Is the P775DM3 cooling just bad or what?

After asking a bunch of questions here on this forum I've gotten most things set up so I'm mostly satisfied with this laptop. However there is one thing, the CPU cooling does seem pretty poor to say the least.

Even after undervolting the CPU (6700k) by -170mV (which I've been told is too much) and putting something under the laptop to increase airflow it can still reach 80c+ (no overclock) if all cores are used 100% for a while. I mean it's probably fine, but it doesn't leave much room for overclocking.

Just setting all cores to 4,2Ghz makes the CPU reach 95c+ when running prime95. So 80c at stock 15c more with all cores set to 4,2Ghz. 10-15 degrees for 200Mhz, seems kinda nuts. No overclocking for me

It does do a decent job at cooling the GPU at least, it seems to max out at 75c even overclocked by 200Mhz (GTX 1070). Does a 6700k require that much better cooling than a 1070? Or is the cooling system for the CPU somehow not working properly for me.

 

If you want to overclock you really need to look at delidding the CPU and applying decent heat paste or at least using good heat paste on the heatsinks. I just did my own machine using Thermal Grizzly Conductonaut and the GTX 1060 GPU has dropped by 8-10C and by delidding the CPU and using the same liquid metal compound my CPU temps have dropped 18-20C, so am now running the CPU at 4.6GHz on all cores with a max temp of 81C using the standard 'Performance' fan profile, can shave a few more degrees off if I set the fans to full.

I also don't use Prime95 or the standard Clevo Control Center as they can cause instabilities even on stock settings. Try using an older Clevo Control center such as the one ending in .50 which you can download from the Clevo website under the p771 chassis section. This removes the XTU software which is rubbish and then use Throttlestop to set the overclocking instead an

My CPU won't go any higher than 4.6GHz without crashing

 

Obvious question here is obvious: What thermal compound are you using (and is the application good)?

Identical chips will not transfer heat identically; from one 6700K to another, there's almost always a small - but noticeable - difference in temps. You may have gotten a hot potato, in which case Adbear's recommendation about delidding will impact the temps greatly (although it it's an involved process that will likely risk your warranty).

 

I have no idea, whatever it came with. Got it from the swedish clevo retailer ( http://pc-konsulten.se/). No idea what they use.

Delidding is out of the question though, I'm not prepared to take any such risks. I also read a horror story where someone screwed it up badly. I remember him writing that "the glue was like concrete", of course that's not the typical case. He somehow managed to break lose some contacts of the CPU. Or so he claimed.

Nah I'll just wait till the 7700K is out and maybe upgrade to that if possible. Not prepared to destroy this CPU. Third time the charm perhaps? The CPU in my last purchase, my desktop was a total turd as well an i5 4670k that couldnt oc past 4,2Ghz.

 

Unless they advertise using their own compound, they may have used what Clevo provides - which works, but there are TIMs that transfer heat much better and can yield a difference of 2 - 4+ degrees Celsius under load (idle temps are comparable).

 

How did you do the undervolt? if you did it through the Clevo Control Center then that only applies it to the CPU and not the Cache which makes a world of a difference. You need to use ThrottleStop to apply the undervolt to both the CPU and Cache and set both to -150mV for stock speeds or -80mV for a 4.4 GHz on all cores.

First system I had was a barebone Clevo P870DM-G with a 6700K that couldn't even run stable at 4.2 GHz x4 (on all cores that is). Next system which was a Eurocom Sky X9 also had a 6700K, this time a bit better, it was able to do 4.3 GHz stable.

I have seen many users here complain about high temps even at stock speeds. The 6700K is a hit or miss, 50-50 chance to get a good one. Even if you don't want to overclock, spending a bit more to get a Silicon Lottery one ensures your CPU will run cooler overclocked or not. Giving you better thermals and longevity since the CPU won't be running hot constantly.
This Silicon Lottery CPU I got (rated to run @4.8GHz in a desktop) can run @ 4.7 GHz not very stable, 4.6 GHz stable but with higher temps than I would like in a laptop for prolonged use (94C) or running super stable @ 4.5 GHz all day long with max temps of 85C no matter what I throw at it.

Also, you need to figure out what thermal paste your reseller used. You see, from my experience, contrary to others, Liquid Ultra was too thin to make good contact with between the CPU and the heatsink so when I repasted with IC Diamond using the X Cross method which provides the best coverage on the CPU and least air bubbles, my temps went down by -10C

see:

Right, so after bloodhawk's recommendation to change CLU for IC Diamond since CLU is not making great contact with the heatsink on the P870DM due to how the heatsink is shaped or fitted

Here is what I did last night.

I removed the old heatsink and put it aside, I then cleaned my CPU using the liquid cleaning kit provided by Coollaboratory. It was super easy to remove with this cleaning kit vs. conventional alcohol.

Next, I used my brand new extra heatsink that I ordered with my laptop as I felt the screws on the old heatsink were kind loose / flimsy.....

The results are simply outstanding.......

After cleaning the IHS...



I used the X Cross Pattern for applying IC Diamond by the way because according to this article, the X Cross Pattern gives the best coverage and the least air bubbles:



This is on a Silicon Lottery factory delidded 6700K CPU @ 4.5 GHz with an adaptive voltage / -65mV offset:

Before with Liquid Ultra:



After with IC Diamond / new heatsink:

 

Most likely they used whatever clevo supplies, since they offer an option to get a different thermal paste. That option wasn't available when I got the laptop. I do have an unused tube somwhere of the stuff Noctua sells, (NT-H1) usually gets decent scores in thermal paste tests. Perhaps that would be an improvement.

 

Clevo uses IC Diamond by default so my guess here is they didn't go a very good job. I hope you can repaste again WITH IC Diamond but using the X Cross method:



This article proves to you that it's the best method: https://www.pugetsystems.com/labs/articles/Thermal-Paste-Application-Techniques-170/

 

Don't rely on test of thermal paste done on test bench or desktop hardware!! Laptop coolers is provided with weaker cooling, heatsink have not the same pressure on DIE/IHS and as well not perfect fit. Noctua NT-H1 isn't more than a medium quality thermal paste on laptop hardware

 

Prime 95 will push your CPU far harder than most tasks. An usual load in a system designed to raise the power limits by defaults will always make things hot. However you are running around 4ghz where as a notebook chip is pushed below 3ghz.

 

Anything else that is good for laptops other than IC Diamond, it seems they don't sell IC Diamond where I live (Finland). Checked several online stores not a single one had it, unless I'm blind.

Not that I'm in any hurry to fix this, I might make it worse. Atm the GPU is kept fairly well cooled. Looking at some documentation it seems you have to redo both GPU and CPU since the whole heatsink assembly seems to be joined together. I'm a newbie at this whole laptop thing.

 

Yeah, Unified heatsink. Look after Gelid Extreme or Grizzly Kryonaut
As well look into http://www.coolerkit.dk/shop/frontpage.html Not far from Finland

 

I'd agree with those recommendations, if IC Diamond isn't available to you. Those seem to typically get positive feedback around here for good temp drops, and are fairly easy to apply.

 

This is what the temperatures look like on this CPU (6700k) at a few clock speeds/voltages (and the GPU as well although that's kinda irrelevant). I just rendered some random 3d model in Catia which put all cores at 100% usage for roughly a minute and checked the temps afterwards. With no voltage offset it runs at just shy of 100c at stock clock speed (and would certainly have reached 100c had it lasted longer than a minute)... yikes.

(Max temperatures, auto fan speed)