*** Official Clevo P870DM2/P870DM3 (Sager NP9873/NP9872) Owner's Lounge! - The Phoenix 2 is here! **

Could someone please try to help me with solving following issues? I'm trying to fix them for months now and I'm stiill unable to find the answer.

1. Fan speed will stop increasing from about 65-70°C (Automatic fan profile) leading to overheating and throttling. Another problem probably related to this issue is that neither HWinfo64 or any other software will show fan speeds (this data just seems nonexistent), therefore, the temperature range 65-70°C is just my estimation as I could not hear any fan speed change after reaching higher temperatures. I asked Prema and he confirmed that fan curve in his BIOS/EC should go all the way up to 100% fan speed. I attached a screenshot from XTU where I ran benchmark (CPU temperature and fan speed were rising as I could hear the fan noise) but you can see that XTU did not receive any fan speed data.
I replaced stock T-shaped GPU heatsink with vapor chamber heatsink + replaced stock thermal pads with Alphacool 14W/mK pads what significantly reduced temperatures of the GPU. However, at full load, the GPU temperature still exceeds the point where fan speed stops increasing and the temperature is slowly but constantly rising (even though that probably only slight fan speed increase would stop any further temperature rise). Also what I observed is that if the temperature reaches cca 77-78°C and stays over this limit for several minutes, fans will suddenly speed up to a very high level (very loud noise, but still not as loud as max fan speed with fn+1) and stay at this speed until temperature decreases below 60°C.
I already had this issue while running stock BIOS, but I thought it might get solved once Prema BIOS will be flashed into my system. I tried uninstalling both XTU and Clevo Control Center followed by NVRAM reset and disabling XTU fan table support in BIOS, but the behavior remained the same even without any of these installed.

Spoiler

2. The screenshot above also shows another issue I have. With replacement of GPU heatsink, I also replaced original i7 6700K with new 7700K. Right after replacement, I performed NVRAM reset (got Prema mod introduction message after powering up) and went into BIOS to check if everything is ok. At first it seemed good, but when I went into first overclocking menu (that should allow me to adjust sepatare clockspeeds for 1-4 core operation), the processor settings were completely missing, only memory could be adjusted. The second overclocking menu has all setting for CPU as expected and they work fine. After that, I went into Windows to check if I can adjust multipliers at least within Intel XTU but no luck there either. I tried performing one more NVRAM reset, but nothing changed.

3. Current GPU driver installed in my system (378.49 on Windows 7 x64) is downloaded directly from nVidia's website, but I noticed strange behavior in GPU clocks probably caused by some sort of incompatibility. With stock driver from Eurocom (very old one), I was mostly getting GPU frequency varying according to load while keeping GPU usage within 70-80% range. Another time while playing the same game without changing any settings, the frequency is still varying but the usage is kept at around 50-60% what yields higher temperatures even though all settings are the same (G-SYNC with V-SYNC is enabled in nVidia control panel). While playing another game, the frequency is more or less fixed (or varies just slightly) to relatively high value compared to load - screenshots below show this behavior in Crysis 3.
So far, I could not figure out what is causing these differences, but I would like my GPU to perform according to first scenario (clocks adjusted to maintain 70-80% usage). All the time, my settings were set to High performance power plan in Windows control panel settings, Performance in Clevo Control Center and Optimal power in nVidia control panel (power management).
Therefore, does anyone have some modified GPU driver that is fully compatible with this system?

Spoiler

Note: Screenshots above were taken just shortly after launching the game, therefore, temperatures are fairly low compared to what I'm usually getting while gaming.

 

Hi. I posted some answers in bold blue font in the quote above. Please expand to view them. I hope you find the information helpful.

 

Thank you for your answer and explanation, at least I now understand why XTU doesn't display fan speed. However, HWinfo was already set to EC support and it still has absolutely no entry about EC or fan speeds, see screenshots below:

Spoiler

Also, the issue I'm trying to describe here is not about fan curve not being agressive enough, but about part of the curve being completely cut out. I tried to draw approximate fan curve from noise levels I could hear at various temperature levels:



Up to 70°C, the fan behavior is completely appropriate for my needs, but after that point it completely stops increasing speed just as if part of the fan curve is completely missing and then just suddenly boosts up by a big margin and stays at that high speed even when not needed anymore until the temperature drops below 60°C. There is something really wrong either with reading of the fan table or fan table itself not being set up properly in my system. OC fan profile is way too agressive for just single GTX1080 with vapor chamber heatsink, I just need a slight speed increase above 70°C and the peak temperature would not even exceed 75°C at max load, plus it is also a bit too noisy for my taste.

I took some photos of the BIOS, please check them below:

Spoiler

It is true that Optimal Power reduces power and I'm using it to keep the GPU temperatures lower while playing not very demanding games. As I mentioned in my OP, I'm using G-SYNC along with V-SYNC (framerate capped at 60Hz), therefore, I don't need more performance as it is there if I need it. Instead, I'd like to limit the power when it's not needed to keep everything nice and cool. I know that you probably have opposite opinion about this, I just want to use my system for gaming, but I don't require benchmarking performance. As long as the game stays at solid 60fps, I'm happy. The only problem here is that the stock nVidia driver is apparently not completely compatible and Optimal Power does not work properly. If you know about any modified driver, I'd like to try that instead.

 

Glad to help.

First, reset NVRAM. Pull the AC power and CMOS battery. Then put them both back in place.

This (below) could be your problem right here. First value (Core Max OC Ratio) is 0. This means your CPU is blocked from exceeding stock core ratios. Set it to 80, F4 to save and exit the BIOS, then go back and see if the first overclocking menu is populated. But, definitely clear NVRAM first, then set this to 80, then save and exit the BIOS, go back in and see if the CPU overclocking options showing multipliers and power limits is available. Setting it to 80 is merely a maximum limit. It must be set to AT LEAST the clock ratio you want to see when all 4 cores are loaded. Once you have the missing menus available, set both PL1 and PL2 to 160000.

If the CPU overclock options are still not available after taking the above steps, then your @Prema BIOS may need to be re-flashed.



-130 offset to stock voltage probably isn't going to provide you with any overclocking headroom and may even stunt performance with stock clocks.

HWiNFO64 settings may still need to be adjusted to see fan speeds in the Sensor Panel. You probably still have a box that it unchecked that needs to be checked in the settings. But, if you find it and enable the fan speed sensor you may see unwanted behavior. If you can get the fan behavior you need, knowing the fan speeds won't matter.

If you feel keeping the NVIDIA Control Panel set to Optimal performance is the best approach, I think it is going to impair your gaming experience. Less demanding games will likely suffer more from using Optimal mode than more demanding games do. They are not taxing the GPU enough to make it run correctly. Try using Adaptive if High Performance is too aggressive. Maybe that will help. You may also find a different driver version works better. No matter what version you land on, I would recommend using a @j95 desktop driver mod and nixing all of the NVIDIA bloatware that adds no value.

 

I did everything as you said but unfortunately it looks that I will need to get the BIOS re-flashed. I remember that with 6700K the Core Max OC Ratio was always automatically adjusted to match maximal value set in multiplier settings (the ones I'm missing), but now no matter what value I put there, nothing changes.

That is just temporary value until I'll be finally able to play with OC-ing, so don't worry

I completely agree. The only reason why I wanted to see those fan speeds was to show you exactly how the fan curve is being cut off. Other than that, I don't care about monitoring these values. I'm still trying to find out what is the cause of cutting off and that hysteresis-like behavior. Do you think it could be hardware related issue or just something wrong with software installed? Just out of curiosity, could you please try to select Automatic fan profile and observe if the fan speed will continue increasing normally after reaching over 70°C?

This could be very useful, thanks. I found @j95's 378.72 driver here: https://www.techinferno.com/index.p...ort-modded-inf/&do=findComment&comment=156716 , but unfortunately I don't have permission to download his modified file. Is there any chance you or @j95 could upload it somewhere else?
Also, are there any news regarding @Prema vBIOS for GTX 1080?

 

It's deliberate, you can argue it extends the life of the hardware and avoids pulsing fans by getting the temps down to a normal range as fast as possible.

 

Sounds really messed up to me, and unless I use the OC profile I see similar behavior. I agree the fan tables are messed up with respect to ramping up with temps. The fans should not be waiting until 80°C to boost up to a useful level. The cool-down part seems good though. Have you tried creating a custom fan profile in CCC yet? It kind of sucks since you cannot create a profile that allows the fans to go past 80%. Not sure why Clevo did that.

 

With my old stock bios/EC my fans would actually go max after a time of being warm where as they stay in the non maxed state now.

 

Ok, so, I need to ask some things from some of you all here.

• Anyone in windows 10 (or 8.1) in UEFI mode (I don't have CSM on, though I don't think it makes a difference I need to ask) with a USB headset, can you please set your display to timeout for me after 1 minute and play a youtube video in the background, wait until it the screen turns off, then wake it again? Tell me if the sound stutters for a bit. This does NOT show up in any DPC latency test, so it isn't that kind of issue. I need to know if the problem is "my" computer or not. It's as if the USB controllers just get high latency for about half a second. It isn't enough to impair mouse movement or cause any lasting issues that I've seen, but the problem has been around over multiple OS reinstalls. If there is some BIOS setting to fix this it'd be great, but if so I don't know why it's not set as the system default.
• Also, I need someone who is likewise on Windows 10 (I do not remember the problem being on 8.1, but I could be wrong and I REALLY don't want to reinstall an OS to check this one thing) who has audio drivers 6.0.1.7824 (the latest available as far as I know from Clevo etc) to test this for me. When I open "recording devices" (right click on the sound icon in the system tray and select it), my laptop makes a popping noise. It's somewhat loud, and it doesn't happen in quick succession if I were to keep opening it over and over (only once every few minutes). Here is a video; you can't see what I'm doing but you can certainly hear the pop around the middle of the video.

Any feedback is greatly appreciated, please and thank you <3

 

@ R1sky999

The fans are slow to react plain an simple. My fans seem to want to let the CPU temps sit in the high 80s early 90s before they think about spinning up enough to save themselves. It's pretty annoying. Try using a custom fan profile and see if that works better for you.

@D2 Ultima
Regarding the pop noise which I use to also get.
Have you disabled power management? That fixed it for me

 

I had not, I just did. Turning it off popped it, so it seems that probably worked? Interesting, and thank you for it. I'll let you know if it remains fixed =D

Edit: seems to be fixed!

 

Thank you very much for testing it, @Mr. Fox ! I started to become worried that my hardware is broken as I have this issue since I received the laptop; it was exactly the same with stock Clevo BIOS, as well. Do you think that it would be somehow possible to fix the Automatic fan profile sometime in the near future? Maybe @Prema could use his magic and find a fix somewhere within EC.

I'm actually using Custom Fan setting when playing demanding games as it is the only way to keep temperatures where I want them to be while keeping the noise level bearable. I don't need to go anywhere above 80% (although I also wonder what kind of reason made them limit the speed option to 80%), the vapor chamber heatsink with 14W/mK pads make wonders to GPU temperatures. The screenshot below shows my stable temperatures after 30 minutes of constant 100% (well, 99%) GPU load with Custom Fan profile set to just 50%. I used 8x MSAA to make sure the GPU is taxed enough to maintain 100% load for the whole time.

Spoiler

CPU temperature is a bit high as about 1/3 of CLU in syringe was solidified when I was delidding. I applied CLU inside, but didn't have enough left to make proper thermal connection between IHS and heatsink. I already ordered another package of CLU, so I'm planning to fix it sometime during weekend.
However, the point I wanted to make here was to show you that I truly need just very minor fan speed increase after reaching 70°C with Automatic fan profile to keep the GPU nice and cool. My biggest concern is that everytime I want to play, I have to go to CCC and manually switch fan speed from Automatic to Custom (Custom is not very convenient for other regular usage) with a laptop costing thousands of pounds. That is just unacceptable.

Although I agree with you in terms of longevity, it still won't justify broken fan profile. In practice, after fans finally speed up, they won't get lower while playing the game as the temperature usually stays at around 65°C (fan speed starts decreasing not sooner than below 60°C). That means, the profile then works just like Custom profile set to fixed value, plus as you can see in the screenshot above, all I need is getting to just around 50% speed to provide proper cooling, not being stuck at 80% with 'Automatic' profile. In that case, there is nothing automatic about that profile.

 

Don't use your CLU yet. Save it until you have a 30mm x 30mm x 0.5mm copper shim to place between the CPU and heat sink. You'll be glad you did. It should improve your temps significantly.

http://www.ebay.com/itm/231683140824?

 

Lol.. I have these on order too. The seller must love Clevo not caring to address this issue any time soon
Did you take forever to get them? I find stuff from China with free shipping to be slow at best (even being sent to Japan).

 

Lool, perhaps Clevo has invested money in Copper Shim production. More money in this than use more money on HS QC

 

How is the shim attached to heatsink and spreader optimally? Also with CLU?

Gesendet von meinem SM-G928F mit Tapatalk

 

The 1080TI might encourage me to build a new desktop and sell the laptop since I'm not out of town much lately. Also, interested to see how the new ryzen chips perform and how well they overclock on water.

https://www.google.com/amp/www.thev.../geforce-gtx-1080-ti-fastest-nvidia-card-ever

 

CPU Die (Delidded) <--> Coollabratory Liquid Ultra (on Die and 'bottom' of IHS)--> IHS <--> Coollabratory Liquid Ultra (on 'top' of IHS and 'bottom' of shim) <--> shim <--> Coollabratory Liquid Ultra (on 'top' of shim and laptop heat sink) <--> Laptop heat sink

 

Lololol. Me as well

 

Ok, W10 has been... mostly tamed.

Now it won't update my drivers on me
Now it won't update on my whenever it wants
Now it won't change my mic volume when I speak loudly DESPITE DISABLING ACCESS TO ALL PROGRAMS TO DO THIS EFFECT
Now it won't do the telemetry crap
Now it won't pop my speakers whenever I open mic settings
Now it downclocks my CPU (but only on balanced... and on high performance it never does I do not understand this ********* OS)
Now it doesn't lag stuff when I open programs for the first time
Now I have a working start menu

I gotta re-pad this machine, install another drive, get everything up and running properly, back up the system again, try to fix the heatsink, use CLU with a shim for my chip, etc. Joy.

Also I am sure my GPU has a low-pitched coil whine. I hope thermal pad replacement will help that. It doesn't even need me to be in a game... just using my browser is enough to trigger it. AND I can hear it through max fans.

 

One of my fans does that low pitched noise. I just taped around it, making sure nothing in or around it was vibrating. And that fixed it for me.

 

Tell me more later? It comes and goes but happens at both stock and max fans. If a lil tape helps then I can try. But I will need to import electrical tape rated over 80c lel

Sent from my OnePlus 1 using a coconut

 

LOL i just used normal packing tape. And the fans housing doesn't get as hot.

Basically the areas i circled here -

Spoiler

Before you panic, that is the old heatsink, and HID replaced it for a new one almost overnight

Ill shoot you a message over Steam tomorrow morning.

 

What is wrong with the old heatsink? Is there a new revision?

Gesendet von meinem SM-G928F mit Tapatalk

 

Take a look one more time

 

@R1sky999 I have the exact same problem with GPU heating and fans speed.
I tried CCC but none of the modes suit me so I got rid of it. (my feedback on the fans curves here : http://forum.notebookreview.com/thr...the-clevo-laptops.800066/page-5#post-10441819)

Unfortunately I didn't find an ideal solution yet, so now it's all about not reaching 84 C that triggers the 3400 RPM fans, meaning to stay in the 2300 RPM level (65 C - 83 C) or less. Which is not hard, using a cooling pad and sometimes limiting FPS in some games just under what the GPU can deliver. It mainly depends on what amount of CPU the game uses, because GPU stressing only doesn't make it reach 84 C.

@nevermore93 what was your method to undervolt your GPUs ? I feel that I wouldn't need much to never go beyond 83 C.

 

Hello gentleman, someone tried already with the shim on cpu? The results? If Someone can post a video i will appreciate it.
Thanks

 

With what CPU?
Use the search link at the top to find 'shim' posts by @Mr. Fox. IIRC, thermal results were pretty noticeable on the i7-7700K.

 

Here is the first report of my discovery: http://forum.notebookreview.com/threads/clevo-overclockers-lounge.788975/page-958#post-10438770

A number of other folks have experience similar improvements using shims, including @TBoneSan, @bloodhawk, @Q937 and @Coolane and perhaps others. HIDevolution and Obsidian PC are already using 30 x 30 x 0.5mm shims to improve CPU temps on systems they are shipping to their customers and Eurocom has begun testing the same to validate the benefit.

Hopefully, the mountain of evidence will motivate Clevo to stop pretending that everything is peachy and they will step up to the plate to fix their sloppy-fitting heat sink mess. Sloppy heat sink fit is an area of opportunity for product engineering that we have been struggling with on Clevos for several generations now, and there's no good excuse for it. We shouldn't have to do this. It's good there is a workaround, but it would be better if the heat sinks actually fit correctly. I'm also pretty sure temps would improve over what they are using a shim to plug the gap. There has to be some degree of lost efficiency using a shim.

Other examples:
http://forum.notebookreview.com/thr...oenix-2-is-here.794530/page-932#post-10470420

http://forum.notebookreview.com/threads/clevo-overclockers-lounge.788975/page-1041#post-10470305

@Donald@HIDevolution @Eurocom Support @John@OBSIDIAN-PC

 

We´re doing the same, in some cases we even solder the copper shim to the stock cooler.
You would be amazed on the amount of work that the P870KM1 needs for good contact.

 

Clevo have to fix this issue ASAP, this is incredible.. i don't think is hard to figured out..

 

I updated my previous post and added Obsidian to the list of vendors taking active steps to mitigate the disaster. Thanks for reminding me you are doing it also. I will not forget this again. Sadly, I probably would not be surprised at how much work it takes. My P570WM had a MASSIVE air gap, so did the original heat sink on the P750ZM. Both had to be replaced at my expense and then modded to improve temps enough for them to be usable.

Step #1 is just getting them to admit there is a problem. Apparently, they think everything is just ducky. It is pretty sad they have so little awareness of the engineering defect their products are shipping with. Maybe it's really a lack of concern rather than awareness.

 

How do you keep the heat pipes from coming loose during the soldering? Are you having any issues with that? I really want to solder mine, but I don't want to destroy the only parts I have to work with by accident. Are you using a butane torch, hot air soldering or some other method?

 

Will ask my metal worker to PM you, that´s one of those things that i actually never did myself, anything that requires "solder" is not my department, there is a good reason why my childhood friends call me "Johnny Flame", and it involved a movie theater and a fire....

 

LOL... thanks. I bet there's quite a good story behind that nickname.

I will PM my email to you in case that would be easier for him. I appreciate it.

 

Loop me in if possible folks!

 

That is totally normal behavior for Balanced power profile. Windows 7/8/10 all do that, but it is much worse with W10 since it is a lame cross-platform OS designed to accommodate trashbooks and tablet filth. One of the reasons I hate Balanced profile, as the name implies, it is not focused on the best performance. But, you don't have to use the defaults for any power profiles. I only use the High Performance power plan, and even then not with Windows defaults. I disable every trace of power management stuff and set minimums on the processor to 100%. For ThrottleStop to function correctly, you need to do that anyway. Otherwise, Windows screws up ThrottleStop.

Fix it with this tweak...


And, then use ThrottleStop and uncheck C1E. After that you should rarely ever see the CPU downclock. It should deliver full performance 24/7, regardless of whether full performance is necessary or not.

 

Oh, I want it to downclock. I just want it to do so on high performance. The power plans are one of the first things I set when I install windows; both high performance and balanced are exactly the same in advanced settings (my tweaks). If I use high perf my CPU never goes below boost (4.4) for more than half second. If I use balanced it almost never boosts! It's weird. Maybe I need CCC installed? No idea.

Sent from my OnePlus 1 using a coconut

 

Oh, I see. I never want my CPU to downclock for any reason. I hate that and always have. If you set the minimum to 100% it should not be able to downclock to save power and that behavior would be by design.

I disable every trace of power management stuff in the BIOS, Device Manager and Windows power profiles and set minimums on the processor to 100%. For ThrottleStop to function correctly, you need to do that anyway. Otherwise, Windows screws up ThrottleStop.

 

Yes, I know. But with my heat a clocked up processor doing little is a waste of heat. Much better to let it downclock when I don't need it. On my P370SM3 I was able to set it up such that my CPU downclocked not in use but when stresses it clocked up to max turbo without even aid from throttlestop. This machine I haven't been able to do this with yet. I'll use CCC later on and check if this helps things along.

@John@OBSIDIAN-PC the CCC from your program is pre-set to not tie into XTU correct?

Sent from my OnePlus 1 using a coconut

 

Yes it is! No XTU, no OC bit.
By the way, i would say that CPU clock is irrelevant, what you ALWAYS should be looking is at the nifty Package Power box in TS, that´s what matters, sometimes a lower clock will be at 15W just like a OC [email protected] is. When i want less heat or less drain i always look for power limiting and i usually ignore clocks.

 

You could also use ThrottleStop for that. Make your battery profile way stupid low clock speeds that are good enough for web browsing and email and it should cool off in your Tobago heat. I use ThrottleStop like a manual transmission and keep it in whatever "gear" I want it to function in, then use hotkey combos to "change gears" when I want to.

That works really well to choke the power down.

 

If I cannot get things working like on the last system (which even kept OC speeds if needed) then I will probably use some TS magic. Easier than swapping power plans all the time

Sent from my OnePlus 1 using a coconut

 

Oh yeah, I know. But since I'm checking the system at stock for now there's no voltage changing and such =D. A CPU at idle at 4.4GHz with 1.28-1.388v is going to use more power than a 2GHz 0.6v chip. Even if the intended load is the same.

And thanks for the clarification =D. I can use the BIOS and Throttlestop for OCing so that's fine.

 

Sadly, Clevo continues to keep their head buried in the sand. Upon inquiry whether Clevo was expending any effort to correct this issue...the simple answer was...No.

We are therefore going to have to continue to develop and implement our own methods of dealing with this issue.
.

 

Well, that's both sad and stupid. Maybe we should not expend any effort to buy Clevo products in the future. If the 17-inch SLI version of the Tornado turns out to be an awesome substitute for the P870 product line, maybe we won't need Clevo any more. All is fair in love and war, right? As it stands now, the primary reason Clevo monsterbooks are the best option is they are not emasculated BGA filth. Their lack of regard for correcting their engineering mistakes is disturbing. I wonder if Frank Azor is moonlighting as a member of their EMT (Executive Management Team) or something along those lines. Pretending the situation is good when it really sucks is an Alienware move. The sloppy fit must be Clevo's answer to the Alienware tripod CPU heat sink engineering defect that Brother @Papusan mentioned again yesterday.

 

I think I still prefer to use a copper chim than Clevo start using the modern TRIPOD design anyway. This below is from latest AcerTurd Predator 17. Mr. Azor has followers, as it might look like

 

Hi guys, I was wondering how much battery life this Clevo would clock with the I7 7700K, I'm pretty close to choosing this over the P650HS-G with the i7 7820HK, just wanna make sure that I don't have too little battery life. I'm gonna be buying from HIDEvolution, which will already have it undervolted, overcloked and delidded, in case that makes a difference.

Thanks for the help

 

Oops...originally read your post incorrectly.

Give me a call to discuss and I will be happy to help you with your choices...see my phone number in my signature below.
.

 

Yeah, I know......I mentioned the P650HS-G as another choice I have.