*** MSI 16L13 (Eurocom Tornado F5)/EVOC 16L-G-1080 15.6" Owner's Lounge ***

Maybe a little, but still better than nothing. The foam is dense (as in not very compressible) but very porous. And, the heat sink has fins on it. I could cut it down so it is only touching the center of the heat sink, maybe an inch long block of foam, but it seems to be working fine. I checked the temps of the aluminum heat sink once with my IR thermometer while running Cinebench a few times and the aluminum was around 65°C.

It sticks without the foam to hold it, but it has no pressure and I think it might fall off in time due to gravity with nothing to hold it.

 

It sounds like your VRM/Mossfet solution is working just fine then. In that case you seem more interested in using the spare heatpipe to cool the GPU further. Ah, 1.3cm deep fins you're talking about soldering to the copper plate that you'd attach to the spare heat pipe which will remain connected to the GPU - well those are are pretty substantial cooling fins you're talking if you've got quite a few of them on there! Things is though, you'd still want to maximise your airflow over & through them, not sure how much airflow is in that area of the chassis - I'm thinking not much airflow?

 

That area is covered by an open grill surface, and most of the time is it on one of my modded U3 coolers, so it would be pretty great I think. Three 120mm fans are blowing directly into that space 99% of the time, even while using it on my lap in the recliner. I almost never use any of my laptops without a modded U3 cooler under them.

That would have even been helpful before I added the heat sink. The air was blowing directly on those naked components all the time.

 

Ah, in that case, I reckon it could make quite a difference!

 

OK just my machine crashed while playing BF1.....I need help to understand what this means, which "driver" is the issue here?
The dump report:

Code:

On Fri 01-Sep-17 00:05:41 your computer crashed
crash dump file: C:\WINDOWS\Minidump\090117-6140-01.dmp
This was probably caused by the following module: ntoskrnl.exe (nt+0x16C560)
Bugcheck code: 0x1E (0xFFFFFFFFC0000005, 0xFFFFF803476E0787, 0x0, 0xFFFFFFFFFFFFFFFF)
Error: KMODE_EXCEPTION_NOT_HANDLED
file path: C:\WINDOWS\system32\ntoskrnl.exe
product: Microsoft® Windows® Operating System
company: Microsoft Corporation
description: NT Kernel & System
Bug check description: This indicates that a kernel-mode program generated an exception which the error handler did not catch.
This appears to be a typical software driver bug and is not likely to be caused by a hardware problem.
The crash took place in the Windows kernel. Possibly this problem is caused by another driver that cannot be identified at this time.



On Fri 01-Sep-17 00:05:41 your computer crashed
crash dump file: C:\WINDOWS\memory.dmp
This was probably caused by the following module: win32kbase.sys (win32kbase!RIMFreeHidTLCInfo+0x214)
Bugcheck code: 0x1E (0xFFFFFFFFC0000005, 0xFFFFF803476E0787, 0x0, 0xFFFFFFFFFFFFFFFF)
Error: KMODE_EXCEPTION_NOT_HANDLED
file path: C:\WINDOWS\system32\win32kbase.sys
product: Microsoft® Windows® Operating System
company: Microsoft Corporation
description: Base Win32k Kernel Driver
Bug check description: This indicates that a kernel-mode program generated an exception which the error handler did not catch.
This appears to be a typical software driver bug and is not likely to be caused by a hardware problem.
The crash took place in a standard Microsoft module. Your system configuration may be incorrect. Possibly this problem is caused by another driver on your system that cannot be identified at this time.

 

Try adding just a little more CPU voltage. For example, if you are using override 1150, try 1160. If it happens again, try 1170.

Edit: If the voltage is too low it can appear to be a driver problem. And, remember, if you are doing something fairly strenuous for a long time, like gaming, and the CPU gets warmer than it normally does, you may need more voltage than you would doing something that is more strenuous, but brief... like a Cinebench run. High temps needs higher voltage to achieve the same stability.

 

I'm running on 1090Mv, if I go 1100Mv might as well OC a bit more to 4.6Ghz.

 

There is a good chance that is your problem. 1.090V is very low. It could run like that forever for most things and never miss a beat. Stock is like 1.250V. Games are a strange thing sometimes. Most are not as strenuous as benchmarks, but as the temps go up and the machine is under a moderate load for a long time glitches can surface that you would never see doing things other than playing games.

Games are kind of like taking a car on a long cross-country trip. You can run it hard and drive it like a maniac on short trips, but the long trips are when things that might go wrong start to surface. The engine and transmission, and other components like AC compressors and alternators, start building heat from running for hours on end, etc.

Try bumping it to 1100 and leave it at 4.6GHz and test. Make small changes one at a time and test between changes. If it never crashes at 4.6GHz with 1100 you can always goose the clocks more tomorrow.

Edit: And, it could be just BF1 does not play nice with an overclock. There is a lot of complicated code and calls made by game software. I've had some games that play solid at 5.0GHz all day long and others that won't even launch correctly with 4.7GHz. And, it's not limited to CPU. Some crash with a small GPU overclock and some crash if you run with the LCD overclocked (custom display refresh rate rather than native refresh rate).

 

I'd be really nervous at that low voltage. I'm running at 1.140, and I found that even down at 1.125 my system was marginal (current problems notwithstanding, my system had been extremely stable at 1.140V before this current round of issues). I wanted to undervolt to keep temps down, but even at 4.2 GHz clock 1.100 V was unstable for me. YMMV, of course.

 

If you have some on hand, you might try some black foam (make sure it's not the conductive kind, of course). Black both absorbs and emits heat energy faster than any other color. (Reference: http://astronomy.swin.edu.au/cosmos/B/Blackbody+Radiation et al)

I don't know if the color will make a measurable difference in this situation, but it would be an interesting experiment.

 

Confirmed: I am able to replicate that long DPC latency if LatencyMon is running when the machine goes into screensaver or sleep. So that high DPC value was a red herring.

 

I do have some black foam that I can use. Those components have no temperature sensors, so I don't know if there would be a way to measure a change.

I will play around with that some more in the near future. The CPU is temporarily in the P870DM-G for my special modding project. I've got another 7700K on order from monoprice.com that should be here in a few days, but the 16L13 is enjoying some well-deserved R&R until it gets here.

 

I've never had any hitches with Battlefield 1 and my OC of 4.8 or 4.9. I would look at those voltages, as you mentioned.

Sent from my SM-G935T using Tapatalk

 

Sooner or later will a massive undervolt (on the edge) backfire. Press down voltage to the edge aka 10 or 15mv lower voltage than what you might have to will not constitute the vast difference in max temp. People should instead fix the underlying problem!! If temp is a serious problem. When I see people talk about 150-200mv undervolts, I mean they rather should fix the overheating. Or clock down a bin or two. But thats me

 

Why does this confuse you so much? You keep saying instead of tuning the voltage for usage, instead leave the voltage the same and overclock more to use the voltage?? It makes no sense

The key with voltage tuning, and power tuning, is to use the minimum power / voltage required to get the most performance, otherwise it's wasted as heat generation.

That's what we are always talking about with undervolting. Tuning for proper voltage use for the maximum performance.

At stock you can usually undervolt more, as less power is needed at stock vs overclocking.

But, when overclocking to the maximum, undervolting - if possible - will give you even more headroom for overclocking. So in this case undervolting gives you higher overclocks.

Maybe it's your way of describing it?

In essense with CPU's you can't OC undervolting is the only tuning you can do, there is nowhere for the extra voltage to be used

For unlocked CPU's you can increase the OC, but the point is to know your proper voltage at each clocking level.

At stock, vs increasing OC settings 1, 2, 3, maximum there will be increasing levels of voltage necessary - or not, it maybe that the default voltage set in the BIOS is too high even for the highest stable clock, as with many Kabylake CPU's that can undervolt at highest OC.

So, when you say "When I see people talk about 150-200mv undervolts, I mean they rather should fix the overheating. Or clock down a bin or two."

They are fixing the overheating by reducing the unnecessarily high voltage, it's higher than needs to be set for that OC, they can maintain that high OC by undervolting - that's how we fix the overheating, by undervolting, get it?

Clocking down a bit would make the overvolt even higher in relation to the CPU performance, they could actually undervolt even further at lower clock speeds.

Try it, you'll like it

 

See below.

Have and will never run stock clocks. Haven't Intel 6700/7700hq BGA. Or a overheating 6820/7820Hk

Have yoo seen max oc'd Intel cpu run with undervoltage? See I mean MAX!! I use TS 4 power profiles. Of course I put lower voltage than stock for some profiles. But I dont need have to squeeze last single mv for making my chips running cool.

Yeah, Some don't bother with fixing the real problem. I would start with whats's under the hood first. The opposite of many others preferred solution.

 

My description had nothing to do with stock clocks, any speed can be used to pick for undervolting, that's the point, it has a use at all speeds, even maximum OC if the CPU is overvolted by the BIOS so high there is room to undervolt at highest OC, which it is for Kabylake.

The real problem *is* the too high voltage set in the BIOS by default. That's the real problem. Not the heat which is only a symptom of the too high CPU voltage setting in the BIOS>

If you can undervolt and run with less voltage, it reduces the CPU load temperature by 5c-10c, it's running stable, and performing the same as when overvolted, then undervolting is a great way to solve too high temperatures. Period.

Why is this so difficult for you to understand? What doesn't make sense to you when you step through it all?

Even novices can follow the description, undervolt, and stop thermal throttling while getting better performance.

BTW, undervolting / setting voltage on LGA CPU's works too. On BGA the offset is often the only way, as the voltage is locked. However you do it, tuning the voltage is a basic part of tuning your CPU.

 

There is a language barrier here and I imagine what @Papusan is stating is make sure your thermals are good from a physical stand point - correct fitment of heat sinks and proper pasting, delidding where possible, and making sure you're getting proper air flow. @Papusan prefers to maximize his system. His approach is to fix the physical limitations, then use whatever voltage he has to get his clocks as high as safely possible. Use all the voltage you have before dropping it by maximizing your clocks. Once you've hit your thermal limit after those steps have been completed, then back off voltage first then clocks second until you are comfortable with your thermals. It seems reasonable. It may not be everybody's approach, but it is a great process in maximizing your system if that's your intention.

Sent from my SM-G935T using Tapatalk

 

Ease up man...that is a bit too far to the rude side. You are not talking to a n00bie, or a child. Show some manners. See the interpretation @leftsenseless makes below and maybe you will see you were missing the point altogether.

 

Max undervolt cant't save yoo from similar as the dreaded TRIPOD MESS People have tried, but with not very good results. Even with 220mv lower voltage, yoo are doomed... If the processor in you brand new laptop can't run properly with stock voltage and need to be undervolted, maybe it's time you start looking for another brand. Don't defend Trash!!

@Donald@HIDevolution Thanks, but no problems Words can backfire. There is a language barrier here. I'm sure we understand each other

Edit. @hmscott When you are at the edge(Near BSOD)... How much can you reduce the max temp if you undervolt 10mV more than what you really should? The difference between RMA or Keep it? Cut vs Shave - What's the difference?

 

Was the brightness control issue ever fixed on these? My friend was asking, sorry if it was already mentioned.

Brightness control on the 120hz display of course.

 

With the Prema BIOS and the latest drivers, yes. Get Prema, get HID. Hahaha. No, really.

Sent from my SM-G935T using Tapatalk

 

My friend bought from Eurocom awhile back since he could only afford the 1070 model at the time and HID did not offer that one.

I'm assuming @Eurocom Support still hasn't gotten brightness to work then?

 

They are a Prema partner. They should be good.

Sent from my SM-G935T using Tapatalk

 

I totally get it. Using the right amount of voltage is what Brother @Papusan is talking about. Depending on the clock speed it might be an undervolt or it might require an overvolt. We both have to overvolt for the kind of overclocking we do because the machine won't boot correctly without more voltage. If a machine has an excellent cooling system there is no reason to undervolt. Intel gives way too much voltage for the lame cooling systems we are all blessed with in laptop land and you can get a nice healthy overclock on 6700K and 7700K using stock voltage with the right cooling system. I just need a tiny bit more voltage than stock for 5.0GHz. With a good cooling system there is no reason to run stock if the machine is stable and cool, just leave the voltage alone and increase the core ratio. The problem is most laptops have a lousy cooling system and if you want to run stock clocks undervolting is required to keep the temps under control. Undervolting is a workaround/bandaid solution, not a silver bullet. By undervolting we are not using the right amount of voltage. The right amount is stock voltage per Intel design, but the kiddos are so OCD about wanting thin and light trash that the ODMs have to compromise too many things, including cooling, and that is how and why we end up undervolting. Unfortunately, cutting corners on thermal management is not limited to thin and light turdbooks. Almost all laptops now suffer from poor engineering when it comes to thermal management. Good example of the " give 'em and inch and they'll take a mile" expression. Half-assed is now status quo and this has created a need for undervolting. It's not the right thing to do, it's just a necessary evil.

 

I don't think they are anymore.

I'll have my friend email them tomorrow asking for any updates but they don't seem to ever provide BIOSes unless you ship the laptop back.

 

Prema's magic requires a physical flash. After the initial load, the BIOS can be updated without having to ship it.

Sent from my SM-G935T using Tapatalk

 

I know you want to say the "real" problem is that it's a BGA CPU and not a glorious "the one and true CPU" LGA

If you are tuning for maximum undervolt for a locked CPU, there are 2 final points of tuning.

The 1st is the lowest undervolt at full load stability testing, let's say it -150mV (crashes at -155mV) and that shaves off 10c from the maximum load temperature.

Then as you know to do ( if you read and listen ), the 2nd stability test is at idle. Let the laptop idle at that -150mV undervolt, stop all the browsers, video playback, exit steam, etc - you want the CPU to downclock and reduce voltage to the minimum.

If you BSOD at idle, reduce undervolt by +5mV to -145mV and let it idle again, it's probably fine now, but it's worth testing again anyway - or drop it to -140mV if you are in a hurry and that should be good to go - if your load test was good.

Then, if you have an unlocked CPU, after doing the undervolt at stock, you know the total headroom available -150mV (assuming that's idle/load stable), so as you increase the frequency you will need to eventually start reducing the undervolt, but maybe not right away.

During tuning you can and should go "too far" and undervolt -10mV past stable, otherwise how else would you know what maximum undervolt setting is stable?

Some have 4.2ghz+ with -100mV more or less on Kabylake - those lucky people

Now I know 4.2ghz isn't 4.3ghz+, so you can project your LGA magnificence in your mirror and bask in the glow.

And, we'll be gaming at 4.2ghz at -100mV, having fun and not caring whether at some random time in the future we will need to replace our laptop motherboard - or whole laptop - to accommodate Coffeelake, like you will

Hey, it's all fun stuff, and that's what it's supposed to be, not a world only populated with uber CPU's that pop out of their sockets at your command.

Mmmm, Chili's good tonight...

 

Hmm, that was only true for the early builds.

 

They do remote flashing, but it is not a Prema BIOS. That is an EVOC exclusive offering for this machine.

Newer drivers fixed the brightness control thing. If I use drivers that are too old the brightness controls are broken no matter what BIOS is used. Under Windows 10, brightness controls are broken on my P870DM3 as well if I use older drivers. Under Windows 7, brightness control works with all drivers on my P870DM3. I think the stinkers here are NVIDIA and Micro$loth more than the BIOS. Also consider that the 120Hz panel is not a stock configuration for the 16L13. It is an aftermarket upgrade that MSI has nothing to do with. As far as I know, those with the stock 60Hz G-Stink panels have not had any issues with broken brightness controls. I value the 120Hz panel more than I value display brightness controls.

 

Yup...what Mr. Fox said.

 

A good example of this rationale is that undervolting was never a thing with Sandy Bridge and Ivy Bridge. Stock voltage was as low as you could go and it did not matter. It wasn't necessary. Everything worked right as long as the machine it was installed in wasn't an engineering abortion. Then along came Haswell. I believe--just my theory, not necessarily fact--that Intel began to allow undervolting using negative offsets from Haswell until now to facilitate poorly engineered systems. They used to build laptops that worked with the CPUs, (meaning adequate thermal management,) but now they don't. Today's strategy is to just de-tune the crap out of junky notebooks instead to compensate... making the CPU work for the laptop. In other words, the tail is wagging the dog now.

 

When I make a undervolt for the lowest power profile in TS. I shave off voltage. I do not cut it all the way down to the leg/bone so the headroom for errors is completely gone. No need for it if the cooling work as intended!!

EXactly!!

I'm sure you also mean this in your heart(but do not want to say it publicly)

Damn, What a decent overclock. I'm shocked this is possible with today's Turdbooks

No hard feelings, bruh +rep

 

Which part? The needing physical flash, or being upgradable in the field after that?

 

@Papusan

Spoiler: The New Seekers...

The Making of 'I'd Like to Buy the World a Coke'
http://www.coca-colacompany.com/stories/coke-lore-hilltop-story

The New Seekers
https://en.wikipedia.org/wiki/The_New_Seekers

Other Seeker's favorite's






The Seekers - I Am Australian, Waltzing Matilda, Georgy Girl (Live, 1994)
https://www.youtube.com/watch?v=v22SPtCFck8

 

if I remember well, on desktop side, cpus can use adptive or offset voltages from nehalem to today's cpus, first was introduced by the Asus uefi bios(raja thing).

Beside thermal things related to cpu's mosfets exposed without heatsink, I'll try to read later and post here the result, with a cheap IR gun, at mild OC of 4,5Ghz (that spikes at 1,30v but load at 1,20/1,25v).

I love the F5 but not its thermal with a 1080GTX, I would think as past in this thread, to accomodate a vapor chamber heatsink on it.
So any suggestion will be very helpful here, I've also contacted mizerable to know if his project goes to trashbin but got no response.

For brightness thing, I'm dual booting Win10 and Win7, got F5 from Woodzstack, obviously with Prema unlocked bios on it (the one with .030 in version nr.) and stock 150w 1080Gtx's vbios, 120Hz screen model, no issues on nv driver 382.53 on both OSes

 

Hotwell, not haswell.

 

Needing the physical flash. Once the Prema BIOS is there, it can always be updated remotely, IF it ever needs it...which would be rare.

 

Did some more digging into the speaker pop issue (because of course I did). The registry keys I mentioned previously were present after installing the Realtek driver but did nothing after changing them. One thing that did work mitigating the speakers powering down/up/popping was going into the microphone properties and checking the box that says "Listen through this device." Something else this does is prevent the pop when the system shuts down, but it still occurs when muting/unmuting the speakers. Upon rolling back to the Microsoft drivers, I saw something interesting. Using those, an extra entry in the sound control panel appeared for the speakers. Enabling that one caused the same popping/power management problem as when the Realtek drivers were installed, but enabling the second set solved all the problems, save for the pop when the system shuts down. This is all very bizarre, but I think I'll stick with the MS drivers for the foreseeable future.

 

Did you try opening the realtek driver app from the status bar at the bottom, ten click on the battery icon , turn off power management and restart the system ?

 

For some reason, the most current Realtek driver (version 6.0.1.8186 IIRC) available from their website didn't install the control panel so that option wasn't there, but the slightly older one from Windows Update did. Regardless, I'm not inclined to pursue the matter any further. Perhaps I'll revisit it down the road after a few more driver releases.

 

An interesting idea. In theory there is no reason ti wouldn't be able to make one fit. But a one-off will likely be very expensive. You would have to convince HIDevolution or Eurocom to invest in it as a mod in order to make it affordable.

In the meantime, have you tried an under-volt curve? My GTX 1080 runs at 0.981 V without downclocking

I ordered some foam copper hybrid heatsinks for the vRam and mossfits. Was thinking of getting some fin copper to put on the GPU as well. With the bottom cover mod, there is much more airflow

Foam Copper heatsinks Datasheet
http://www.mouser.com/ds/2/733/VTL-LowProfileHeatsink-DATASHEET-MARCH2015fv-552995.pdf

 

These were took at 26° ambient temperature, today afternoon,

R22/R23 chokes during 3 loops of Cinebench 11.5, 6700K@4,5 1.20v load(1.35v spikes) multiCore maxed temp at 41.7°










The only capacitor present maxed at 46.2°


and the transistors (mosfet) maxed at 45.5°




3 test runs were on Cinebench 11,5, with laptop downside and no bottom cover (obviously)

 

We will be VERY interested in your results with the foam copper hybrid heatsinks!!

The copper fins will do nothing for you as @Mr. Fox has earlier reported, and our testing has confirmed.

 

Well, I guess heatsinks are not needed then.

 

I thought @Mr. Fox testing was done before the bottom cover mod? I was hoping with the much improved airflow, copper fins would be viable

Edit: Use case is also a little different, as I do not use a laptop cooling pad.

 

Well, no harm in giving it a try. When you do it, please be sure to test with just that mod alone done so that there will be no other contributing factors.

 

Those temps are fine then, praps not worth cooling them in your case, although they're bound to run hotter with the cover on - by how much I do not know.

 

Thanks for clarifying.