NP9380-S and Intel 4800MQ - Extremely High Temps

I upgraded my NP9380-S to Windows 10 on release hoping it would solve another issue that I have. Once on 10 I noticed a slight increase in my overall system temps so I decided to replace the thermal paste on my CPU and GPUs. After doing so, my CPU temps rarely went above 88C under a full load with all the latest Windows 10 drivers. A few days ago I decided to do a clean install of 10 due to various other issues related to the upgrade that required a clean install to fix them. After doing so, I immediately noticed my CPU fan was constantly ramping up and down prior to any driver installs. I installed HWMonitor and noticed my average CPU package temp at idle was up from 48C to 61C. I then installed all the same drivers I had prior to the clean install - no change. I then decided to run a benchmark with XTU and prime 95, my CPU temps would jump to 80C the moment the test started and would hover around 94C, occasionally it would hit 96C and begin to throttle. Oddly, prior to the clean install I was having an issue with my CPU not down clocking from 3.49GHz and would stay in the 3.5-3.7 range at idle, and the temps would sit at around 48C with no load and 88C under load. Now I'm seeing my CPU go from .99GHz to max 3.59GHz and with much higher temps (under load). I've enabled and disabled speed step and no change in temps. This is all on a bare install of 10, and before that was an upgrade to 10 from a bare install of 8.1.

 

My 4800MQ was having the same problem with a clean install of Windows 10. It got much better after I flashed PREMA's mod though.

 

I'm using his 1.3.3 version, I'm wondering if there's something I need to change in the BIOS to make it more compatible with 10?

*Edit* I forgot to mention earlier, I had to take out my 2133MHz set of memory and put a 1600MHz set in because the CPU throttled the entire time under load.

 

I highly doubt that it is Windows 10 in particular that is causing the temperature issues. You probably want to undervolt your CPU and may want to underclock your CPU.

 

I can't see a reason as to why I would need to do that, I've tested all my hardware and physically inspected everything. I've been using that BIOS version for almost 8 months. The issue didn't start until I performed a clean install of 10. I was initially running an upgraded version of 10 on a clean install of 8.1. What would you say would be a possible reason I would need to down clock my CPU? I have a Windows 8.1 on a bootable flash drive and I don't have these temp or clock fluctuation issues while running it.

 

Because I literally checked a week ago, and my temps on Windows 8.1 versus Windows 10 were the same. I don't see why you wouldn't want to do that when you can reduce temps (which seems to be what you want), increase the life span of your chip and even system in general, and you don't lose anything from doing that if you get low enough temps from undervolting and not needing to underclock. Temps would be a VERY big reason for you to underclock your CPU if you are getting high temps and even undervolting doesn't help enough. I underclocked (and undervolted, of course) my CPU, I didn't notice any performance decrease in games, and my temps dropped by 15-20c on my CPU, my CPU won't even hit 70c now. Higher clock speeds = higher voltages = higher temps. If you want, you could obviously lap your heatsink if it's not getting proper contact, you could do some heat pipe modding, etc, but undervolting and underclocking are simple enough and can always be reverted if you decide that you don't want them. I mean, you could always just run your system at high temps but I wouldn't recommend that because it's going to just reduce the life of your system, depending on how long you want to keep it (I like to keep my laptops for years and years, even after I get a new one, that way I have a backup just in case).

 

What power setting were you on on each OS?

 

I'm definitely going to attempt to under clock to get my temps down for the mean time. But what I'm trying to figure out is what's causing the issue so that I can try and solve it. This is what I don't under stand. Under load CPU automatically down clocks, but at idle it sits in turbo boost. It down clocks even before hitting 90C. Another thing I just noticed is I no longer have any control over my BIOS settings with XTU or any other programs. For example, I disabled Processor VR efficiency mode in XTU but in the BIOS is shows enabled. With Intel Speed Step, it show enabled in XTU and disabled in the BIOS. I can change these settings and they don't apply to the BIOS. Here's a snapshot of the issue that's got me confused.

Idle


Load

I'm running in balanced, but I've tried all power modes and custom modes and nothing seems to change. Its like my OS doesn't have control over my BIOS.

 

What bios are you running?

 

I've been using it for quite some time but haven't had any issues up until I decided to do a clean install of 10.

 

Well, for why it automatically down clocks, that's just how Intel CPUs work since they generate a lot of heat, even with Prema BIOS there is still throttling which I believe even happens before 90c, at least I get TDP throttling when I have stock clocks stock voltage. If you are going to underclock, you should definitely look into undervolting. That's very weird that XTU doesn't let you control the bios, perhaps you could look into using ThrottleStop instead of editing things directly in the BIOS assuming your CPU is compatible with ThrottleStop. A lot of people don't like running extra software like ThrottleStop but I actually prefer it because then if an undervolt is unstable, you don't have to reset the CMOS battery etc you literally just change the value in software.

 

Its just odd that in the 2 years I've had this computer I've never seen my CPU down clock for any reason, and as I said before it only down clocks on Windows 10 and not when I run a different OS from a flash drive or ssd. And to undervolt I just adjust the core voltage offset? What would be a good amount to start with, -100mV?

 

Yeah, core voltage offset, and yeah -100mV is a pretty good start. Of course, there's no guarantee that -100Mv would be stable. I can run -100mV stable on my i7 4710MQ, but some people can only get like -60mV.

 

-100mv is probably going to crash. Try -80mv, most Haswell chips are stable at that voltage offset. The problem, however, is that undervolting really just raises the core clock because the TDP window gains extra headroom. For example, undervolting my 4940MX brings it from 3.4GHz under full load to 3.6/3.7Ghz (it fluctuates between them) with a -85mv undervolt. If you really want to drop the temperature, drop your TDP to 35W in XTU or go into the BIOS and turn off the turbo entirely. Some machines won't let you turn down the TDP, I don't know about yours.

It is possible that the thermal sensors are being read incorrectly by Windows as well. Check and see if Intel has released a chipset update for Windows 10.

 

Except you can use something like ThrottleStop, or hell even the good old BIOS to set your maximum clock rate. That way, you don't have to deal with something like wattage, you can just deal directly with voltage. Wattage is unpredictable for determining maximum clock rate, much better to just set it manually to what you want it to be.

 

ThrottleStop is a pain. XTU is the proper way to do it. If you tweak the wrong setting in TS, you have to CMOS reset to fix it. The TDP adjustment in XTU is the best way, especially if you set it to something low, start the stress test, and adjust it up until you find the perfect balance of performance and temps.

By the way @Troy_Carnahan here is Sager's driver page for your model. Install the Windows 10 drivers and see if that changes anything. Chipset and management engine.

http://www.sagernotebook.com/drivers.php?cat=583

 

I tried -100mV and it wouldn't POST, -80mV I got a BSOD for Machine_Check_Error, but it wasn't until about 5 minutes of idling after I ran the XTU bench test. I did notice however that lowering it to -80mV boosted the overall score almost 60 points. My core clock went from 3.15 under load to 3.59. It still sits at 3.49-3.69 at idle though. My temps still hit 96C before throttling. But with the throttling I saw in XTU it hit 56%, but the clocks never went under 3.25 during the test. With that being said, and in regards to your comment about the OS displaying incorrect temps; does the OS control the fan speed and throttling for the CPU? Or does the OS report back to the BIOS and in turn the CPU throttles if need be? Only reason I ask is because the CPU fan does not blow any hot air at all, even at max RPMs. But, it doesn't seem to ramp up to its max rpm until the throttling engages. I'm currently using the 10.1.2.10 chipset driver. I was initially using the Sager driver, but with that driver I had an issue with my USB 3.0 ports not initializing any 2.0 devices, the later driver solved that issue for me. The issue I'm having existed with the 10.1.1.9 driver as well.

 

Have you tried the TDP adjustment? Your board should support it. Locking the TDP is the easiest way to drop temps.

 

That's the next thing I'm going to try when I get off work. But in doing that will I see a noticeable performance hit? I'm running Sli 780Ms with a 120hz panel and I plan to replace my 840 Pro SSD with 2x mSATA 850 evos. I want to avoid bottlenecking as much as possible.

 

Basically, fire up the stress test in XTU, set it for 10 minutes, then adjust the TDP down until you get to a point that you're happy with. You definitely need to check your mounting and such though. It is really easy to mess up the heatsink on these machines due to how its designed.

 

Uhh, I think you have that REALLY backwards. In ThrottleStop, you can ALWAYS just manually boot into Windows safe mode and remove your ini config file and thus you will have clean settings on reboot. With XTU, it saves settings to the BIOS, and we all know when BIOS settings are messed up, you need to CMOS reset. TDP adjustment doesn't allow for easy selection of a maximum clock speed, unlike manually setting a clock speed. ThrottleStop is great, and perfectly safe.

 

No, I don't have anything backwards. There are indeed settings that stick, like for example when you tweak the power that goes to the iGPU. XTU allows direct adjustment of multipliers as well and is the Intel official solution. Having used both tools extensively, XTU is the better option. Its definitely more user friendly on top of that.

 

How exactly does ThrottleStop save settings? XTU DIRECTLY saves your settings to your BIOS. I know XTU is the official solution, but official isn't always necessarily better - Prema has shown this time and time again with his amazing mods. I would rather not modify settings in my BIOS that may require a CMOS reset when I can just use ThrottleStop and when I use terribly unstable settings I can just boot into safemode and delete the config file and safely start over again.

I think @unclewebb needs to come in here and chime in on this.

 

Well I'm telling you that I deleted the INI file, deleted the entire program, and the power balance settings I modified showed up when I reinstalled the program and opened it. Reflashing the sBIOS, resetting to defaults, and opening the program again, everything was back to normal. I haven't used it since. It wasn't a huge deal because it didn't actually affect anything other than causing problems with XTU but I also haven't had XTU make my system unable to boot either. I'm sure if I messed with memory timings, I could easily achieve that scenario and with how difficult it is to get to my CMOS battery in my laptop, I opt not to even try.

 

I thought XTU had a built in fail safe that if you ever modify anything that causes severe instability to the point it won't POST, it configured the BIOS to reset any modified settings upon the 2nd attempt to POST? Besides, XTU and any other 3rd party BIOS modification utilities are irrelevant to me due to the fact they don't have any control of my BIOS. I can see all my BIOS settings, but if I modify and save something it reverts back to the original value both live and upon a reboot. I can even make a change that requires a reboot, go into the BIOS and find the setting unchanged. I have to manually change them in the BIOS if I want them to stick. At first I thought this was due to a bad clean install of 10, so I performed another install and the same issue. Again, this is not an issue if I plug in a drive that has Windows 7 or Windows 8.1 on it.

 

XTU reverts back to default settings after a crash and auto restart. It's part of the undervolting process I'm going through at the moment.

 

I know XTU watches for POST failures on the CPU (this has been around at least since my Q6600 Kentsfield chip) but if you screw up the RAM timings, you always had to pull the CMOS battery. I haven't messed with memory timings since my Q6600 because the Bad Axe 2 was a real pain to reset when you screwed up the timings (there was a jumper that was located in the worst possible location - surrounded by all the ribbon cables and towards the edge of the board which you had to pull off of two of the three pins and move to the other two then back again) so I suppose they could have worked that out.

 

What do you think would be the reason I can't control my BIOS settings from within Windows 10, but I can on any other OS?

 

Did you install the Windows 10 version of the Management Engine driver from the Sager site? It won't save the settings without the correct ME Interface. If that doesn't fix it, just save the profile and apply it when you boot the computer. You can also try upgrading to the latest Clevo BIOS for your machine to get the latest ME then flash back to Prema. The ME doesn't downgrade.

 

I'm using the one from Sager's site but Windows updates automatically installed a different version after I rebooted. I haven't had any experience down grading a BIOS version, should I watch for anything? And where would I get the latest BIOS for this model (NP9380-S)? All I can find is the version mine was shipped with which is 1.03.02 (December 2013) and that's the same one posted on this forum. Sager doesn't have any posted on their site. When I try looking for one, the Prema versions show up everywhere haha. Just out of curiosity, what's the reason for flashing to a Sager BIOS versus Prema just to get ME, then flashing back?

 

Prema has MEs also. https://biosmods.wordpress.com/me/

 

Get Shutup 10 and disable automatic driver updates - http://www.oo-software.com/en/shutup10

Then install the Sager ME again.

As for the BIOS, I went and looked and he doesn't have a Windows 10 ME firmware version posted under the ME Firmware section and your BIOS version is the same as the latest he has on his site so I don't know. Maybe @Prema can weigh in on it.

Worst case scenario, you can set your settings in 8.1 or 7 - they'll stick in 10 unless you overwrite them.

 

Wow awesome thank you, I have the version previous I believe but I'll have to double check in about an hour.

I didn't realize there were different firmware versions for ME. I'm glad you pointed that out to me, btw thanks for all you help thus far and for everyone eleses as well. I'm going to wait and see if Prema chimes in before I go through the process of flashing my BIOS multiple times.

 

I updated the ME firmware and I'm about to start under volting and adjusting the TDP. In the mean time, what could be causing my northbridge to be so hot? This is while running the XTU bench test

 

What do you mean he doesn't have a Windows 10 ME firmware version? You literally go to his website, go to ME Firmware, download the right one for your Clevo, extract it and run the exe. Works perfectly fine on Windows 10, I did it yesterday. Works perfectly fine with his Prema BIOS and vBIOS mods that I am using.

@OP: Your temps are high because you're running your CPU at 1.2v. I have my i7 4710MQ @ 2.5 GHz and literally 0.75v. 0.75v versus 1.2v is literally like a 20-30c drop. My CPU doesn't even hit 70c. Voltage makes a huge difference for temperatures and unless you are undervolting and maybe underclocking you are not likely to see your temps decrease if you have done everything else except for those two things.

 

Wow why is my voltage so high compared to yours? My system isn't stable if I try to do anything over -60mV? Are you using turbo boost?

 

Because you are running at stock clocks. I have my CPU heavily underclocked - 2.5 GHz which turns turbo boost off. Also, the lower your clock speed, the more you can decrease the voltage. With that said, I am running a huge GPU undervolt too which makes me less able to decrease the voltage on my CPU, so it's a bit of a trade off. I'm very happy that I get less than 70c on both CPU + GPUs which means my system should last for years to come.

 

I've decided to disable turbo boost entirely now due to the excessive heat issue I'm having in Windows 10. With that said, since I just updated to the latest Prema ME firmware I lost the ability to adjust my reference clock. I wanted to try and OC the base clock to the max 2.9 and leave it at that, but since I can't I would like to lower my max TDP and voltage. When I disabled Speed Step I lost the ability to adjust my lower and upper TDPs. Is it possible to lower it from the 47? It's only using 30w at 2.7GHz and .88v rather than the 1.2 with Steed Step on.

 

The version numbers he has up there aren't for Windows 10 but if they're working, that's great.

You do realize that his P370SM3 uses the same heatsinks that my P377SM-A does and my blistering hot 4940MX doesn't even hit his temperatures running at almost 1.3v right? I don't know why you're so insistent that the only acceptable fix is to undervolt and underclock his machine. Its obviously one of four things - the heatsink isn't mounted properly (most likely), the thermal paste isn't acceptable, there is insufficient airflow around the machine, or the thermal sensors are jacked up in 10.

Personally I'd be rather annoyed not having turbo boost but then again my chip cost me a grand on the build.

There's no need to adjust the TDP if you turned off turbo altogether, it won't ever come close to that TDP.

 

I am very annoyed actually. I can run my system cool on my 8.1 drive, but I really enjoy the performance boost in games I've seen on 10 so far. In an attempt to compensate I now have my memory clocked at 2400MHz from 1600 and the XTU benchmark is actually about 10 points better than what I score in Windows 8.1 at 1866MHz with Turbo Boost enabled. Probably because my temps stay under 76C at full burn without turbo boost (90C in 8.1 w/ turbo boost). I did take a look at the mount and swapped out to a different thermal paste, barely a change in temps there. I'm starting to get the feeling its an issue with how Windows 10 is reporting temps on my setup. I suppose since my CPU is now running far cooler and at less power consumption I can try overclocking my 780s, most everything I play is GPU taxing anyways. I would hope 2.7GHz wouldn't bottleneck my GPUs if I overclock them.

 

Actually, you're missing another posibility - it's possible that his heatsinks are literally mangled/bent and need to be lapped which is quite common with Clevo heatsinks of all types. Thermal sensors are fine for me in 10, and for lots of people running 10. Not all heatsinks are built equally, and if you take the time to lap a heatsink properly when it really needs to be lapped, you can drop 10-20c easily.

These chips just run hot, and decreasing the voltage and clock rate (and thus, voltage even more) will reduce the temperatures a lot.

@OP: I have seen tons of benchmarks and CPUs really don't affect frame rate TOO much (10-20 fps stock vs an overclock on most Intel CPUs in some games, 3-5 FPS in others), and you would likely gain that performance back from overclocking your GPU as long as your temps aren't too high on your GPU you should be fine.

 

It just doesn't make any sense that 10 runs the CPU hotter than any other OS... I'm assuming you've done direct comparison with forced max fans (FN+1)? In all the time I ran 10 on my 9377 from Insider to the week or two on RTM, I actually noticed lower temps due to better resource management in the new NT kernel.

You don't have to lecture me about Clevo heatsinks. I've had my fair share of fighting with them - Foxconn can't make a flat heatsink to save their life. The biggest issue is that temperatures are reporting so much higher in 10 than in 8.1 - how do you explain that one? I'm wondering if its an EC issue which is why I asked if OP has forced max fans and compared temps. Its very well possible that Windows 10 isn't communicating with the EC properly resulting in the fans not ramping up when they should. I had a similar issue with Windows 8.1 on legacy mode when I didn't have the proper Hotkey version for my EC version. I fixed that by grabbing the latest Hotkey from Clevo.

 

This model unfortunately does not have that FN key, just the standard keys seen with any 17" laptop. As far as I know, there's no way for me to manually adjust any of my fans.

 

Your keyboard doesn't have an FN key? You hold down FN like you would to adjust your volume or screen brightness and hit the number 1 on the number row (not the keypad). It should trigger the fans to go full speed.

 

On my keyboard setup it turns my touch pad on/off. I don't see anything that resembles a fan on any of the FN icons.

 

Hold down the FN key and press the number 1

 

This is what my key does:

 

Because you're hitting FN+F1. Hit FN+1 - as in 1 2 3 right under the F keys. The number keys.

 

Derp! My bad, its still early haha. Yea that definitely works, awesome!

Is there a key for the GPU fans too?

 

That forces your fans to their maximum (Clevo EC usually won't go over 80%)... Try setting your CPU back to stock and run XTU stress test in 8.1 and 10 for 5 minutes each and see if your temps are more in line with each other. Obviously let the chip cool down between tests. Or even just see if you thermal throttle in 10 anymore.