz but couldn't get it stable going up to 1.36v and temps where already in the 90s running realbench. So, I figured I'd settle for 4.8, which is also stable for hours running realbench and xtu at 1.28v and a max temp of 84c which I can live with. Problem is these setting fail instantly when trying to run occt. Occt fails after 5 minutes of 4.8ghz at 1.33v in bios with temps hitting the 90s, including cputin. The best stable oc I can get running occt is 4.7ghz at 1.28v. Anything more fails, even with huge voltage bumps.

Is cputin supposed to match the hottest core? Also shouldn't a h115i keep this chip cooler than this? And lastly I read that every 7700k should be able to hit 4.8ghz around 1.3 volts, yet I'm unstable at anything past 4.7. Is there something wrong with my chip that temps are so high and there's so little overclocking potential? I know you have to play the silicon lottery, but this is ridiculous. Thanks.

 

[geofelt]

You may be overly optimistic.

As of 1/13/17
What percent can get an overclock at a somewhat sane 1.4v Vcore.

I7-7700K
4.9 74%
5.0 56%
5.1 26%
5.2 5%

One can delid the processor for lower temperatures if you are into that.
I would not.

Running OCCT or other testers that may use AVX instructions, you can get hot very quickly.
For whatever reason, avx heats the cpu dramatically.
Motherboards should have a bios setting for avx which will reduce the multiplier when those instructions are detected.

On liquid cooling, your radiator needs a good source of fresh intake air to do it's job.

The orientation of the radiator will cause a problem.
If you orient it to take in cool air from the outside, you will cool the cpu better, but the hot air then circulates inside the case heating up the graphics card and motherboard.
If you orient it to exhaust(which I think is better) , then your cpu cooling will be less effective because it uses pre heated case air.

 

[mrobscura]

I have my radiator receiving cool air from the outside and a 140mm exhaust on top and 120mm exhausting out the back. And with this same cooler people are running occt OCed with an average temp of under 80c. But ok, perhaps they have golden chips, I was still expecting more than 4.7ghz at 1.28v. That seems terrible, especially considering temps in occt reach close to 90c with a h115i. As for the avx offset, unfortunately I don't think that's an option on z170 boards. It would be nice. And yeah, delidding isn't happening.

 

[geofelt]

kaby lake can run hotter. The previous peak vcore for skylake was around 1.3v; with the more efficient kaby lake, 1.4v is nte new peak.
If you want to approach 5.0, be prepared for vcore in the 1.4 range.

I would think your liquid cooler would do it, but perhaps it is not running well.
Perhaps a liquid blockage or pump problem.

 

[mrobscura]

i dont think its the cooler considering the chip did the same thing with my cryorig h7... random spikes into the 80s or 90s. and the cooler seems to be working fine since idle temps are low and steady, which means the block is seated properly, and the water temp starts a little above ambient rising slowly as the cpu heats up, which means the pump is working just fine and there is proper flow.