[i7-6820HK] Overclock Settings for 4.0GHz to 4.3GHz (Stable)

Hi everyone, I just got my 15r3 past week and testing the temps and overclocks, I am new completely to benchmarking, getting around 750 marks on intel XTU on stock clock @3.6 GHz, max temps 79°C is that normal or ok?

 

I can run 4.1, but I keep hitting the Current Throttling Limit.

 

I'm confused as well : while OCCT'ing temps barely go higher than 65°C (with a good 60° average) but ingame I'm almost 10° higher...

OCCT (short test) :


Back to idle :


I'm currently running at :
1-Core Ratio Limit Override: 38
2-Core Ratio Limit Override: 38
3-Core Ratio Limit Override: 38
4-Core Ratio Limit Override: 38

Core Extra Turbo Voltage: 0

Power Limit 1: 100000
Power Limit 2: 100000

I'm lucky : my CPU can whistand 4.1 with ease but I see NO difference while gaming. So I rolled back to 3.8, which seems a good compromise.
I went LM a couple of days ago (yay!) and so far so good : still dropped 10-15°C max temp while gaming (from low-mid 80°s to barely 70° now), not much changes while idling though. I'm still with 1.08. And best of all : core temps difference is slim to none.
GPU will get a steady 65-66°C while playing Witcher 3 on Ultra and native QHD resolution.

Room is at 27°C, I'm currently living in Libreville

 

guys do you notice a increase in core temperatures @ 4 GHz OC compared to Stock 2.72?

 

plz post your screenshots of the XTU settings for oc @ 4GHz

 

OCCT isn't a good load generator, CPU intensive games are better, but it's tough to make those runs repeatable.

I use prime95 small FFT tests on all 8 cores (default) for 5 mins to 15 minutes.

Use 100% fan's for that test, let them run for a bit before and after the test run to provide a cool baseline temp.

If you use auto fans, you will need to clear the hwinfo64 readings after 30-60 seconds to get rid of initial Red ink and Thermal throttle - especially in thin laptops as the CPU temp will rise much faster than the fan curve will respond on auto.

There is a button on the bottom of the hwinfo64 dialog that will clear all the readings, I do it after I hear the fans ramp up on auto - I do still recommend 100% fans instead from before the start of the test.

You can use hwinfo64's logging feature set to sample every 2 seconds to see the correlation between CPU speed / utilization and power to temperatures per core along the whole run of a benchmark or a game run to see temperature peaks and more importantly running temperature during the extended run.

Quite often the temperature peaks are few and brief, usually at the beginning of the run while the fans are ramping up. Otherwise temps are good throughout the game / benchmark run.

 

I'm having trouble keeping the my 6820HK in my AW17R4 at any speed over stock speeds. I've treid 4.1, 4.0, 3.9, and 3.8 on all 4 cores with PL1 and PL2 set to 100000. I keep getting TDP dips from 47 down to 26 in seemingly very regular intervals causing my games to freeze for a few seconds making everything unplayable. This only happens when gaming. Ive tried over 20 different games from BF1 to Farcry4 and they all experience this phenomena. I put turbo power limits to unlimited in XTU but nothing seems to work. I've tried overvolting, undervolting, pretty much everything under the sun. Currently running BIOS 1.0.12 but also have tried every single previous bios with same problem. I did a repaste with thermal grizzly conductonaut and am having GPU max at 72C and CPU max at 86C. What am I doing wrong?

 

Maybe <disabled fan performance mode> may affect EC firmware power settings? Put enabled. Dell is known to tamper with the power settings in firmware for their models. Next step. Start all over again. Flash and use the bios version others have had success with.
Edit: Remember the thermal throttling point, can be lower than you expect. Have seen 88C as a number. Maybe lower for different bios/Ec versions. Also check PCH temp when this happens.

 

@Papusan do you note that with the last update of 3Dmark the bug of different scores come back ??? with the 5.0.609.0 is a difference of 5-10 points but is there

 

I'm not sure what you mean. Latest 3DM suite v2.2.3509 from 15 Dec work as intended. The bug In previous GUI versions was finally fixed.

Edit: You meant new System info? I haven't tested but it's unlikely this piece software should make a difference.

 

test it with the last update, i have 10 to 5 difference. before it was 0 difference on benchmarks. also is really strange because my cpu score raise and gpu score down and in the another is the opposite, but the final score is 10 to 5 points different at the end

 

5-10 points is a very small difference. The total score will drop more with slightly lower graphics score vs. increased physics. I would think a optimized system info as Futuremark claims will influence on the processor (less wasted processor cycles). = Increased physics As expected. But I don't understand this with the graphics score. Tested with same Nvidia driver? And no other changes? You could do as my testing for Futuremark... Use a ups or kill-a-watt to measure if you can see a difference. If you see it, maybe post your results in Futuremark support. Give me feedback if you try it.

 

85-90 °C

I tried everything in XTU and BIOS. Apparently is a motherboard defect.
P.S. I tried bios from 1.0.2 to 1.0.12 it doesn't help too

 

Do all AW17R4 motherboards suffer from this? If not I'm going to give AW a call and have them replace the mobo along with a new screen without this horrendous banding water mark BS.

 

See New vbios and PCH temps through the roof! And Dell do as they started with first Aw BGA models early 2015. No PCH COOLING aka No need for it although all complaints. See also posts from sub-threads regarding the older AW BGA models with exactly same problem. @judal57 and @VICKYGAMEBOY know about the PCH TEMP MESS
Thin machines will always have problem with hotter internal components (passive cooled hardware). See also SSD's gets boiled in ALIENWARE as well.

 

@Papusan dammmm 82°C on an SSD !!!!! that is new to me !!!!

 

You can fix it changing from Slot 1 to slot 2. Someone said, I can't remember their name.

 

@Papusan @Mr. Fox can you teel us if this heatsink works
http://www.ebay.com/itm/ALPHACOOL-H...732658?hash=item54424f1832:g:o50AAOSwuxFYwwHi
i see your youtube channel and i saw you are using this heatsink

 

Great fix Tragic. What if you want all slot filled? I expect you have 2x PCIe 1x crippled M.2 slot and one 2.5"

No guarantee that it will fit in your machine. But you can modify the heatsink. Reattach the upper metal cover with electric tape around the ssd. Aka remove the clips. Several have done it.



Edit. Another M.2 ssd heatspreader

 

Yes those passive cooled method sucks, cheapo alternative, specially for those who lives in humid places.. trust me even ssd gets throttling, when there is heavy load, mine went upto 92c on ssd.. it started throttling, FYI its PM951 256GB nvme.. so yeah thermal pads where ever you can, try to keep overall internal components cool.. no other options.. for 70% of those people who never faced issue, you guys must be living in cold weather countries or atleast not so humid..

 

A lot of pc users do not even know how to watch hardware temperatures. You think it might not... But it's true. All people care about is thinner, thinner, thinner, but forget that passive cooled hardware need to breathe.

 

@VICKYGAMEBOY: I used wooden plank to lift up the base to reduce those temps. I know because we live in more or less in same humid places.

 

I found that heat sink to work very poorly Brother @judal57 and do not recommend it for NVMe. It works OK for normal M.2 SSD, but my NVMe was still hitting 90°C with that piece of crap on it. How much space do you have available? I am using what is shown in the photo below on the NVMe in my Tornado F5 and it works amazing. But, it will not fit in tight places.

I used half of the Alphacool thing on the underside of BPX NVMe (yes, I moved the 960 Pro to the DM3) and the finned sink on the hot top side, then used rubber o-rings to sandwich the stuff together. I used the other half of the Alphacool thing on the hot top side on my X400 SSD with nothing on the bottom.

On the P870DM3 I have a fat thermal pad squished between the M.2 and the motherboard and that is doing an excellent job (no heat sink). If you have limited space, you may want to try that.

As you can see in the photo, the finish on the Aphacool thing is not durable. It is easily scratched by the clips. Simply snapping them on and off one time scratches the crap out of the finish.


So, with the setup shown above I went from 90°C to never hotter than this...

Spoiler: New temps with finned cooler...

 

It has no fins and minimal surface area of dissipation.

On a desktop it will work, on a laptop it will most certainly not.



I believe there isn't much clearance on SSD for the 17R2/R3, but using thermal pads underneath the SSD works, also try to thermal pad to contact the access bay door.

 

@iunlock i saw your temperatures using thermal grizzlt kryonaut and seems to be better on gpu, so i want to ask you if you can make a hybrid repaste, conductonaut on CPU and kryonaut on gpu and see what happens

 

I think his laptop can't use NVMe drives. Too old. M.2 ssd heatspreader maybe is a better choice and fit older AWBOOK's. And this HS isn't sandwich version. Regardless... Change the pads to a better one.

 

Yes, I agree. With limited space that should be the best approach. Also could apply a wide strip of aluminum tape to the inside of the access door so the pad is contacting metal instead of plastic and that may also help.

 

See previous post regarding Heat spreder for AW17R2

Yes try it. The heat problem is the processor. Gpu can use most of the good paste.

 

@Mr. Fox thanks, i was just curious, but i am nor facing heating issue on my ssd´s. they are SATA 3 and i have fujipoly thermal pads above them and they are really good.

i tried many thermal pads, unless 17W/mk fujipoly and i can say the more durable at the moment are the fujipoly 6W/mk (Sarcon tmkk402) 2mm thickness. i like that dont dry quickly

 

Very good, sir. Glad the temps are fine now.

Edit: If you use one heat plate from the Alphacool kit on top of each M.2 SSD with a thinner pad between the heat plate and access door that might work better than 2MM pad. But, if your temps are already good enough, then there is most likely no need to waste money on the Alphacool kit. As Brother @Papusan said, you can use electrical tape to attach the cooler plates instead of the clips.

 

@Mr. Fox later i would like to make a mod and use the 2 slots of m.2 drives that are covered, making horizontal holes with a motor tool and this flex

 

@Mr. Fox also, have you ever seen this type of heatsink? i always wanted to see those on laptops
https://www.aavid.com/product-group/pulse-jets
http://www.thermacore.com/news/active-heat-sink-technology-in-high-power-electronic-devices.aspx

 

No, I have not.

 

maybe @iunlock that have money can afford one hahahah i can barely afford Chinese thermal pads hahahahah

 

Yes, use Kyronaut on the GPU and Conductonaut on the CPU.

The GPU with a bigger die surface area has a larger contact point, allowing a good paste to be sufficient enough to keep it nice and cool.

The GPU is also not as sensitive as the CPU with 4 cores under the silicon to where it matters greatly to have great coverage. (ie... Larger surface area to dissapate the heat and doesn't get as hot.)

The CPU has a lot going on so having a complete coverage with the best thermal conductivity medium is important.





::iunlock::

 

Dell Russia refused to replace the motherboard They said that temperatures above 80 degrees are normal for a laptop.
I bought an expensive heater instead of a laptop.
P.S. Only 2.72GHz with LM it's terrible

 

What about a new heat sink?

 

I replaced all the thermal pads...
Maybe Grizzly Conductonaut сan damage the CPU crystal?

 

You mean the surface of the die? I think No it will not damage unless it stays on die and doesn't flow/leak towards the board. To be on safe side, tape all 4 sides with a tape.

 

If he can't get ok temp with liquid metal.. Bad contact between Die and HS. The results in the end can be leakage and Liquid metal will dry up. The next level will be even worse temp or fryed hardware.

 

You could try Kyronaut or Gelid Extreme as alternatives.

 

I tried. Imprint perfect. The temperature is 95+ during games.

The surface of the cooling system is perfectly flat (I polished it to a mirror shine).

 

Time for a new replacement Tripod

 

I always get an image of a three legged dog when I hear the word tri pod. Poor dog, but also poor design.

::iunlock::

 

Try using sand paper to smooth out the heat sink surface and sand it only in one direction.

 

YEAH, Reminds me very much about 10 Worst Car Features and Flaws Ever!!

 

That is exactly what my system does!!! But it happens regardless of temperature.

 

I have a few questions I'm hoping you tech savvy people can help me with. But first, I wanna pat you on the back a little bit, iunlock. I just got done following your re-paste guide before I came to this thread and realized you wrote this one too. You're the Dell representative the people deserve and Dell needs. They should pay you for this ****.

So as I said, I followed the re-paste guide. Everything was fine at first, but now my temps are hitting close too 100 again. I have come to terms that it is my fault, though. I half assed it. I didn't pay enough attention when I went over the guide the first time to realize that it was suggested to replace the thermal pads as well. All I did was wipe off the toothpaste and replace it with the Thermal Grizzly paste, non conductive. Apparently that is no match for Dells shoddy design. The reason I'm asking my question in this forum is because, like someone before me, my 6820hk will not go above 3.6, or more specifically 3594 MHz. I followed the instructions for the BIOS OC of 4.1 exactly. XTU even shows that the max turbo cpu speed is 4.1, but 3.6 it stays. I just got off the phone with Dell, the rep told me that the engineering department told her that the cpu will only reach 4.1 with one core. If more than one core is engaged, it will not pass 3.6...

Side note: I spent $85 on everything needed to re-re-paste using the liquid metal Thermal Grizzly, along with the pads this time. I called Dell because I thought my temps did permanent damage to my system and I wasn't about to re-paste just to have them replace the whole board after. She heard my case, put me on hold for a bit, then came back and told me that they're sending me a new system (third so far) because the new models are no longer having the overheating issue. It'll probably take 3 weeks like last time, but I'll try to remember to come back to this thread or the re-paste thread to share the results. I'm definitely not gonna cancel my re-paste order from Amazon though. Even if the temps are better, I'll probably still re-paste for maximum cooling.

TLDR; Why won't my damn cpu go past 3.6 GHz?

 

Have you increased PL1 and 2 ? Follow the guide a second time.

 

I did, yes. I set them both to 100000. I tried resetting defaults in my BIOS to defaults a few times but I still can't get past 3.6. I'm running 1.0.12, I'm thinking about rolling back.