Is the thermal compound on the die of a 6700k this bad?

[DarthSmartt]

I have a custom watercooling loop with a 360 and a 240mm rad. I am cooling my 6700k and my gtx 1080. The 1080 is at stock clock speeds and the 6700k is at 4.5 at 1.45v. With my pump running at a little over 50% and my fans at 100% my gpu stays at 52c but my cpu is jumping from 80-90c. This is while stress testing both the cpu and gpu at the same time. I know my thermal compound application is good and the waterblock is mounted solid. The only thing I can thing is that the thermal compound between the die and the heat-spreader is absolute shit. I knew it was bad but is it really this bad? I really don't want to delid.

[Aereldor]

It's that bad. Delids yield improvements of up to 25 degrees celsius on 6th and 7th gen chips. I would highly recommend delidding, the risks are low with a delidding tool (3d printable, cheap on ebay), and will increase both the performance and the overall lifespan of the chip.

 

 

[JoostinOnline]

It's not actually that bad, but there are other issues. I wrote an explanation a few years back, the link is in my signature.

[DarthSmartt]

3 minutes ago, Aereldor said:

It's that bad. Delids yield improvements of up to 25 degrees celsius on 6th and 7th gen chips. I would highly recommend delidding, the risks are low with a delidding tool (3d printable, cheap on ebay), and will increase both the performance and the overall lifespan of the chip.

 

 

Also just noticed that I have the flow direction wrong on my cpu block. That may be part of my issue but it is such a pain in the ass to fix that I am just going to leave it. If it ain't broke, don't fix it. My cpu is bottom of the bin in terms of overclocking performance anyway so I really can't care that much.

[DarthSmartt]

2 minutes ago, JoostinOnline said:

It's not actually that bad, but there are other issues. I wrote an explanation a few years back, the link is in my signature.

That is quite the study! It does make a lot of sense. That adhesive is thicccc.

[Aereldor]

Just now, DarthSmartt said:

Also just noticed that I have the flow direction wrong on my cpu block. That may be part of my issue but it is such a pain in the ass to fix that I am just going to leave it. If it ain't broke, don't fix it. My cpu is bottom of the bin in terms of overclocking performance anyway so I really can't care that much.

It's a GTX 1080, depending on the use case, you could stand to gain some performance from pushing the 7700k, which is realistically only possible after a delid. 

[Spotty]

16 minutes ago, DarthSmartt said:

the 6700k is at 4.5 at 1.45v.

1.45V is way too high for 4.5GHz. You should be able to hit 4.5ghz with <1.3V. 

I did some tests with my 6700k a while back here. 

[TrigrH]

8 minutes ago, Spotty said:

1.45V is way too high for 4.5GHz. You should be able to hit 4.5ghz with <1.3V. 

I did some tests with my 6700k a while back here. 

@DarthSmartt are you using LLC? without load line calibration set correctly you need to use a lot more vcore than you actually need.

[porina]

4 hours ago, Spotty said:

1.45V is way too high for 4.5GHz. You should be able to hit 4.5ghz with <1.3V. 

I did some tests with my 6700k a while back here. 

I'd agree in the sense that 1.45v is excessive for 24/7 use, but what clock at what voltage will also depend in part on stability criteria. For Skylake, based on overclocking two 6700k, one each of 6600k, 7350k, 6100, I typically reached 4.7 GHz at <=1.40v Cinebench R15 stable but I didn't long term test it. For the K CPUs, Prime95 was borderline long term stable with 1.25V at 4.2 GHz. By that I mean you might see detected error rates on the order of one a week to one a month, in the end I dropped to 4.0 to be sure (stock running for 6700k, OC for 6600k).