Clevo Overclocker's Lounge

Here's the weird thing, it should not be doing this if everything is set to stock and is unchanged after a CMOS reset.

 

When I used Mr. Fox's static voltage settings and even going as low as 1.185V the system would either overheat so much or hard lock.

 

thats the finicky part, u cant just take other user's settings and apply them blindly. ull have to test out each individual cpu and find its respective sweet spot. otherwise ure giving away tons of tuning and thus performance potential...

Sent from my Huawei Mate 8 NXT-AL10

 

MeowMasterXP

 

Hello Scerate

if you are the owner of a prema modded PM870, then get rid of XTU, reset Bios to default settings and have a look at the

wonderful video made by ( Master ) Phoenix with different overclocking options: it's very complete, accurate and simple.

For example, i'm currently @ 4.4 Ghz on all cores, ultra-stable.

By the way, Is your CPU a SL one and is it delidded + CLU repasted for better temps ?

Best regards

 

To be fair, there are two parts to this. First, they are on skylake and us on haswell. Their adaptive voltage settings are more mature and may help their cpu better. I can't say for sure as I haven't played with one yet. Second, the reason static is a benefit and a curse is that it idles at a higher voltage, but it also doesn't allow for the voltage to blow past to the next voltage the cpu says should be used for a higher mhz, a built in "benefit" of adaptive so that if a hard running bench is ran, it instantly views past the offset voltage it should be running. This may be fixed in skylake. I do not know. Asus benchtest caused this on haswell.

Static allows for safe stable guaranteed work without the higher voltage so that idle temps may be higher, depending on the chip, while temps are lower on load. It's a trade off.

Sent from my SM-G900P using Tapatalk

 

on haswell tho it wasn't a Problem cause due to enabled C-States the cores were mostly sleeping anyway or be turned off completely, so static wasn't as problematic as in previous Generations and like you say, even prefered for OC. For example my old 4790k couldn't even undervolt -50mV Offset stable, got crashes when just browsing, so i guessed some lower Clocks are the issue and with static voltage i could lower my voltage from stock 1.22v to about 1.08v and with Cstates on i absolutely had no issues with Wattage

 

Not the big change in load temp with Statick vs. Adaptive voltage for skylake. Only in idle. Longevity is neither, a big topic

 

He also tested with one I'm not sure called for extra voltage when started. I know p95 28.x and realbench did. That is where a difference can be made...

Sent from my SM-G900P using Tapatalk

 

oh and back to my issue.

I reset NVRAM CMOS yesterday again then flashed Prema Bios then after the initial Setup i reinstalled Windows completely, only installed main Drivers and redid the OC, still no go sadly, same throttling as before, CPU only is still no Problem tho, only when GPU get's involved the Mainboard ****s his pants. Maybe power Distribution issue on a Hardware Level, namely Mainboard Power Delivery, you guys have a P870DM i guess that issue isn't there

@Mr. Fox have a vid on YT what exactly describes my issue on the most Point LINK

 

Either ways, it shouldn't happen at stock. Specially after a CMOS reset and without touching anything. System should be fine. So its either the EC issue or some hardware problem. Doubtful about the EC, because Prema's BIOS fixes that.

 

yeah shouldn't but every clevo is an Adventure somehow for me, P150HM got the Voltage unlocked for the 7970M and 6990M woes, P150SM Powerbug with the 4700MQ, P771ZM killed screen due to Win10 Update, P771DM no OC with GPU, let's see what the P870DM3 brings for me

 

Do you have a second adapter?

 

I do not know what to say. Same problem with stock settings? Also try with different voltage offset for your OC.

 

@Scerate

Are you online when your installing your drivers? If your using W10 that can cause problems as it will more than likely beat you to the job and install what it wants before you get a chance (and without telling you) . Also the Anniversary Edition caused me Major problems.
One more thing.. Which version of the Control Center are you installing. If it's the newer one XTU could very well be reeking havoc in the background even if you never installed it.

 

not with me atm, have the original 230W Adapter at home

yes mostly the same Problem with stock, but mostly i get power Limit throttle instead of the weird PL4 current throttle.

Well right now i don't have any Control Center installed, and no i always install without Connection then after installing MY Drivers i connect.

Edit: cool side effect tho, i repasted for the sake of it and noticed that the Thermalgrizzly Kryonaut isn't a good paste for CPU, just applied some quick Prolimatech PK3 i had with me and my CPU temps dropped to about 50 now instead of 60

Edit2: CPU only seems very very nice now at least temp wise, and no it doesn't work yet, CPU only worked everytime

 

ok i close my case, i think it's definitely a P7xxDM(-G) thing, big thx to all who helped me ( @Papusan , @bloodhawk , @TBoneSan , @Phoenix , @Mr. Fox and everyone else i sadly forgot to mention) to adress it but looks like i have to wait for my P870DM3, i just found some posts from @godfafa_kr who has the P75xDM and have EXACTLY the same issue.

So sadly i have a half Turd

 

Did you disable C States, Set Turbo Power Max and Turbo Short Power Max to 200000, then set the VR PSYS Current limits to 8191 and same 8191 for the other 3 VR Current limits like Core/Ring/etc AND set the power limits 1 , 2, 3, 4 in CPU Configuration to 300000??

 

sadly yes, i already tried everything, no dice.

 

when do you see the throttling? is it when the GPU is used at the same time or any CPU stress test?

 

Also tried to put VR voltage limit up to 2000 on core, ring and SA domain?

 

yep, put everything to the max, in the Bios Amps to 8191 and tried Voltage Limit too to 7999. CPU goes down when GPU gets involved, if i exit the Game to CPU boosts to my values.

Even tried reverting the Prema VBIOS to the stock one, at first it seemed to help but then i noticed my GPU throttles pretty hard and that was the reason the CPU could handle the turbo better but after a while even there the CPU gets punished.

 

HWINFO64 tells you its an error with ring power?

Have you tried to underclock/undervolt ring as a worst case scenario?

 

I'm going to assume that's a typo, otherwise there is your problem.

 

not yet, but i could try that, just tried Deus Ex on that NB, holy crap my 980M sucks and that with OC, 41fps on average on mediumish/high settings.

nah not a typo, but didn't matter i tried 1000 i tried 2000 i tried 1300 well just nothing worked sadly.

 

Dont worry about it. Its a tough game.

 

Yes it is the mobo limitation. I now use p870dm and i have no throttle whatsoever

 

Stock both P7 and P8 have the same firmware throttle.
It's just that we had a P870DM with the 'permission to kill' during development and as such could properly test and finally go for higher values for its Prema Mod than for the P7 Prema Mod...

If I find the time and you guys are adventurous we can also push the P7 Mod further, no problem...

 

Thanks again and I always appreciate your work, Prema.
BTW, are you planning to make Prema bios for the new p775dm3?
And would your bios for p775dm3 be more like p870dm bios with "higher values" ?
Or only p870dm would have the "higher values"?

Please let me know. I am considerrng buying p775dm3 but if it also has the throttle, I will not.

 

We can do whatever needs to be done, but we will first have to see where people hit their RL usage limits and then test on actual hardware to know how much more would make sense.

 

Would be great if you can find the time for a sideproject like this, but it isn't necessary i think cause most enthusiast already have the P870DM and i don't know or see if other P7xxDM users need it besides me now. It would be much more important to get these P870DM2/3 beasts going , and ofc your future killer machine. You know i'm extremely grateful for everything you do and appreciate it a lot you put so much hard work into it to make some of us sooo much happier with the Clevos we got . I still on the verge about getting a P775DM3 or a P870DM3 but now it tends to be more on the P870DM side, just to be more safe than sorry. Oh and about the P7xxDM Mod you mentioned, i don't even know if the P7xxDM could handle it, cause e.g. my 980m still kicks the bucket sometimes and getting black screen even at stock when i'm not capping my fps

 

Prema has to do that for his partners. I think he was saying if someone is willing to push their machine to death and put in the work, he'd get a good working firmware for you guys... But work comes first...

In any case, that doesn't sound right on heat even for ZM/DM machines. What have you tried to get the heat down, it is the gpu heating up because you have a cpu running hot?

Sent from my SM-G900P using Tapatalk

 

well cpu is pretty cold now at, least with the 4ghz i can achieve, maxes out at 55-58 °C, gpu get's to about 70 °C and pulls about 130w-140w according to HWInfo64, i even repasted today again with the Thermal Grizzly and cleaned up a bit of residue i had left from the liquid metal, applied some more preasure than before and it seems to help with that low viscosity paste. So besides the no OC while gaming issue i have, my book runs superb.

 

Do temps slowly climb over time? If they do, contact issue. At those temps, heat isn't an issue unless contact isn't being made on a component without a sensor or feeding the gpu too much voltage... This isn't addressing your throttling, but the black screen, btw...

Sent from my SM-G900P using Tapatalk

 

yeah yeah i know you mean the black screen issue, but no it mostly gets hot pretty quick to about 68 then slowly to 72 and stays there, and when i repasted everything sits nice, checked it. Seems like my VRM gets too hot, i had that black screens when doing some firestrike runs at higher voltage with 1330. But i could try a slight overvoltage too for the GPU, black screen is mostly VRM what i noticed but i had the same when trying to undervolt my 980TI.

 

make sure all your thermal pads and their placements are fine!

Sent from my Huawei Mate 8 NXT-AL10

 

Building on that, @Scerate, check whether the bar got bent. This can effect thermal pad contact as well. Kinda makes you wish they put a contact screw up there, but also an easy fix...

Sent from my SM-G900P using Tapatalk

 

Recently Ive upgraded to redstone and installed 372 nvidia driver.
Ive noticed some down score of CPU benches after these two changes. Anyone experiencing similar problem?

 

how much of a difference? anything in the range of 1-3% cannot really be considered significant...

 

Nope, scores are actually up by about 5%, compared to previous stock clocks.

 

might be time for a fresh install then maybe? @godfafa_kr

always good to have a nice clean system

 

Weird thing is that the scores improved on my daily crap install of Windows. Wish i could run these on my Scientific Linux install , that baby is fast and smoooooth.

 

The difference is about 3dmark FS physics 200~300points. Number is little but I am sure it is significant.
Before a clean install which I really hate to do, I think I might try to uninstall avast free.
I saw some article about avast and restone conflicts, I am not sure it is true though.

Even after uninstalling avast free ver., every cpu benchmark scores didnt come back. I think I should clean install redstone and see if it works. Thanks.

 

Avast wouldn't cause it to slow down. I use to run avast all the time. Now i even run Symantec endpoint and still i'm good for benching.

 

To be fair, symantec endpoint (just as all Norton AV) has low overhead. The two lowest, I believe, are kaspersky and bitdefender. In fact, according to AV testing agencies (not the mag reviews), those three regularly arrive in the top five, usually top three, and all have low overhead. I haven't checked avast out, but I hated sophos endpoint!!! Large overhead, horrible av scores, and didn't even find an old virus that my friend with bit defender Mac found on my external... These AV testers also give certifications and one recently published either their June or July review...

Edit: but he's right, most likely not the culprit...

Here are a couple companies I look at for reviews:

https://www.icsalabs.com/products?order=field_version_value&sort=desc&tid[0]=4216&items_per_page=30

https://www.av-test.org/en/antivirus/business-windows-client/
https://www.av-test.org/en/antivirus/home-windows/

http://www.av-comparatives.org/performance-tests/

Sent from my SM-G900P using Tapatalk

 

How about redstone? Does it influence any cpu performance or bench score in general?

 

Please read bro. I have been in touch with KOKKO on Futuremark for over a month now(debugging for him). Finally something will happen.

Spoiler: Low physics in latest 3DM UI

 

http://www.3dmark.com/fs/9858701

 

Okay so you experienced similar thing too?

Oh now I can see the img.
Okay so it is the 3dm UI?
I am experiencing the same with 3dm11 too though.

 

Futuremark v/KOKKO < can confirm that this is now reproduced in-house and investigation is ongoing.>

Same 372.54 driver and stock graphics. The only difference is 3DM UI versions. Futuremark crippled last 3DM version. Something in the latest (now 3 versions) eat processor cycles. Will be fixed. They have released two new versions already without fixing the physics bug The first one came 18. Aug , and the second one 19. Aug.
http://www.3dmark.com/compare/fs/9848576/fs/9924347#