[Undervolt Results] 6700HQ ? 6820HK (What's your MPG?)

Greetings,

This thread is to compare our chips to see the different fuel economy results we can obtain with our CPU's. It should be interesting to see the results, because the results will really show how not all silicon's are created equal.

Let the games begin...

Special thanks to: @Mobius 1, @DeeX, @VICKYGAMEBOY, @judal57 and others for all their knowledge and support.

-

Please copy / paste and use this format below:

Vehicle: AW 17R3
CPU: 6820HK
CPU OC: 36x
Core Voltage Offset: -200.2mv
Cache Voltage Offset: -65.4mv
Application: ThrottleStop 8.10 beta 2

Optional: (If applicable...)
Thermal Paste: Grizzly Conductonaut
Thermal Pads: Fujiploy 17.0mk/W

*Please include screenshots*


-

 

2

 

Vehicle: AW 17R3 BEST BUY SPECIAL RETARDED EDITION
CPU: 6700hq
CPU OC: 31x / 26x
Core Voltage Offset: -115mv / -155mv
Cache Voltage Offset: -65mv / -65mv

Optional: (If applicable...)
Thermal Paste: Grizzly Kryonaut
Thermal Pads: Dell toilet paper


I recommend OCCT for 30m for preliminary stress, and 4+ hours for stability stress

 

@iunlock i think we need more ... like a stability test (i dont know if AIDA64 extreme edition is a good choise) ... 15 minutes at least; and you dont unvervolt the graphic voltage offset ? i do ...

 

iGPU downvolt bit risky, I can recommend only -10mv to start. Mine only goes down to -20 or -25 before bsod entire computer

 

Vehicle: AW 15R2
CPU: 6700HQ
CPU OC: 31X (4core turbo)
Core Voltage Offset: -175mv
Cache Voltage Offset: --175mv
Application: AIDA64 extreme (30mins run)

Optional: (If applicable...)
Thermal Paste: ICD 7
Thermal Pads: Nope

 

OCCT > AIDA

 

mine runs -100mv GPU offset, but yeah don't bother touching, create's issue on gaming.. there is no point undervolting that..

 

where is your result ?? i can't see.. and OCCT results coming soon

 

Good idea...let's include that...after all if it's not stable then it won't mean anything yea....

 

don't have the AW anymore

 

OCCT 10mins test, -175mv on core,cache.. max temp 66c, Max fan speed 2400rpm, almost dead silent.. i could have tested 30mins with fan manual speeds which might have given me better temps.. but naah.. not botherd for now..

 

nice

 

Vehicle: MSI GS43VR
CPU: 6700HQ
CPU OC: None
Core Voltage Offset: -195mv
Cache Voltage Offset: -195mv
Application: Prime95 Small FFTs for max FPU stress (30mins run)

 

Those are some great UV numbers on the 6700HQ. Good stuff...

 

i undervolt my friend laptop, a msi with the i7 6700hq -150mv looking you .. thanks .. if you found that -175 is stable, i could try

 

looks like all the i7 6700hq have too much voltage at stock... the undervolt is insane on this models ...

 

the thing is 6700hq stock clock speeds are different, they left the voltage settings used on haswell 4710hq, the reason why its pretty high on most models, Yes for me -175 stable for gaming and rendering videos on adobe premiere.. check the link I sent u on skype.. for the PCH heatsink.. its perfect sized, and also doesn't touch the bottom base.. it comes with thermal glue.. and yes please apply artic thermal pads on ur main heatsink, remove those crappy dell ones..

 

Actually, -195mv wasn't too stable for my 6700HQ after all. I got the whea_uncorrectable_error BSOD message after running the Rise of TR built-in benchmark. The benchmark ran fine but the result screen at the end (where it shows your FPS result for the run, etc.) generates 100% GPU load for some reason (my CPU+GPU fans all ramped up due to heat). I let it sit there for 2 minutes and Windows 10 BSOD'ed. I did another run with -185.5mv (on both core and cache voltage) and it has been stable even if I let it sit at the result screen for 15 minutes.

Has anyone got this BSOD error before? Is it related to undervolting your CPU too much? Since my CPU was at 79C and GPU was at 83C before the BSOD, it should not be related to CPU or GPU overheating, correct?

Now, I highly recommend running Rise of TR's benchmark and letting it sit at the result screen for stability testing of your CPU or GPU

 

too much undervolt





use occt for at least 6 hours

 

inthat case I had -225 mv for just 1min lol, those WHEA error related to insufficient voltage on ur hardware, Yes -100 to -175 is the best scenario here, for me I ran on Core + Cache + graphics. and ur temps are not so good with undervolt, try to repaste..

 

Thanks for the advice @VICKYGAMEBOY. I will try to re-paste the CPU and GPU with IC Diamond when I have time. Btw, what's your stable undervolting settings (core+cache+GPU) when running on battery? Does it give you more undervolting mileage when you are on battery?

 

Here is my testing for the AW17 battery test, (4k screen). -> https://docs.google.com/document/d/1o-bp11u5RLIe0R9U8jLj7wM1NSDV1nIL2dLObh399po/edit

I can get around 4.5 hours with custom power profile and a lot of undervolt. Closer to 6 hours if you control the brightness and use a lighter video player such as VLC or disabling madVR.

FHD screen will have more battery life compared to 4K, but not sure by how large of a margin.

I recommend Kyronaut instead of ICD, much easier to apply and remove. Or if you're feeling brave, go for Conductonaut.

 

That is a common error with either over volting and/or under volting...ie...not enough power or too little power.

Yes, use OCCT as @Mobius 1 has suggested. That's the real CPU bench test. It really does make every other CPU bench pretty obsolete IMO.

OCCT should be the new standard. It's the real deal.

 

I'd recommend GELid Extreme over ICD. From my experience GELid is a much better paste and has a stronger integrity. ICD provides okay results in the beginning, but it is integrally weaker than GELid. ICD does scratch the die and will leave little spider web like scratches, which won't hurt, but just the thought of it can be irritating depending on the level of your OCD.

It's up to you, but I'd opt for GELid any day over ICD without even thinking twice.

 

And, don't forget that the last undervolt tuning will be for a stable "idle".

Load testing is one end of the tuning. Once stable there you may still get a crash while in programs normal usage, or simply sitting at idle for a period of time.

If you aren't tuning anything else at the same time, it's usually the undervolt is dropping idle voltage too low. It will tick down just enough to be unstable.

Bump up undervolt say from -195mV to -175mV for a sanity check - it should be stable at idle with a -20mV correction.

Then after a few days, go to -185mV, and see if it is stable there for a week.

As you get to the limit of such a large undervolt, it's usually going to take days of running at a new test setting to find the max undervolt.

 

I was about to suggest him the same, stay away from ICD.. and regarding undervolt battery life, I'm yet to test it, I get around 5-6hours of normal browsing + watching movies.. 3-4 hours if i use photoshop and adobe premiere.. there wont be much of a difference on -150mv on all core/cache/graphics.. except the temps.. i got around 3-5C max temps reduced..

 

Vehicle: AW 17R4
CPU: 6700HQ
CPU OC: 31x (4core turbo)
Core Voltage Offset: -150.4mv
Cache Voltage Offset: -150.4mv
Application: ThrottleStop 8.48

Optional: (If applicable...)
Thermal Paste: Grizzly Kyronaut
Thermal Pads: Dell Standard ones

20min small FFTs test with prime95, i know OCCT is recommended, but for the purposes of my needs (gaming) it works fine, only game that pushed close to those temps was guild wars 2 (63c peak, 55c most of the time). i did also run this test for 8 hours and temps never went above 66c

 

I had 3 mobos with 6820HK:

First:

• Core -180mV
• Cache -180mV

Second:

• Core -140mV
• Cache -140mV

Current:

• Core -150mV
• Cache -150mV

Using XTU & TS and stressed with OCCT.