Clevo Overclocker's Lounge

pleaase go through 100000 emails and delete the attachments you don't need.

Also, don't forget to empty your trash can

 

Yea, these laptops are really for enthusiast. I mean most resellers wont even do a proper Top quality paste job and you need to repaste yourself/delid/change thermal pads and even the BIOS is handicapped and you need to unlock it yourself (and by yourself I mean Prema )...
I wonder how Prema is doing with his bios mods and only doing some final touch ups or it is still a long way to go until they are finished.

 

Good thing I've sent some large PNG files when I spammed you with mails

 

Yeah...LOL! Funny thing is that I bought a google play card the other day as they have that option for paying when selecting their storage upgrades option.
The thing is AFTER you register it they tell you that its NOT even applicable for additional storage! What a scam!

 

i guess it takes @Prema a longer time with the DM2/3 machines, since he chose to wait for his own phoenix 2, on which he can mod in peace and with more self-dependency

 

There where a few (e.g. supply) problems causing a delay in receiving my system...
Last time we had one of the first P870DM units in US, which was used by NVIDA themselves to demonstrate the 980...
That was a good month before MP units where shipped by factory.
At the same time we had the permission to do 'whatever it takes' to make it work right. This included advanced and dangerous EC Mods which are a 50/50 chance to turn out to be a one way ticket...this time the risk is on my system, so I am triple checking each bit of code before flashing.

EDIT:

Ah, VR just got serious:

http://steamcommunity.com/games/465240/announcements/detail/663547937788619287

 

Ayyyy i see what you did there...

 

so it IS the same issue i had before you made the new mod for my P771DM-G? Srsly why they do that, completely the same opinion like @DRevan

Appreciate your time you dedicate for for godsent mods, and about the OC thingy do you mean due to power limits or something else, i only noticed so far in Overwatch at least when OCing my Clock rates get's used much more often from the 850-900mV range instead of the higher ones 950-975mV

 

To be fair the hardware is a lot closer to the limits than it used to be.

Putting constraints on power does need to be done on a mass market device, removing those limits needs to be done under the supervision of an experienced user aware of what they are doing (I.E don't remove the limits and then do lengthy runs of prime/furmark with overclocked settings in a warm room)

 

Furmark throttle is build into every vBIOS since GTX580M, doesn't matter what BIOS/EC/vBIOS/ME Mod we use.
Every CPU getting too hot will also automatically throttle on it's own.
The DM3 can not even use more than 480W on any stock version without firmware or hardware mods, while we tested the board with a continuous power draw of ~700W without breaking a sweat and no hardware damage even at over 820W...

I can not talk about other ODMs, but Clevo actually makes very good hardware and we intend to make it the best that it can possibly be.

My own mobile GTX1060 for example easily runs 2100Mhz on a bare 1.05v...
BUT on it's stock vBIOS it power throttles like 500Mhz down at ANY temperature even in a gpu-z render test, since it has been designed not to become something more than THEY intended it to be.
Similar goes for mobile 1070 and 1080, which (as we know) in addition to the 1060s power throttle on top also temp throttle at very low temps (47c).
Having GPU OCing unlocked at all in those stock vBIOS is a simple scam to make the regular user believe that it would actually perform better when doing so, while in reality it barely does scale at all under stock conditions...
They simply don't want the owner to make better use of it in order to sell us their next best thing...

 

Nvidia know very well... If the performance gains is too small from current vs. next gen graphics... A lot of people will skip one gen of their graphics

 

We always gave you ~20% extra with the Mods on top of what we had yesterday and intend to do the same with today's hardware.
So the above would only hold true if they stop us from making the best use out of our future hardware (aka shooting themselves in the foot) or just sell us another stock overclocked re-brand, which I would expect them to no longer do...

 

mobile 1060? so what ARE the specs of your hardware now? still curious here i thought u had a 1080 sli setup? or did u get a second machine for on the go? or wait, even better: are you testing an unreleased system / P640 successor?

Sent from my Huawei Mate 8 NXT-AL10

 

There is no 1060 in the P64xRx successors...

Stock vBIOS and a 5 year old system: https://goo.gl/4Jn76E

I invested in a few other things to work on compatibility for you guys...

Spoiler

 

oooooh, ive seen those posts before, just didnt realize that that was your own system nice haha!

concerning the ram, now i understand why u had problems sourcing hardware thats not even properly available yet anywhere else @ddr4-3200 sticks gskill, i presume? but only single channel...? or is that just for testing overclocks on a 3000 stick?

no 1060 in the P640 successor eh....? could either mean good news (1070) or bad news (1050Ti) for prospective owners...im betting on the latter

Sent from my Huawei Mate 8 NXT-AL10

 

Backing the @Prema mod.
Mine.

 

Just did a BIOS reflash and a Windows reinstall and tried FireStrike with basic drivers and without Clevo app. Everything related to teh CPU is on default factory settings.

Result of a FireStrike run:



(CPU temp during throttle was in the 60s)

What in the name of God is going on here ?!

 

Strange thing is that no matter what BIOS I try now, when I set back factory settings, CPU clock ratio gets set back to 42-40-40-40 correctly. However with the PCSpecialist default BIOS when I used restore factory settings, the CPU core ratio was set to 3x. Also, I did not see as bad throttle like this with the PCS bios.
It's version number was also interesting: 1.05.01 PCSN
Despite having the default Style Note logo. maybe it is a special modded BIOS ?

 

I have a P775 so I cannot really help you as much as I would like, but from my experience:

Setting the frequency to 4.2 in Bios works for me but increases temps with 15C.

Setting the frequency to 4.0 works and there is no drop at least evident.

What caused drops in performance for me:

- setting the power limits to anything, even increasing those will cause throttle (anything other than auto creates throttle)

- setting the voltage to static increases voltage a LOT and there is a lot of throttle - very evident

- setting the voltage to auto and using offset will cause no throttle - it works like a charm

- setting the A to anything other than auto causes throttles.

So the only things that will NOT cause throttle for me are to set voltage to adaptive and to set voltage to offset, and changing clock multipliers. This of course, on stock BIOS.

Anything else will cause throttle. Maybe this helps you out! On my P775 I am running only stock BIOS, stock vBIOS, with drivers mostly from Eurocom site and a few from other places.

 

Really looking forward to it!

Thank you for taking the time and the effort to do this!

 

@DRevan and @Georgel

have to buy or get a 3d mark firestrike advanced key then i will look into it too if my CPU throttles, using 1.05.05 Bios myself so stock from Tronic5 (XMG Barebones)

oh and @Georgel i'm using -130mV via Throttlestop for 4.2Ghz and barely touch 79 °C without delid using auto fans, if you have some questions about your temps and i can help you drop me a PM

 

Thanks!
For some reason I experience the biggest throttle when vertical sync is set to "FAST" in the NVCP.

 

Thank you!

I'm the type of crazy tinkerer that does play with his stuff and I noticed that if I try to decrease the voltage too much my windows either restarts or my Bios won't boot. I think my exact sample of 6700K is not best friends with undervolting and my work does require 100% data protection.

When I say 4.2, I mean 4.2 on all cores, for 4.2 on 2 cores and 4.0 on all cores I also get about 80-85 CPU temps depending on what bench I run but it passes Intel Burner test just fine.

I haven't noticed but the Nvidia driver might be broken. I had very bad problems with broken Nvidia drivers before, the only fix is installing another version. Something something .06 seems the most stable right now and doesn't seem to induce any problems at all.

 

Don't recommend it anyways, i didn't tried it on my new NB yet but on my previous P771DM and my Desktop PC i get immense stuttering when using it and not be in the 200-300 fps range.

Gesendet von meinem Nexus 6P mit Tapatalk

 

Maybe NVidia driver is broken but how can that affect the CPU speed ?

 

From what I've gathered is seems the EC (Embedded Controller) firmware is causing this... To prevent too much current going through the controller and/or pulled from the PSU at one time. I have zero throttle issues with running just the CPU overclocked to 4.7GHz, but add in GPU load and I get throttle even at stock speeds.

It's a limitation to help prevent the user from damaging the machine. Given a knowledgeable user, however, these "stock" limitations can be removed and much higher current driven without damage. That is what @Prema will be hopefully delivering to partners soon with unlocked BIOS & vBIOS!

 

It doesn't even come close to damaging the machine. It's just Clevo being ultra conservative about the limits to minimize their repair costs as well as the slightest risk of anything going wrong.

 

I'm not a knowledgeable user by far.... Should I keep on stock and settle for low performances?.............

 

Yes

 

That's cold man........

I'm trying to joke around, don't burn me!

I'm sure I can handle it... (..crying in the corner in fear...)..

.
.
.

By the way, how was I able to break CCC?

I don't have overclock fans in CCC... I needed those........

Any idea what did I do wrong or how do I fix this?

 

Is anyone actually able to set the DM3 LID LEDs to stay on instead of breathing?
Personally I would like them to be static…
What would you prefer?

 

I think I remember reading that the older .50 version doesn't have the OC fan profiles.... See older post here

 

Nope, just tried and it seems like it there's not a static color mode available on mine either that will stick... Would definitely like that option

 

I'd like if it could be static or off and have it be a choice by me. Pulsing *shrugs* not for me.

 

I generally have static color on all my devices, breathing can be distracting on any device.

Thank you, I had no idea about this!

I'll see to it.

I just tested a FireStrike with fans on max and 4.4 - 4.2 on CPU, fresh install nvidia driver, everything set to quality, no Vsync, no Gsync. Temps problems are there only if you think about them.

Though I gotta ask those who do tinker a lot - I've no idea if the numbers reported are good or bad, especially voltage and things that I'm still reading countless tutorials on. I'm on basic Bios, basic vBios, just basic with the poor try to change clock multiplier to 4.4 on 2 and 4.2 on 4 core from bios. That is it.

Please help on this : Is the voltage way too high?

I'm way over what is considered a safe voltage, and I am on stock everything....

 

Back at it, seems the overclock profile is back after an Nvidia driver reinstall.

@Phoenix - don't install SVP with Nvidia drivers.

It breaks the drivers entirely, I almost had to to a complete system reset to clean it...

 

well well well, would you look at that... today i took my sweet time and did a repaste with GC Extreme (replacing the short-lived kryonaut) and also replaced all the stock thermal pads with Arctic Cooling pads and Phobya Ultra. as per usual, lots of components werent really properly covered or making good contact with the heatsink and guess what? not only did my stock firestrike score increase, but the CPU clock dips I mentioned before are gone now as well under load, GCE provides about the same cooling performace as kryonaut before, with only slightly higher deltas (range of 1-2C).

however, the current throttling at 4.8Ghz and above is still present, i guess its just voltage related and theres nothing i can do about that. still, its good to see those cpu clock dips disappear during simultaneous cpu/gpu loads during gaming, cpu temps were about 5C lower on average, that was also nice to see.

now the only thing left is to check longterm performance of GCE (havent had any bad experience with it before, but now im just a bit sensitive to that after the kryonaut excursion ) and also see if i still get black screens on the gpu side. i got those before when i was playing some shadow warrior II, at stock clocks and voltages! figured it mustve been some overheating components that werent properly covered by the stock thermal pads....

tse tse tse, these days u gotta do eeeeverything yourself to even get stable STOCK performance

 

GCE stays put pretty long. At least when i had it on my GPU. But on the CPU both Kryonaut needed very careful flat placement. Once that was achieved, none them had any issues.

 

trust me, ive never had such a perfectly fitting heatsink like i have now:

Spoiler

that was my last paste job with kryonaut, see how thin and even the layer was? used the cross method on the cpu and pea on the GPU. right after repasting temps were absolutely perfect, but only 2-3 days later they would shoot up by 10-15C without any reason whatsoever

i noticed that the gpu corners werent always perfectly covered, this was due to the stock thermal pads on the vRAM being too thick (2mm instead of the correct 1mm). should be better now as well

 

That is some sexy contact.
But im not sure why there are these inconsistencies with Kryonaut.

 

ikr?! im completely baffled as well!

 

Finally made some progress which I did not even expect. Will make a Windows reinstall to tripple check that it is stable.

 

woohooo, @Prema to the rescue!

as it seems, my board power limits were still not raised high enough in the regular Prema MOD, so he raised those even higher and Bob´s your uncle: no more current throttling at 4.8 Ghz!

new personal CB15 record: 1046 points CB11.5 will be following soon

as for the GPU black screens, those were still happening when running FS Ultra stress test turns out, my stock voltage is probably too high, cuz ive got quite the insane ASIC value for a 980M at 78.0% @Prema again helped out there and provided me with an undervolted vBios. Now its not black screening anymore, but either Freezing, BSODing or switching off, so I guess Ill have to manually up the voltage again and find my sweet spot for stock clocks *lol*

so much work! but hey, lucky me that I love tinkering with my hardware and even more lucky to have @Prema to help out in such endeavors with his magic!

 

Clearly, you need mosfet modifications and stuff

 

yes! exactly! and im not even overclocking the gpu! its ridiculous...im just glad that i barely have time for gaming anymore. my main focus nowadays is on cpu, ram and storage for everyday stuff and work

Sent from my Huawei Mate 8 NXT-AL10

 

WHAT? I'd get that RMA'd instantly if I could lol

 

UPDATE: It is looking like the G.SKILL 16GB 300MHz RAM is OK and the issue is buggy BIOS/firmware. Just be aware that they probably won't work correctly until you have the right firmware to support them. More to follow.

---

Based on some behind the scenes testing, I would like to discourage anyone considering purchasing G.SKILL 3000 RAM in the 16GB per module denomination from doing so until further notice. We have multiple examples of P870DM2/3 owners (actually every person identified so far, so that is a 100% failure for those identified) that are finding it unstable. From what we can see the 8GB G.SKILL 3000 modules are fine. @Prema and @Johnksss@iBUYPOWER are having no issues with their 8GB modules. We do not know if it is the G.SKILL modules themselves have issues or a system compatibility issue with those 16GB sticks, but they fail Memtest86+ and TM5 testing with hundreds, sometimes even thousands, of errors.

Here are some benchmarks with Kingston HyperX Impact 4*4GB 2400 modules overclocked to 2800 and 3000.

2800:


3000: (still tinkering with timings at this memory clock speed... not 100% stable yet, but nice numbers)

 

For the record. Mine now work. Still trying to work out the details as to why they work.

And @TBoneSan seems rock solid at stock 2400 for now.

 

I tried using 41X on all cores and 43X on 2 cores - now I have max temps of 87C in stress tests.

Seems this is the max my 6700K can go.

If I bump it to 42 on all cores and 44 on two cores, I ged temps of 98C.

Tested cinebench differences, with this I get 870 points, which is satisfying.

 

that is very interesting news indeed, especially since i somehow have the feeling that the corsair 16GB sticks are affected as well. i havent seen anyone with those sticks that could run them at 3000 mhz without any errors showing up in memtest or tm5...

may i ask why youre so hellbend on keeping a higher clock on two cores? this makes finding a stable voltage much more demanding and ull have to increase offsets more to satisfy two different multis at the same time...

Sent from my Huawei Mate 8 NXT-AL10