Clevo Overclocker's Lounge

That can only really happen when you use early CPU microcode (and block microcode updates in Windows) which will use too much voltage and due to being broken won't throttle on battery...so first system AC Adapter trips and then system battery gets fried when switching.
Killed my on W230SS battery with the broken code by simply running it in battery mode. That's why I never made such mods based on it...

 

I did attempt the OC mod, but it didn't work properly, and I reverted my microcode to the latest using mcupdate_genuineintel.dll. I can't OC now anyway. I'm just hoping that it's my AC adaptor, and not the charger port that's gone. I'm planning to replace this 120 W one with a Chicony 150 W adaptor.

 

Oh wow, so then Furmark did that all on it's own...
At least our systems use of the shelf two PIN power port, so you should be able to replace it yourself.
Just test the AC Adapter with a multimeter...

 

That's exactly what I was thinking. Voltage ought to be roughly 20 V, right? And if the AC adaptor is gone, can I safely use a 150 W adaptor instead of a 120 W one and expect that my charger port can handle the excess power if needed? And any recommendations for a 150 W, lightweight but reliable adaptor?

 

Only thing that matters is correct voltage and pinout and that the maximum power output is higher than what you need. Unless you have a laptop with a sense pin in the jack.

 

I am using this one (19v): http://www.coolermaster.com/mobile/portable-power-series/usna120/

 

Thank you!

I'll look into it!

Even if i go for desktop, I should be able to use all components in it, no?

This is why everybody in the world reocmmends to avoid FurMark... I mean, I never started it in on P775

 

And those fans is very good if you go for water cooling.

 

Best benchmarks thus far:

FireStrike and its suite - average benchmark - good for everyday usage

Cinebench - very soft on evertyhing

AIDA torture tests - this is the hardest test to pass if configured properly

As for stability tests:

FireStrike

Dota2

Dying Light

Metro 2033

Mirror's Edge

If you can play 3-4 hours of each of those games you have enough free time to enjoy your life and you're having lots of fun! You're also blessed with a stable setup.

Bonus stability test:

Facebook / youtube / folder browsing / idle computer for at least 40 hours. If it doesn't freeze or restart in this period, it usually means that the idle clocks / voltages are golden.

Strictly GPU burn tests for both stability and heat :

madVR configured to use 100% of GPU. Change videos 20-30 times on mid video and seek through video. If the computer doesn't freeze / restart and it doesn't overheat, it means that the GPU is generally stable and doesn't overheat either. This is a truly heavy test that will never damage the hardware, but will push it to its limits.

 

I could run Witcher 3 for dozens of hours on end, is that considered OK?

 

Yes.

Games are the best kind of stability tests and anything that plays for 6 hours straight means that the setup is stable on that game. I take games with different engines most of times to test how the setup fares when tested for stability on varied types of loads and algorithms.

 

Yep, the AC adaptor has kicked the bucket. To be frank, I'm surprised it has stood up this long after being absolutely abused for nearly three years. At least the charger port appears to be alright.

 

Well... so far is looks as if my expectations for RS2 were accurate. Seems like it sucks... as usual. So much for DDR4-3333. RS2 took care of killing that joy immediately.

 

YEEES. As I said to BRUH @Phoenix More Trash from the M$ - NSA factory!!! WINDOZE X Redstone Crapstone 2 Build 15063

 

Thanks for giving me the kick to school myself

I always thought adaptive was something that was fairly constant and adaptive +- offset = static so why bother

Reality is quite different.

Right now, 4.5 perfectly stable in IBT (no throttle!) at "1.160V" (which actually settles just below 1.25V under peak load) and falling plus the scores are a couple of GFlops higher than with 4.4 as it should be

 

Crazy good!

 

It looks impressive but I don't even understand what I'm looking at. What are the copper blocks and tube for?

 

Dry ice.

 

Anyone want to try this ?

http://overclocking.guide/the-truth-about-cpu-soldering/

 

I'n the process of making another one of these setups myself.

Temporary heat sinks. The two smaller ones are for my Kingpin cooling LN2 Pot. The round one is a dry ice/water cooling/air cooling pot. Made by Nate the block man over at evga.

The smaller ones are 2.3 lbs pounds each. And at the time I was testing what happens when you "really" add more copper to a heat sink.

 

What are you talking about man. Adding copper will just increase the Inductance and it will not do any good!! /s

On a serious note, never tried placing dry ice directly like that, any draw backs or risks other than condensation?

Im working on some to scale 3D Models of the GPU's and the CPU Socket (will need to 3D Scan this) , to send to NateMando along with the P870, and see if he can create a giant mono block or maybe even 2 separate water blocks than can fit the DM3.
IF i decide to keep the laptop by the time the models are dones.

 

Yeah, that's what i was trying to explain wayback when. It took longer to heat up and even longer to cool down.

Yes. risk to your fingers if you're not careful. And as to the condensation. I was using a blower to blow it out if I thought it was starting to get wet.

If he has the time and the parts...He can do it.

Other news.... In no way shape or form is this stable.

 

There is no way you are not running a super custom version of @Prema's BIOS LOL.
But wow. Is this the same chip?

 

Yes, it's the same chip, but I can't validate it yet. It's my 5.4 ghz chip

 

Out of the 3 i had , only 1 was able to do 5.3Ghz (and that is now with Q937) , other than that the second one couldn't even stay stable @ 4.9Ghz, the third one that i have now can validate at 5.2Ghz, but only benches at 5-5.1Ghz (5.1Ghz isnt all that stable) . Still trying to see what im doing wrong, or if its the chip. Going to try it in a desktop on Sunday.

 

Lol, copper helps just fine when on a big open setup and it's that much, though for air cooling a tower heatsink would have done much better.

 

Might not be the chips at all...Might be the machine itself or the motherboard rather.

 

Hmm gotcha. I want to find out what it will do under water. Even though those voltages might end up being pointless. Will update if anything changes.

 

I think you are going to need to use chilled water to make a difference......I could be wrong though.

 

Good LORD! I've never seen anything like that. I thought I was getting gutsy when I used CLU. I don't think dry ice would help Alien Ware now.

 

Sure it would, but it would also help it to the grave faster.....

 

In desktop news-

Spoiler

 

Yes, entirely possible. I'd maybe even be more inclined to think this comes into play more so than variance in how the CPUs are binned. Irregularities in motherboards and the quality of their components may also explain why some are more stable with memory OC than others are. I think the QC and uniformity of component quality and engineering R&D that goes into even the best laptops like the P870 is next to nothing and with desktop boards they are more worried about their reputation selling crap than laptop ODMs are. The people making laptops are barely doing enough for them to work right stock, and not even trying very hard to make them good at overclocking. In contrast, I have to believe that the desktop enthusiast board makers know without any doubt that the first thing many of their customers are going to do is overclock it. If it doesn't go well, there is a very good chance that they're going to lose that person as a customer for life. It's really unfortunately the laptop manufacturers don't have similar pressure driving them to take the same deliberate approach to delivering excellence.

 

Have you tested on both your Clevo's and if so, seen a difference? Regarding @Johnksss@iBUYPOWER's reply?

 

Actually have. And emailed Prema about it xD
He suggested something to try, there is more to this than what i have posted.

 

You know, I think this is really well-said, with the exceptions being the recent nVidia GPUs: desktop power in a notebook means the best bins drawing the least voltage but outputting the same amount of power goes to notebooks. The CPU side of things is a rat's nest, and component integrity on notebooks is terrible. The sigma of component lifetime is far too wide, from the simplest transistors to entire CPUs. But then again, 90% of notebooks sold are hardly pushed to their limits, so why would manufacturers bother?

This is why my next PC will probably be a desktop, and I'll use my current notebook as a device to game on the go, or maybe purchase something with a 1050/1150 Ti inside it.

 

Otra ves! I beat my last one this time @Phoenix.

 

@Mr. Fox benching @ 3333Mhz: http://www.3dmark.com/3dm11/12077508#

 

And i'll be next on that list after a few more adjustments....

 

WOW! I didn't know the chips could handle that much voltage.

 

Testing how hot the system gets under AIDA 64 @ 1.2V running 4.8Ghz .



For some reason RealBench only wants to be stable @ 1.245+ V. Every thing else, including my work applications have passed 1 Hour+ of stress testing.

 

I don't think it is actually safe for them to run with those voltages for long

Intel recommends avoiding voltages above 1.45 for long term usage as having voltage higher than this might degrade the life of the CPU and other components faster

 

Here's how 3200 versus 3333 looks with XTU with the CPU at 4.7GHz

3333: http://hwbot.org/submission/3502494_

3200: http://hwbot.org/submission/3502414_

Even better than 3200 @ 4.8GHz
http://hwbot.org/submission/3502426_

@Papusan

 

There's also temperature, and thermal cycling tied to voltage. I would be happier running 50C under water at 1.45V than 90C at stock voltage

Voltage based degradation is basically a logarithmic scale. Going by the paper on 14nm by Intel,
"A 14nm Logic Technology Featuring 2nd-Generation FinFET , Air-Gapped
Interconnects, Self-Aligned Double Patterning and a 0.0588 m2 SRAM cell size"
(Google it for link)
As a rule of thumb I would say every +0.1V increases degradation by about 10x. But that is coming off an incredibly low base, extrapolating the graph to the left means they would be testing CPUs for YEARS at stock volts under lab conditions to observe enough failure data. CPUs basically never ever fail due to electromigration alone at stock voltage it's always something else going pop well before it gets there.

 

OK, you can count me in as member of the '3333 club':

http://valid.x86.fr/w27m99

Spoiler

 

You guys are using the 3000 mhz modules, right ?

 

We have now been able to push 3000 & 3200 samples to 3333...
When we started with the stock firmware even 3000Mhz was giving us memtest errors,
it's been a long journey of flashing, testing and repeat ever since...

 

Damn nice!

 

They can handle far more with the right cooling.

Not all programs will use the same voltage and be optimal...Some use a tad bit more while others use a tad less.

The chip cost all of 280 dollars. Who cares about degradation over 10 years? I wont even have this chip then.

I am, they are using 3200 kits.

Side note. If 3466 mhz was unlocked, I think I can boot that as well.

 

Yeah, its just weird since on the DM1, i dont seem to have the same thing happening. With the same chip.

Chip degradation isnt really a concern to be honest, i wont be holding onto this system for too long as it is. SLI doesn't seem to be my cup of tea, works great for benching and all. But for anything else that i do, its just being a pain in the rear end. Already have a second 1080Ti coming in, might as well deal with it on the desktop.

Whoa, didnt even know there were 3200Mhz kits out there yet. I remember reading about it, but didnt think you guys already got them. The most i have been able to push my 2800Mhz 64GB kit is to 3100Mhz.