Clevo Overclocker's Lounge

You have dips in clock speed. I would try an other driver or change a bit on voltage. Up or down. Skylake is very sensitive to voltage. Same as old Haswell. Remember different benchmark and workload can behave differently on processor. Try what I suggest. Even try with static voltage

 

not coming near static voltage with a 2 foot pole. It almost fried my CPU last time.

But what I noticed in the above graph (thanks alot for telling me about this graph, never knew about it) is that the clock speed dips are for a second or so, nothing to worry about I guess

 

@Papusan

Please change your custom user title saying BGA bla bla it's old now. Please change to Overclocking Doctor

 

NOOOOO. I want to show what I feel about BGA junk!!! Maybe I change to a better BGA text
Maybe you could try your former OC profile? I hate dips in clockspeed and shouldn't be there. Or try other driver.

 

nope, all drivers to this. Also, as I said, I still have this dip with a 4.4 GHz OC

Try it on your system and post your HWiNFO screenshot to show min CPU and also post the 3DMark details.

PS: Don't use HWMonitor, it hasn't been updated in ages and doesn't display GPU clock speed correctly. Shows 504 MHz always

 

Default graphics aka zero OC graphics . This was rarely done by me. Only tried this default clockspeed on graphics one time(When I got my machine), LOL
http://www.3dmark.com/fs/8227051

 

what the heck, please post every screenshot of your BIOS settings

Also, were you using Throttle Stop during this benchmark or not?

 

A new Little Baby Girl OC on graphics (200/400). Here is Hwinfo64 monitoring with score. No dip in clock speed @ 4.8GHz + max fans that I always use in bench. + I use Throttlestop.
http://www.3dmark.com/fs/8227210

Be happy with what you have. Dont use Hwinfo or any OSD under bench. Then there will be no frustration + you get better bench scores I will delete all of this Firestrike results in Futuremark site. I dont like a lot of bench results with baby OC on graphics in my result page at 3Dmark
An other Firestrike with stock gpu speed 11929P http://www.3dmark.com/fs/8226665

 

@Phoenix + a lot of others should put up Hwinfo64 more like this in the image below. Maybe some want an other Hwinfo setup than what I have, but It becomes much clearer/easier when you want important info. It is very strange no one has heard @Mr. Fox talk about this importent topic. I also preached about this a long time. If you want Hwifo on your computer, do it properly...

 

please send me your INF file of HWiNFO so I can make mine look like yours.

 

PM

 

A little higher score now, I overclocked the GPU by +100 Core / +200 Mem

3DMark11 - GTX 980 (364.96) CPU 4.6 GHz-GPU OC 100-200 [W10]



Fire Strike - GTX 980 (364.96) CPU 4.6 GHz-GPU OC 100-200 [W10]

 

Have you tried to disable all the c-states?
I dont own p870dm but a p750dm as you know.
When c-states were enabled, my clock speed fluctuated as yours but not now after I disabled it.



It is very hot here in Korea, look at the min temperature
With stock fan without any other ext. cooling system, the min temp is above 50 and max 85 after playing "the division" for 3 hours with "OC fan table".

 

it reaches up to 48C here in Dubai bro. Right now it's around 33C

Room temp is always around 25-26C though due to the AC

I will try disabling C-States, totally forgot about that.

 

when using premamod bios do we still need xtu? or it is redundant? also, any videos on showing how to run/use the bios?

 

No need for XTU. Uninstaling it though will not fully get rid of the damage it has done. On a complete format will. It messes up the fan tables. Like with my previous systems, the fans would start spinning up on full blast randomly even if the temps are low

As for the walk through, check the videos / settings in this thread:
Clevo P870DM n00bie's guide

 

So you suggest a clean install of windows? Uh oh. I did it already when i got my system. Will do it maybe tomorrow its a pita to do...

 

well you can try without formatting, see if you encounter any abnormal fan spinning even when not doing any heavy tasks

 

very nice video. just halfway through it. now i wish i purchased the 870dm rofl. very informative

 

So my GTX 980 Died today morning. Was playing Dark souls 3 for about 2 hours and then a black screen with audio in the background. Force shutdown the system, and have had a blank display since then.
Tried each RAM stick individually, and cleared CMOS, but the display has been blank.
When i turn the system on, both power and battery LED's are green, the HDD's are spinning up and doing their thing.
As a last resort and to make sure it was not a dead LcD i put in a GTX 980M from my friends Sager NP9870 (Gsync) , and the system booted perfectly fine.
I have not seen a GPU die like this before. Day before yesterday, when playing Division, the temps were around low 80's after 3 hours of playing. Which is pretty good for the game maxed out on a 4k display.

 

I know you got a hold of rma before even coming here to report it right?

 

Nope. Unfortunately not.
At this point im just hoping they dont ask me to send the whole system over, which will be stupid tbh.
And are i able to cross ship me an new one.

 

Hummmm, if it was clevo. They use to want the system to come to them so they could verify it. Not sure if it's still like that. And if it's a reseller, i guess you could always ask, but they might want you to put a card on file just in case you forget to send the dead card back....

 

Yeah i dont mind that. But At this point, id do just about anything to get a replacement card shipped as soon as possible lol.
This was bought from HIDEvolution. I was busy all day so couldn't try the 980M before their closing time, so might have a reply from them tomorrow.

 

Sounds good. Now what you need to do is take 30 minutes to an hour out of your day and go over that card with a microscope. And if you don't have one of those, then a pair of binoculars will do. Just flip them upside down and grab a light. See if it looks like a bad vr or something that may or may not be easy to fix or find for that matter. Then lets us know if you find anything.

 

Will give that a shot tomorrow.
Btw. Does it matter for RMA if i remove that black gasket around the GPU core? I thoroughly cleaned the card today with Isopropyl Alcohol, and there some serious amount of sticky thermal paste gunk mix under it. So i took it off.

 

Yeah, well....that black gasket was kind of there for a reason...., but I think you should still be ok.

 

Just heard back from HIDEvolution, and hey want me to send the whole system in. Which is rather stupid.

 

sorry I missed it, what is the issue again? why do you need to send your system in?

 

GTX 980 go bye bye. Ted say send system san back. I ask if can send only card back.

 

hmm, maybe because it's like a major component? not sure, did you ask them why can't they send you the new GPU and you send the old one back?

 

sorry lost here bro, black gasket was the reason for......?

 

hi dudes yesterday i was playing around with the gpu.. i tried running it at stock voltage oc'd to +100 / +200 (core / mem). benching went ok but the division crashed after a while (black screen). then i set the oc to +100 / +150, launched the division and after a while the game froze forcing me to hard-reboot. i then went down to 95 / 120 and it ran without any problems for the time i tested it. i then raised the voltage +12.5mV and tried 100 / 200 which ran, but i noticed in the benchmark the gpu score went down a bit. now other people overclock this card (gtx980) a lot higher (i.e. +100 / +200) with stock voltage.
also i set the system agent voltage in throttlestop to an offset of -30mV. does this setting have anything to do with the gpu? meaning can this cause instability? or do you think i may have just had bad luck and my card is just not the happiest overclocker? ^^
cheers from switzerland

 

Well back in business. http://www.3dmark.com/3dm/11707054
I took the card out, plugged it into friends system and VOILA it works. But doesn't want to boot into Windows .
So flashed stock BIOS, and it booted. Weird. Flashed premas vBIOS ,and still working fine. Ran firestrike , temps are also great. No issues on my friends system.
I put it back in my system, it boots and has been running Unigen Valley just fine for the past 40 mins.
I have no idea what is going on .
@Prema
Any thoughts ? Or idea what it could be ? Can it be the 4k panel dressing it too much?

 

Sounds like the GPU internal falcon security chip disabled the card (a protection measure) as a result of the crash, but couldn't complete its cycle.
It's the same chip that throttles the GPU to 2D clocks after a crash at very high voltage/clocks.
Taking it out from your system power cycled it and the vBIOS re-flash triggered a full reset.
Test the card properly, but it should be back in business now!

We will see a lot more of this 'magic' with the next gen...

 

Gotcha. Yeah im running 3D Mark Suite and Unigen right now. Seems to be A-OK for now.

Such a tease you are ...

 

@Prema @bloodhawk Rep added. Learn something new every day. + rep on both .

<It's the same chip that throttles the GPU to 2D clocks after a crash at very high voltage/clocks.
Taking it out from your system power cycled it and the vBIOS re-flash triggered a full reset.
Test the card properly, but it should be back in business now!
We will see a lot more of this 'magic' with the next gen...>
Can you do this with BGA junk?

 

*Please not on desktop cards.... or in general. More security is never a good thing. Nvidia is being silly again and AMD is not much better with the OC on fury x. Stop locking down cards....

@Prema any idea if cutting the traces to the falcon chip would disable it? I am completely open to doing PCB surgeries and voiding warranties if I dont have to deal with this non sense.

 

So this is what im bummed about, i can OC my GPU core easily to +250Mhz with a +.50mV , and it runs and benches fine, but the moment i start overclocking the memory i start getting driver crashes and black screens.
So is my memory just not good enough at overclocking or could it be something else?
I made sure the thermal pads and heat sink were making good contact with each and every component on the GPU.

 

Does ıt crash too when you try underclockıng the memory? Could ıt be somethıng wıth changıng memory clock?
but probably just bad luck.

 

Im not sure how to do that. Nvidia Inspector wont let me go under 0Mhz

 

Try a 362 driver...that's still the one NVIDIA uses for 'every' ODM out there:

362.19 LENGDA
362.13 LENOVO
362.09 ACER
362.04 GRID
362.03 ACER
362.03 CLEVO
362.01 FUJITSU
362.00 GENERIC

 

Did you try OC'ing the VRAM with the Core non-OC, +0?

It's likely some sensitive VRAM...

Didn't your 980 die just recently? Is it already replaced, or did you resurrect it?

This might be why it died, some VRAM chip is flaky.

 

Well swapping it around into a different system and re flashing BIOS, MUMBO JUMBO. Made it work again.
I'll give the memory only OC a shot.

 

http://www.3dmark.com/3dm/11713973
Well just passed with a +250Mhz on the memory.

 

That's interesting

If it was VRAM, it should have failed to OC, but if it's power related...

See how far you can OC the Core with the VRAM OC @ +250

 

And now with +400Mhz O_O

http://www.3dmark.com/3dm/11714048

With +125Mhz / +250 Mhz

http://www.3dmark.com/3dm/11714130

Any idea how much power the CPU draw at max load @ 4.5Ghz ?

Finally Black Screen'd @ +175Mhz / +250Mhz with +.25mV. Temps were low 70's

 

Probably a good time to buy a Kill A Watt, or some such from a Home Depot like place.
http://www.p3international.com/products/p4400.html

IDK what the power draw is, but it looks like you found the optimized MXM card *limit*, which I thought was 200 watts.

So it's surprising that you found a power limit before a GPU / VRAM limit. I am assuming the card isn't up to spec on power handling? Others, including you have OC'd the Core further.

It looks like you have debugged to root cause and can report your findings for a replacement 980

 

HWINFO64 should tell you. Look for the package power draw. It also should have data on GPU power draw.

If you are prema vbios and overclocking, you might by passing the 200w mark.

 

Yeah im guessing its probably the MOSFET's topping out. Or one of them is heating up too much. But im just not sure if they will give me a replacement, given that it does this only when overclocking and works fine normally.
Lets see if i can get a KillaWatt.

One thing to note however is that im using a single power brick right now. Second one is at work. I dont think im hitting the power limit on the adapter @ 150+Mhz ?

CPU package power draw is topping out at 65W and the GPU goes upto 220W with a +150Mhz OC on the core.
With a +100Mhz/+200Mhz it tops out @ 198W.