Clevo Overclocker's Lounge

LOL, thanks guys, obviously I have something wrong with my eyes or I was thinking about a different forum because I swear I looked there and didn't see it.

 

@Phoenix odes the uncore voltage has to be lower at 4.4 than 4.2? im confused

 

yes, the higher the overclock, the more you need to tweak the uncore voltage. for 42x4 the default is fine so leave it at 0. For 44x4, set the uncore to -10mV, if temps were high, try -20mV

 

@Phoenix

Did you get the chance to try the new settings? Would really love to know how well your chip clocks.

 

I will in a couple of hours. I will try Mr. Fox's new settings and yours

 

Keep us posted!!

 

What has everyone been getting for high gpu clocks I have reached +175 core and +300 memory but I can't push the core anymore without black screens.

 

did you increase the voltage on the GPU? Does your GPU have the Prema VBIOS or stock VBIOS?

 

d00d your settings are awesome! this is with the fans set to auto not full blast!!



wPRIME score improved from 4.7 seconds @ 4.4 GHz with my previous settings to 4.55 seconds @ 4.5 GHz with your settings.

What's crazy is my 4.4 GHz overclock was with the cache Ratio @ 4.4 GHz now even with the cache ratio @ 4.2 GHz according to your settings, we are still seeing a huge improvement!



Thanks a lot man, I cannot thank you enough for this.

What would you recommend changing in regards to voltage for increasing the cache ratio to 4.5 GHz or is that not recommended for stability?

 

@bloodhawk

Temps while running a backup from my SSD/HDD to my external USB Drives using multiple backup profiles (locations that is) at once

 

Can you share those settings

 

What have been changed in the settings?
Anyway good to see your improvement.

 

It is 2133 cl15 ram. Samsung ram has been known for good OCability.
Corsair uses samsung ram. And I think kingston uses hynix.(at least.they used to..)

I really dont see any difference between corsair and stock samsung exept corsair allows spd and xmp write. but since our friend prema has opened all the memory options we just dont need spd and xmp in practical use.

 

Dang! Nice drop in temps!! I still need to fiddle around a bit and try to tighten these more, if possible.
Personally im not sure if it is really worth it to bring the cache to 4.5Ghz because it will increase the thermals quite a bit with almost no real world performance game. You might be able to get that milli-second ahead in benches but other than that i haven't observed any performance gains. Maybe try bringing down the cache under volt to -80 to -50 mV. It might stabilize , but IMO not worth the headache.
I would say try to get a 4.6 to 4.7ghz for daily use, that would be way more useful.
But next week im going for the 4.8-4.9 Ghz Bench-marking. Gonna put CLU on the IHS as well. We might see a further 4-5C drop in my temps once i do that.
btw. https://i.gyazo.com/8dcf732c9558f02289f905c31fccf675.png

@Juang1985 @godfafa_kr
He is referring to these : http://forum.notebookreview.com/threads/clevo-overclockers-lounge.788975/page-104#post-10231478

Pretty good for 24/7 clocks. Not that competitive for benching. Since the power limits are pretty tight.

 

Nice, my default SK Hynix modules can do 2400Mhz @ CL14 CR1. Now looking at your results, im thinking of going for 2666Mhz CL15 or 2800 CL16. I doubt these will be as good as your Samsung modules though.

 

Since you are not using the IGP then minimising the overall latency is your best bet for performance.

 

Yes I have both bios's but I do not have the option to add any voltage to the gpu in afterburner or the stock program

 

Use Nvidia Inspector.

 

Of course we need to be active here on NBR first, too, in order to write PMs and do enhanced customization stuff. That's the norm in most places in order to fight spam...

 

Thank now I have the option to up the voltage what is the voltage Max? What should I stay under?

 

I see new BloodHawk Magic, what about the Fox Magic.

 

It more depends on temperatures, remember always use small steps for voltage.

 

Didn't work on my CPU. The clock speed was constantly jumping between 4.7 and 2.3 GHz

 

Different current limit, Turbo power max, so many to list

 

Soon

 

anyone who's interested in minimalist passive cooling, this decreased my max temps by 6C:

http://www.amazon.com/Steklo-Aluminium-Adjustable-Portable-Universal/dp/B00VIZ60QY

I have a 15.6 but theres a review (third one down right now) where some guy says it fits his 17.3" Alienware M17xR3

 

Help us to shape the future:

https://www.techinferno.com/index.p...ll-the-future-of-ac-adapter-what-do-you-want/

 

Yeah... Bigger the better said the girl

 

Personally I would prefer to have 2x450W, so I can travel lighter with only one unit and have a second one @home for when I need more power.

 

I think it fit better with a big one. I avoid to have a power connector on the table behind my laptop. Because it is better for me without any extra hardware on the coffee table in the living room where I sit normally and use the monster .

 

am I allowed to post the settings for 4.5 GHz that you gave me in the OP of this thread?

 

Oh yeah yeah, go right ahead.

 

thank you, I will post yours and my new 4.6 GHz settings as well

still in testing mode now to ensure stability

 

@bloodhawk

while the 4.6 GHz settings were fine in normal day to day usage, they failed on the stress test, after 3 mins the temps rise up to 97C. Lowering the voltage more than -45mV would cause instability so I think I have reached the limit on this CPU @ 4.5 GHz

Going back to 4.4 GHz on all cores + 4.4 GHz cache ratio as it gave me excellent temps for a 24/7 overclock and great performance

 

Have you tried the new 4.6GHz settings with Wprime 1024M, Cinebench 11.5 + Firestrike? And what voltage you put in the stress test ? Skylake processors can have a voltage difference about 0.100mV with stock clocks. I don't know what cache speed you tested with, but try to lower a step and try again.

 

no, first stress test I always run is AIDA64 for 5 mins, if that fails, I don't test anything else. I need my system to be stable during extreme conditions

voltage is adaptive / -40mV core / -150mV Ring with stock ring settings @ 41x

 

Yeah but what is your stock core voltage?

 

0 = default with a negative offset of -40mV for that 4.6 GHz overclock on all cores

 

I know that but what is the stock core voltage for your chips? As I said in previous posts, those chips have variations in stock voltage. Up to 0.100mV difference + some chips clock better than others. Test reviewers said 4.6/4.7 + 4.5GHz cache is a sweet spot for i7-6700K. You could try with stock voltage without negative offset.

 

I don't know, this is how HWiNFO looks right now, -80mV offset @ 4.4 GHz on all cores:

 

Well I think I got every ounce outta her might be able to get more with tweaking.

http://www.3dmark.com/fs/8086200

But at least I got this.

http://www.3dmark.com/sd/3983838

 

Cool!
I stabilized mine at 4.6Ghz with a -50mV (about 1.275V) and cache at 4.2Ghz -150mV .
Other than wPrime and memory benchmarks didn't really find any use of the cache being higher clocked. At least for my work loads.

 

Mine passes those at 4.6Ghz easily, just that im more concerned about my render engines working flawlessly. Gonna have good benchmarks later this week though.

 

Because benching has it's own following....

 

Also, i have seen bench stable settings crash at the least demanding tasks sometimes.
For example the other day my static voltage setup crashed playing Counter Strike, while it runs perfectly stable while doing heavy rendering for 24 hours non stop. The same setup passes Prime95 small FFT's and blend tests for 30-40 mins, after which i stop it.
Buuutttt..... Im going for that 4.8-4.9Ghz mark once i get some time this week

 

Is anyone's frequencies dropping under load? I will be stressing at 4.7 and the freq will drop to high 3's low 4's immediately. no thermal throttling no power or current throttling. I just don't understand it.

 

Switch to high performance power profile and edit it. Make sure the minimum power state and max power state are both 100% for the processor power management section.

 

I think you misunderstand...
Benching is for numbers or testing against other setups in your class. It's not about benching for stability. Or rather....Stress testing one's system which is totally different.
Same goes for gaming...
And in your example. CS does not run the cpu in the same fashion as your prime stability testing might do. If you get a bsod, those codes help you to make the system more stable....Once you figure out how to decode them.
All this stuff starts happen when you start getting into higher overclocks and higher temps. Room temps are a direct influence on your olver clock/stability. Today you are stable, tomorrow you get nothing but crash after crash with the same clocks.... When you start going above 4.2 ghz...nothing is really set in stone. And just remember... Just because you are stable at 4.5 ghz means nothing if someone with a 4.2 ghz overclock setup is far more optimal than yours. Better numbers, better results, better numbers crunching...etc...etc...etc. Something to think about...

 

Agreed on most of it. But CS uses Source 1 Engine, which mainly only uses the CPU. I have had more crashes playing CS than overclocking With CS its mainly either RAM or CPU based system if your system is OC'd. At least on the desktops that i have owned.
But yeah, as you said, everyone's chip is different and the performance will vary from person to person!

 

Because your not running enough voltage to compensate for CS. And in your case, you are more than likely trying to run the uncore far higher than it needs to be.