Intel Skylake Core i7-6700K IPC & Overclocking Review @ [H]

thesmokingman said:
Nice, grats on the delid. Did it lower your volt requirements for those clocks?
Click to expand...

Nah, but it improved my temperatures significantly. Way more than I expected, I honestly thought it wouldn't do much. But I always see mixed opinions on it. It was actually much simpler than I thought it would be, I just stuck a razer in the side and put the razers back side on my table and pushed the cpu down arching the blade away from the PCB, literally got it off in 2mins doing all 4 sides with light pressure. Then just scraped glue off with my finger nail, removed it off the heat spreader with a towel and artic cleaner.
 
Man I should do this. I feel like mine has a bad paste job under the lid.
 
revenant said:
Man I should do this. I feel like mine has a bad paste job under the lid.
Click to expand...

Well if you end up doing it, just be mentally prepared to buy a new chip if something goes wrong. I really didn't care and just wanted to see if there was any difference regardless and I live within a few miles of a Microcenter. Personally I think it is easier than most people think, just don't cut anything except the glue, you would know if you were cutting anything else and would have enough time to prevent damage to the PCB unless you were applying too much force.
 
revenant said:
Man I should do this. I feel like mine has a bad paste job under the lid.
Click to expand...
The thermal paste Intel uses is one of the best in the industry. The issue is that the layer of glue they use is too thick, holding the IHS physically too far from the CPU die to efficiently transfer heat away. This was pointed out almost four years ago, not sure how people still think it is poor quality thermal paste. Delidded my i7-3770K, loaded temperatures drop by 20°C at 4.7GHz
 
My 6700K has 2 cores 10C hotter than the other 2 when under load. Edit:: (by 2 cores I meant 2 cores + 2 HT cores, the difference was 10C on 4 cores vs the other 4)
Looks like mine also has a lack of paste across the whole surface.
Possibly due to the above mentioned distance of the IHS from the die causing the paste to not spread far enough.

I've been thinking of delidding too, I can feel my arm twisting...
 
Last edited:
Nenu said:
My 6700K has 2 cores 10C hotter than the other 2 when under load.
Looks like mine also has a lack of paste across the whole surface.
Possibly due to the above mentioned distance of the IHS from the die causing the paste to not spread far enough.

I've been thinking of delidding too, I can feel my arm twisting...
Click to expand...

you only live once
 
Lmah2x said:
you only live once
Click to expand...
I used to ride power bikes, my life must be percentage based now because the on/off switch broke a long time ago.
I still get urges to push boundaries though....
Stop it !
 
jedimasterben said:
The thermal paste Intel uses is one of the best in the industry. The issue is that the layer of glue they use is too thick, holding the IHS physically too far from the CPU die to efficiently transfer heat away. This was pointed out almost four years ago, not sure how people still think it is poor quality thermal paste. Delidded my i7-3770K, loaded temperatures drop by 20°C at 4.7GHz
Click to expand...

Thanks for the warning. I delided my opterons so easily with a box cutter and fresh blade. But yeah, if something goes wrong I know the consequences. I get 4.6 with 1.36v which is pretty good but my chip seems to run warmish and none of the heat seems to make it out of the radiator really. It's strange. It is very stable but the warmish temps kind of bugs me. After multiple reseats I am guessing bad contact between the die and lid is the issue.
 
revenant said:
Thanks for the warning. I delided my opterons so easily with a box cutter and fresh blade. But yeah, if something goes wrong I know the consequences. I get 4.6 with 1.36v which is pretty good but my chip seems to run warmish and none of the heat seems to make it out of the radiator really. It's strange. It is very stable but the warmish temps kind of bugs me. After multiple reseats I am guessing bad contact between the die and lid is the issue.
Click to expand...
As long as it's not getting over 75-80ish then I wouldn't worry too much, but generally, cooler is almost always better :D
 
Lmah2x said:
you only live once
Click to expand...
It got the better of me.
Idiot that I am, I forgot to take before / after temps.
Its at least 10C lower, possibly 15C.

The temp difference between cores has changed.
Instead of 4 cores being 10C apart from the other 4, now its only 2 cores 6C lower than the rest.
Not done any overclock testing, only powered up 12 mins ago.

Interestingly on first power up the motherboard reported a new CPU had been fitted and insisted I enter bios setup.
 
Nenu said:
It got the better of me.
Idiot that I am, I forgot to take before / after temps.
Its at least 10C lower, possibly 15C.

The temp difference between cores has changed.
Instead of 4 cores being 10C apart from the other 4, now its only 2 cores 6C lower than the rest.
Not done any overclock testing, only powered up 12 mins ago.

Interestingly on first power up the motherboard reported a new CPU had been fitted and insisted I enter bios setup.
Click to expand...

Yea mine said new CPU as well. Not really sure why it does that, but everythings been working fine for me ;)
 
Lmah2x said:
Yea mine said new CPU as well. Not really sure why it does that, but everythings been working fine for me ;)
Click to expand...
Good stuff :)

Perhaps mobos have a profile of the CPU stored.
Either that or it detects a CPU has been removed or fitted, not so easy without power though.


After further testing, mine is at least 10C lower avg temps under load, hard to tell cos today is quite hot.
Idle seems to be similar to before, perhaps it will be a bit lower given todays room temps.
I cant overclock any higher nor reduce my voltage.
(4.6GHz @ 1.408V, "proper"voltage reported by Asus AI suite. CPU-Z says VID of 1.322V)
But I can finally Prime Blend @ 4.7GHz. None of the other Prime tests pass so 4.7GHz is still a no go.

10C + is a good drop, at full tilt it used to get toasty, well over 80C.
Now its just over 70C running Prime small FFTs, cool :D
Not the best chip ever but its damn quick anyway.
I gave it a good shot.
 
jedimasterben said:
The thermal paste Intel uses is one of the best in the industry. The issue is that the layer of glue they use is too thick, holding the IHS physically too far from the CPU die to efficiently transfer heat away. This was pointed out almost four years ago, not sure how people still think it is poor quality thermal paste. Delidded my i7-3770K, loaded temperatures drop by 20°C at 4.7GHz
Click to expand...

I was thinking more like uneven application or something. I never thought they used cheap paste. The distance between the die and ihs issue makes sense. I probably won't bother doing this as even with the avx2 prime i was maxing at 81ish - normal use never goes over the low 60s. I just though it would be nice to optimize temps and maybe go for a higher oc maybe.
 
I am building a new hard line water cooled gaming PC. I was planning to go with the 6700k based on the very positive reviews here at [H]ard OCP. However, this tidbit from PC Gamer/Maximum PC has me second guessing.

The best high-end gaming processor: Intel Core i7-5820K

“Is this truly still better than Skylake’s crème de la crème of processors, the i7-6700K? Yep, absolutely. In computational benchmarks those six cores run rings around its Skylake cousin.”

Thoughts on this? Will the 5820k be a better choice for an enthusiast gaming PC than the 6700k in the $300-$400 range?
 
They didnt state its better for games.
They stated its the best for computational benchmarks.
 
Nenu said:
They didnt state its better for games.
They stated its the best for computational benchmarks.
Click to expand...

That's true. But they are also labeling it the "best high-end gaming processor."

You don't have to infer much to suggest that the 5820K > 6700K for gaming.
 
If its worth reading, the proof will be in the article.
 
They stated the best gaming processor is the 6600K.
I found some modern games are better with the 6700K (FO4, Tomb Raider, GTA V ...) but there can be thread contention issues in older games with more than 4 CPU cores which reduce performance slightly.
You can disable HT on the 6700K/5820K or limit the number of cores the game uses if you have performance problems, not that I've seen any.
The 6700K has a slight cache size advantage which may help in some situs but I have no proof of what.

If you are building a high end machine for other CPU data crunching purposes, the best CPU for gaming is the 5820K.
It wont clock quite as high as the 6700K and will use more power/create more heat.
 
revenant said:
I was thinking more like uneven application or something. I never thought they used cheap paste. The distance between the die and ihs issue makes sense. I probably won't bother doing this as even with the avx2 prime i was maxing at 81ish - normal use never goes over the low 60s. I just though it would be nice to optimize temps and maybe go for a higher oc maybe.
Click to expand...
Well, the only way it will get an uneven application is if the IHS is not flat on the die, which is caused by a layer of adhesive that is too thick ;)

Hell, 81C with an AVX2 Prime load is great. I'd just leave that :D
 
Lmao, just updated the latest beta build of handbrake. It now has profiles for QSV in x264/x265. I'm testing by re-encoding a 8gb ts to x265 avg 2.5kbps balanced and its taking about 45 minutes and hardly using any cpu. LOL.
 
So I decided to try out CLP on the die today. It dropped temps an additional 6-8c at load. 20mins of IBT max temp was 62c @4.8ghz 1.35v. Still had no luck at 5ghz boot, tried 1.5v. 4.9ghz 1.4v seems to be the max stable OC with this chip. Batch L523B541. Overall improvement with Delid+CLP is 21-23c roughly, didn't keep track of ambient temp the entire time but I generally keep the room at 23-25c.

I guess my IHS just had some really bad epoxy/sealant job.
 
Nenu said:
...After further testing, mine is at least 10C lower avg temps under load, hard to tell cos today is quite hot.
Idle seems to be similar to before, perhaps it will be a bit lower given todays room temps.
I cant overclock any higher nor reduce my voltage.
(4.6GHz @ 1.408V, "proper"voltage reported by Asus AI suite. CPU-Z says VID of 1.322V)
But I can finally Prime Blend @ 4.7GHz. None of the other Prime tests pass so 4.7GHz is still a no go.

10C + is a good drop, at full tilt it used to get toasty, well over 80C.
Now its just over 70C running Prime small FFTs, cool :D
Not the best chip ever but its damn quick anyway.
I gave it a good shot.
Click to expand...

Update for those that delid.

My temps on over 1/2 the cores started rising until they were getting close to 90C, a good 15c+ higher!
I took it apart and found that the lid wasnt exactly central with its original position on the die.
When pulling the lever down on the CPU it moves the lid forward a few mm.
Holding it in place while locking the lever fixed it and actually dropped the max core temp about 3 or 4C and have been able to drop load vcore to 1.392V (down from 1.408V max).
It now is 2 to 3 degrees under 70C when running prime.

Moral of the story, make sure the lid doesnt slip into the wrong position when closing the CPU lever.
Keep the spacing even on all sides once it is fully locked in place.
 
I just upgraded from a 2500K to a 6700K today, but the 2500K will live on as my media server/HTPC. I mostly just upgraded for video encoding, with a bit more gaming performance as a nice little "bonus" on top. From my testing so far the 6700K @ 4.5ghz is more than TWICE as fast as the [email protected] for x264 encoding (35 mins to encode a video vs 71 mins)

I've also tested a couple of the games where I was CPU limited and it made a pretty big difference in those as well. ARMA 3 50 man servers my minimum framerates went from about 30FPS to about 50FPS (with much higher averages as well). I also tested at an "amp station" in Planetside 2 which has had performance problems the past few months, and my minimum FPS went from 40-45 up to 60-65. Most other games I am probably GPU limited but since Planetside 2 is my most played game I'm happy for the extra performance.

Now it's time to actually try to push the 6700K. I installed Windows at stock clocks, then bumped the voltage slightly and set it to 4.5ghz and it was stable first time and haven't even bothered with anything beyond that yet.
 
Nenu said:
Depends how long you intend on keeping it.
If you are sensible, you might reduce its life from 20 years to 15 years.
If you go mad you might reduce it to 2 years.
If you are balls out mad you can kill it in weeks or instantly.

I harmed a 2500K by accidentally highly overvolting for 1/2 an hour.
It was already running at max sensible volts and max overclock.
It still overclocks now but by a lot less and is running in my Dads PC.
Thats a 6 year old CPU, highly abused yet still functions perfectly.
Click to expand...

Yep even if you manage to harm a CPU you have to do some pretty extreme stuff to just kill it instantly. In most cases it will just degrade. I ran my 2500K @ about 1.38-1.4V for 5+ years and my max stable overclock went down by a whopping 100mhz in that time.

So far with my 6700K I am just exploring how low voltage I can go and still be stable at 4.6ghz. Down to 1.29V so far and still stable (with Realbench). Seems pretty good. :)
 
Nenu said:
Update for those that delid.

My temps on over 1/2 the cores started rising until they were getting close to 90C, a good 15c+ higher!
I took it apart and found that the lid wasnt exactly central with its original position on the die.
When pulling the lever down on the CPU it moves the lid forward a few mm.
Holding it in place while locking the lever fixed it and actually dropped the max core temp about 3 or 4C and have been able to drop load vcore to 1.392V (down from 1.408V max).
It now is 2 to 3 degrees under 70C when running prime.

Moral of the story, make sure the lid doesnt slip into the wrong position when closing the CPU lever.
Keep the spacing even on all sides once it is fully locked in place.
Click to expand...

My delid is still going strong 2 months later after applying the CLP. I have CLU as well on my other chip both are holding up well.
 
Lmah2x said:
My delid is still going strong 2 months later after applying the CLP. I have CLU as well on my other chip both are holding up well.
Click to expand...
This makes me wonder if my non de-lidded 6700K has a bad connection with the die. I reapplied Gelid Supreme to my H115i block and cpu because i thought the thermal past was maybe not distributed well. When i run Realbench i still reach 80C on only 1.30V running 4.5GHz.
This should be way lower on an AIO like the Corsair H115i. I even use a push pull (4x 140mm SP). The coolant reaches 40C after while. When i up the fanspeed it doesn't exceed 35C. That's with a 25C room temp.
 
Yeah I am convinced the heat is not making it to the IHS well on these at all. my AIO (Thermaltake Water 3.0) barely barely gets warm while I am stressing the cpu and it's at 70+c. My old Koolance used to dump and I mean DUMP heat out of it cooling my old system.

I am about to upgrade my daily rig to a z170/6600k with an h100i v2 and am curious to see if this one behaves the same way. At least I will be able to get some real numbers from the Corsair util.
 
You must log in or register to reply here.
Back
Top