Clevo Overclocker's Lounge

Just use wPrime and Cinebench, or 3DMark 11 physics test for CPU tuning. If it is not tuned right it will crash in those tests. But, the nice thing is that none of these tests are overly brutal or dangerous. They do require stable settings though. The popular exception is Fire Strike. It is fun, looks cool, and it is a decent test of GPU performance. However, it is a lot more forgiving than those other benchmarks. Passing Fire Strike is not necessarily a good way to verify CPU, memory or GPU overclock stability compared to those other options because it might pass in Fire Strike with flying colors only to crash really hard in those other benchmarks. At any rate, if the CPU or GPU gets too hot in any of those tests you know you need to (a) choose less of an overclock, or (b) take steps to keep it cooler. And, yes... 5 minutes is more than enough to know if your settings are stable or your temps are out of control. In fact, 2 or 3 minutes is usually more than enough time to find that out.

Another area where people sometimes get off track is having unrealistic expectations. Running a 3 minute benchmark at 4.8GHz is one thing. Rendering a huge video for an hour probably ain't gonna happen at 4.8GHz. Even if it does not overheat, chances of the video and/or audio being corrupted due to CPU errors and glitches at such a high overclock is very likely. Playing games at your max CPU and GPU overclock is also unrealistic and probably won't turn out that great. But, 4.3GHz is certainly within the realm of possibility if the temps are OK. Some games just don't play nice with any kind of overclocks. Even LCD refresh rate overclocking can crash in some games.

In order to be ready for every scenario, you have to know your machine well and find out what settings are good at 4.0GHz, 4.3GHz, 4.5GHz... so on and so forth. Once you figure those things out, ThrottleStop is your digital Swiss Army Knife that is ready to help make the rest of the job easy for you, with 4 on-the-fly CPU overclock profiles that can be applied instantly with nothing but a keystroke combo... thanks to @unclewebb for that.

 

Thanks a lot for the very detailed response! I think I do understand these a bit now. I will try to always use Cinebench then for CPU testing. My GTX860M does run 60fps cinebench, when it tries to use GPU, so it might not be conclusive. I might no fully understand how to use it.

Is furmark bad for GPU testing? (I read somewhere on NBR that furmark is dangerous and it might fry up your GPU)

I downloaded Cinebench R15. Is it goo enough for GPU testing? If not, what is the best GPU oc testing utility? All 3Dmark11 seems to be paid, with no option for free testing.

 

Well, Furmark, MSI Kombustor, or EVGA OC Scanner benchmark tests (and each of those suites includes a few of them) are generally brief timed tests and not too rough on GPUs unless you are having thermal management problems to being with. It's the run-forever Furmark "burn in" and similar testing that are severe and can cause damage. Those remind me of stupid teenager games like trying to decide who is the biggest tough guy based on how long a moron can leave his hand sitting on a stove burner for the longest time, or who can stare at the sun for the longest, or who can drink the most alcohol before passing out or dying from poisoning. AMD and NVIDIA have implemented protections to cause GPU throttling to mitigate the potential for damage with those crazy and meaningless burn-in torture tests.

There are different types of GPU tests. Cinebench and a number of other utilities are OpenGL tests. There are tests for different versions of DirectX, Mantle, Vulcan and other APIs as well. Each type tests a different aspect of GPU performance. OpenGL is generally not useful for SLI and CrossFire performance testing. Cinebench has CPU and GPU tests. To run them, you click on the buttons labeled "CPU" and "GPU" (respectively) and there is also an option for testing single-threaded CPU performance. This is where some lesser CPUs can sometimes keep up with more robust CPUs since there is no hyper-threading and only one core is tested. Other examples...3DMark 11 and 3DMark (2013) suite are focused on DX11, of which 3DMark 11 is more strenuous with heavy emphasis on PhysX and tessellation. 3DMark Vantage is a DX10 benchmark.

In spite of it being roughly 2 years since Windows 10 came on the scene, there is almost nothing out there for DX12 performance testing, or even gaming... I think you can count them on one hand and have some leftover digits. Can you say "bait and switch" boys and girls?

 

Talking about rendering videos :



This is without a delid, on a rather hot day (Around 25C+).

@Mr. Fox How much of a hit on performance does keeping the cache @ a lower clock than the main cores lead to? I haven't noticed anything different yet.

 

Thanks a lot for expanding. I will try to look into 3DMark as soon as I can. Seems like there are quite a few versions of it out there, and it is a bit hard (as it is with all software) to get it without trial and error.

On the other hand, I think that this benchmarking and testing and OCing can get a bit addictive after a little while .

I am not going to do torture tests, because I see little reason for it. I will have anyways lots of work to do, and to analyze, so I will know if something was not up to the task.

Are those your temps while idling, or when doing light workloads?

I see some usage, but way under 100%

 

Not much most of the time. Similar to higher memory clock speeds, it typically only matters much when you are splitting hairs or just need to eek out a few more points to take a higher spot on a leader board away from the guy ahead of you.

Nice job on the temps. Those are really good.

 

Light loads, but with the System fans set to minimum or whatever setting they use at lowest settings. And im kinda sitting next to the window with sun shining on the system. Not sure what After Effects uses for video rendering , CPU or GPU (i do have CUDA support enabled) . But things are super snappy.

System idles around 25C-30C when the ambient is around the low 20ish and im not basking in sun light

 

Isn't big difference with a lower cashe vs core clock. With a lower Overclock, you normally can hold core and cashe 1:1. But with a high overclock this will be a bit more difficult.
If you want to show how your processor do it in load... Show also processor package power. This monitoring in Hwinfo is very useful for you.

 

This was on load. I didnt bother checking the draw, but im guessing its well under the 200W mark.
This isnt from today btw, i have dropped the VID by another 20-30mV since then.

 

3dmark11, cinebench are good tools to see if CPU OC is stable or not. But sometimes even OC setting that could run those tests without crash can cause crashes while gaming.

A few days ago, I had some crashes while playing "the division".
It wasn't temperature problem, but merely voltage problem.
I had to give more voltage to avoid the crash which didn't happen with cinebench, 3dmark11 or 3dmark.

I think we need another or better tool for CPU OC stability and maybe prime or linx would be a possible solution.

 

So, we have another world record holder among us! Fantastic job, Brother @Prema! WOOT! You're basically 105% OC, LOL... crazy.

http://hwbot.org/submission/3168328_prema_cpu_frequency_core_i7_4702mq_4500_mhz

 

Most of the tools that hammer the CPU real hard, arent really suggested to be used with laptop computers. The thermal dissipation is just too much IMO. But out of all these tools, i do prefer Prime95 (smalls 15-30min) runs to stability and IBT/Prime95 longer runs for desktops.

 

Wow do you have any idea how the multiplier lock was unlocked?

 

Nope, not sure how he did it. All I know is that is a pretty amazing feat for a 4702MQ.

 

Could this be Prema majik? Or could this be love ...
♥♥♥♥♥♥♥
Edward? ... Thankyou. ​

 

@bloodhawk

I see you are using a 1.275V static voltage. How is that going? what is your Cache Ratio? how are your temps?

I never had good luck with static voltage.

 

All A-OK. Im actually at 1.235V right now with a +20 uncore.
Cache ratio is at 4.2 with -150mV
Temps are amazing - At the automatic fan settings web browsing and light workloads - 25-35C (When im not sitting under the sun lol)
With Maximum Fan Speed and 100% load around 75C-79C max on the hottest core.
Im just waiting on my display cable to come in, so that i can delid and change my LCD in one go.

 

http://valid.x86.fr/bnygnh

Using a bios that contains the original Microcode 0x1 is the key. Intel likes to pretend that the maximum multiplier is fused at the factory but the truth is out. Run a modded bios with the early microcode and you too will be able to convert a locked CPU into overclocking heaven. @Prema definitely has some experience with modding a bios or two.

Congratulations. Impressive overclock for a mobile CPU.

 

Yeah MSI patched and published a version, based on such a microcode, for their boards and was instantly greeted with cease and desists order...so don't expect this to come from me in a public package either...
It's also only really good for privately playing around as in not being very practical (needs a reboot into BIOS for every two steps you want to raise the multi) and misses a lot of fine-tuning and corrections Intel implemented for these chips with later patches.

 

Unlike previous clevo laptops, static voltage mode doesnt work properly as it should in dm skylake models.
Static voltage worked great in zm models and as a result it was very easy to make things stable because voltage doesnt fluctuate. Hope this can be solved in future bios.

 

Me, too. Not sure why it doesn't stay pegged on whatever you set the voltage to. What we do not know is whether or not that would cause throttling. Hopefully, it would not, but Skylake is very voltage sensitive and it might.

 

did you ever use Adaptive with a core voltage offset? I know you prefer static voltage always but unlike Haswell, setting to static voltage in my system causes temps to go wild

 

So i attempted to overclock to 4.3 @ 100 mV . It seems to work fine until I do a benchmark. It drops back down to 4.0. I also noticed it does it @ 4.2 @ -150 mV. It doesnt say im throttling though. Any ideas?

 

what are your power limits (ie. Cache ICC Max) and what are your Turbo Power Limits and Short Turbo Power Max?

If everything is set according to the pics in my guides, then I have a feeling it's your BIOS limiting something

 

120.00A and 130W I thought it was sorted at my BIOS update but maybe not.

 

I am thinking that the sensitivity of skylake is due to this voltage fluctuation. In desktop skylake where voltage stay static, it doesnt seem to be this much sensitive.

 

Ive tried both static and dynamic mode. and I am thinking they are very similar in dm clevo laptop since static mode also act as if it is dynamic mode.

 

http://www.3dmark.com/3dm11/11115261

 

You are bringing Cats to a bird fight. Freaking insane.

 

Check the power draw with my awesome @Prema BIOS and vBIOS.
I logged the values from my HWiNFO64 sensors, including the UPS power meter and put the data into an Excel pivot table. I did this with quad 330W AC adapters just to make sure it would have more juice than it could ever consume. You can see my AC cooling setup in the foreground and the quad 330W setup on the long table to the side.

 

Holy ****.

Last i saw these type of figures was on my desktop rig running the 3930k @5 Ghz with 2 x GTX980's (+450/+500).

Wth do you even do with 4 Power Bricks lol.

 

Not really necessary, since it doesn't draw that much. Just wanted to test it to run without any AC adapter capacity limitation with 2 by using 4 for good measure.

My palm rest is so freaking cold right now, LOL.

 

I just noticed the AC cooling setup, literally went - *snort* Giggle.
That is as ghetto as it gets, let me go get some duct piping and set it up through the window.
I just cant freaking wait for my display cable to come in , so that i can do my panel swap and delid in one go.

 

@BJ > Panther's finally endowed with the pricey 4960x you wanted for it. Awesome sauce. (woot?!)

Spoiler: beware of klaws & (easter) bunny kicks

 

Ghetto? C'mon man, don't say that about my work bench, LOL. I've got my hot air soldering station, alcohol, aluminum foil tape and box of goodies like copper modding stuff, assortment of thermal paste, thermal pads, thermal epoxy and what not. It's a computer work bench... not supposed to look pretty.

 

Haha ma bad, i meant it in a good way. Id stick my laptop to a vent going to a window with an intake fan if i could lol.

That some insane amount of hardware, cant wait to get my own place and setup a workshop.

 

You are guaranteed alone inside the bench room I think . Your wife don't even dare to come in with coffee for you. She'll freeze to death if she is not accustomed to a regular Norwegian winter . Better that you're all alone in there .

 

...

Don't forget the 'bat-phone' line to custom Prema Mods...

 

It's on speed dial...

[parsehtml]<iframe width="640" height="360" src="https://www.youtube.com/embed/QQjE3U1Oq5Q?rel=0" frameborder="0" allowfullscreen></iframe>[/parsehtml]Edit: that music is being played on a flat-ended shovel, LOL. For real...

Spoiler: Rob Scallon Music

Also check out his "Enter Sandman" video play backwards.

 

These look like new settings brah, I am tempted to test them but the last time I tested your settings it corrupted my boot loader on my RAID 0 setup. I should theoretically have a better CPU than yours since it's a delidded Silicon Lottery 6700K but seems not. Unless the ambient room temps are making all the difference in our overclocking scenarios.

PS: can y'all imagine if we hadn't created this thread, how crowded/overflooded would the regular owner's lounge threads be with all these overclocking settings/results

 

@Mr. Fox

What IME driver version are you using? What does one gain in updating it anyway?

@Prema

Like I notice with every Prema BIOS update, it says it's a new ME FW version, what does that mean anyway, should the driver installed for IME match the BIOS IME? Can you explain this as it's causing me a lot of confusion.

 

Another drivers question guys, I noticed if I ran the Clevo Driver installer from the disc, the first thing it does is disable UAC then reboot, then after all drivers are installed and you hit the EXIT button, it re-enables UAC and reboots.

Is that an essential step for installing drivers or what? I mean I always right click on the setup file and choose RUN AS ADMIN then accept the prompt, what difference does it make if I disabled UAC?

I don't disable UAC as some programs don't behave correctly with it disabled like Internet Download Manager for example.

 

For the best results use an ME driver version equal to or greater than the ME FW version. ME firmware is what enables some features, including some for overclocking. It doesn't necessarily always enhance performance, but using the wrong versions can hinder performance. If everything works well, unless it adds features, there is generally not much benefit to newer ME firmware. Installing updated ME drivers if you already have a driver version that is a match to or newer than the firmware version doesn't really do much as far as I can tell.

I have not found anything that misbehaves with UAC disabled except some of the Windows 10 crapware. I remove all Windows 8 and 10 crApps and Windows Store filth with DSW anyway, so it doesn't make any difference one way or the other. IDM works fine for me with UAC disabled. What is it doing to you that is not correct? I use Windows 10 maybe 1 out of every 40 hours of computer use, so maybe I just haven't noticed.

But, UAC is extremely annoying to me. Any time it prompts me to approve an action I am instantly pissed off by it. I've despised it and run with it disabled ever since they first released that piece of trash in... what was it... Vista? When I do a clean OS install the very first steps I take, before installing the first driver, is change folder options to always show menus, show all files including system files, restore open Windows after reboot, show all desktop icons, disable Windows Updates in Group Policy, disable balloon tips in Group Policy, disable all Toast Notifications in Windows 10 in Group Policy, disable "Limit local account use of blank passwords to console logon only" in Group Policy, enable Legacy Boot Menu and disable DSE with the command prompt, disable hibernation and sleep features, disable all power saving features, set my lid to do nothing, set my screen to never turn off or dim, set power button to shut down, processor 100%, set Windows Sounds to none, show all system tray icons, disable Action Center/Security Center system tray notification flag nag-o-rama icon, disable the Charms Bar and all hot-corners for Windows 8.1, and disable UAC. After all that, then I start working on drivers. Once the drivers are done, I give ownership of all folders and files except those in the Windows folder to "Everyone" with full control access over all folders and files.

 

There is a little OCD Devil in my head telling me, when you have UAC enabled and you run an installer or program as admin, it does actually give it full usage rights/permissions so it installs/works correctly. If I have it disabled, yes I won't get the UAC prompt, but maybe the program is installing/running with something missing since I didn't give it that permission to run in full mode as admin.

When I first install IDM, if UAC is disabled, it gives me a popup that IDM cannot work properly as.......wait a minute, my bad, I think that was when I was running Windows 7 using the built in admin account...
my bad

 

Tested SuperPi - 1M (4.8 GHz) Score 7.574 sec

http://hwbot.org/submission/3172040_papusan_superpi___1m_core_i7_6700k_7sec_574ms?recalculate=true

 

What the heck is that? Never saw that benchmark before.

 

@Papusan

Did you try the Windows Defender Performance tweak? It makes a big difference:

Fix for high CPU Usage in Windows Defender

 

You're welcome!! http://hwbot.org/benchmarks/

 

No worries. I have never seen that prompt with IDM, so I cannot comment. I do not have IDM installed under Windows 10, so that might have something to do with it as well.

 

@Mr. Fox have own mod of this version http://forum.notebookreview.com/thr...od-for-780m-880m-and-980m-mobile-gpus.767644/