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

I would have to agree here, I have a single 512GB M600 in my Blade and I'm perfectly happy with the performance of that aspect of it. I might just buy a 2nd one and configure them in Raid 0 when I pull the trigger on this unit.

 

So I repasted my CPU, finally. More Kryonaut. D did it 3 days ago now. It's funny I know there is no "set time" with kryonaut. But immediately after pasting maybe 1 hour. Played some league and was hitting 92C,wasa little upset because it was 85 before the re-paste. Now 3 days later same game setup, actually playing for longer I am hitting a high of 78... I am happy I didn't immediately repaste. And after tweaking my fan profiles I am now maxing at 73C

 

Does anyone know if the newer models shipping with Kaby Lake CPUs has an updated motherboard, or is it the same one as before? I'm just trying to get the exact model and chipset so I can look at some reports of hardware support under Linux and BSD. I imagine whatever Eurocom is using ATM will be what I get from HIDevolution.

 

That's awesome!! Noone else has had the patience to try this... wow, think of all those people that repasted immediately - they shoulda waited!!

@Phoenix @iunlock @Papusan @Mobius 1 @D2 Ultima @Mr. Fox

 

The EVOC 16L is using the C236 chipset.

 

Yeah I know it goes against everything for non metallic thermal paste... I remember a couple hours after the repaste my wife was like "Why is your laptop so loud now"... Then last night, "I see you fixed your laptop"... I was like wah? Realized my fans were running quietly, popped open HWMonitor and was amazed... Played for 6 or so hours last night with some buddies and was at 78C (This is before I played with the fan profiles)

I am running my 6700K at 4.2Ghz on all cores with a 150mV undervolt on core.

 

Would like to know the CFM rating if you can find that. 4k RPM sounds great if it's pushing 75 to 150 CFM, but not so much if it's only pushing 15 CFM

 

So I have been investigating occasional crashing in games for the GTX 1070 version of this notebook. Specifically, I am talking about a sudden black screen, with a "LiveKernelEvent" 141 entry in the Windows Reliability Monitor. Hopefully this information will also help anyone else who is affected:



The key to diagnosis is to find a way to reproduce the problem, and I have finally done just that. I have experienced these crashes in Gears of War 4, Battlefield 1, Deus Ex: Mankind Divided, and now I will demonstrate in Black Ops 3.

Here is a video of the crash:


Here is what I have discovered:


Underclocking the GPU core gets rid of the crashing. It would appear my GTX 1070 is not stable in this scenario, even at stock settings. I am building a case that the issue is hardware related. Is this enough evidence? What else should be tested?

The last thing I want to happen is if I RMA my notebook, and the technician is unable to reproduce the issue (due to the inconsistency of the problem). That would waste time and money (shipping back and forth). I am hoping that the evidence provided here is enough to warrant a GPU exchange. Although it DOES crash in other applications, the only way I have been able to get it to crash every time is in Black Ops 3.

This error occurs even after changing NVIDIA drivers and a full reformat of Windows.

 

I just removed the Kryonaut from the Tornado F5. It was the worst temps I have seen on the Tornado F5 so far. Coverage was flawless, too. So, not sure what the deal is. I found the same problem on the P870DM-G... Kryonaut was almost worthless to me, almost like that worthless NT-H1 garbage paste. That's a bummer, too. I like how you can just wipe it off real easy. Clean up is a breeze.

I'm not ready to go CLU yet because I am not done taking it apart, so this time I used Gelid and the temps are a lot better than they were with Kryonaut. I don't know why because I don't like Gelid GC Extreme either, but I did not expect it to do better than Kryonaut. IC Diamond is better than either of them in my own testing.

 

Kryonaut is the worst thermal paste I ever used. Glad you're happy with it

 

So far i have taken mine apart about 10 times and the Thermal Grizzley Conductonaut is still there in liquid form. Might be something to try later on....

 

Ah yes with Kyronaut, interestingly enough there seems to be a cure time. Funny isn't it? I have kyronaut on some machines and it's still going strong.

Strangest thing eh? What's interesting is that I remember a 1.5 hour long video that I had watched when Kyronaut first came out with the developers talking about how Kyronaut is the only traditional paste with the highest content of "metals," in a traditional paste, not enough to be conductive, but an amount that exceeds all other traditional paste. It was also mentioned that the binder used was specifically formulated it stand some serious abuse as this was made for extreme over clocking....

It doesn't surprise me that it has characteristics of its own....like this whole cure time that we may be seeing....we'll I've seen it and there have been quite a few reports with the temps improving with @cavell219 being another testimony to this....interesting...

Hmm...perhaps there's a contact issue going on with the IHS -> HS. Kyronaut as you know is very thin, even compared to the average traditional paste.

For uneven situations and with Kyronaut being so thin, it's clearly not the best choice. However, as I've mentioned before I think what's important is attending to the uneven situation first to get that rectified, instead of patching the issue with thicker thermal paste like ICD. - But whatever works eh?

 

Just thinking out loud and you very well may already have, but you did DDU before installing the new driver yes?

Have you tried an older-older driver?

Also check in Device Manager to see if there are any unknown devices being reported. I've seen this before..

 

I don't re-paste. Put on liquid metal and thats it

 

CLU can be re-used the same way. I've taken it apart and put it back together without using more CLU and it worked fine. Just took the brush and smoothed it out again. I need to order more of it.

I have some Conductonaut left, but I am just too scared to use it. I used it twice on the P870DM-G and both times found it ran off the edge of the CPU onto the heat sink with a stream of it headed for the edge of the copper plate. Had that happened on an Alienware with the CPU on top of the motherboard, it would have run down the side of the CPU and potentially gotten into the CPU socket. That just scares the crap out of me.

I did a test fit with Phobya Liquid Metal the other day on the Tornado F5 (which I also have a couple of tubes of) and a stream of it ran sideways on the heat sink the moment I went to flip it over and screw it down, LOL. The Phobya Liquid Metal, Coollaboratory Liquid Pro and Conductonaut are all too watery/runny for me to feel comfortable using them on a laptop. I've never once seen CLU move from the place I brushed it on, which is why it's the only thing I will use now.

So, here are the fans. CPU and GPU are the same part number. I'm doing some online searching and not finding CFM or other specs so far. Aavid Thermalloy does not list this MSI laptop fan on their web site and their part number decode table doesn't identify it.

 

Yes, I've tried many, many different driver versions always using DDU. In fact, I just did a fresh windows install yesterday, and it is still happening. No unknown devices in the Device Manager. To be honest, I really do not want to RMA. It is such a huge risk (shipping) and may take weeks to be returned. I am currently exploring software solutions, and I may be on to something... I'm not sure if anyone else is following, but I will report back with my solution after testing.

 

Very nice. Thanks for sharing.

I do agree that CLU is the most well behaved LM. It stays put. The rest are a lot more tricky to work with and to find that balance of how much to put (just the right amount) is a challenge, but what makes it even more of a challenge is when there isn't a solid contact. So one variable on top of another...

Those fans look like they'd be in the ball park of around ~14-15cfm? How think are the fans?

Great stuff...exiting thread...

ps...Have you heard anything about the F7?

 

My go to regular paste is PK-3. After a short trial with Thermal Grizzly I've gone back to PK-3. I find it to work well in laptop applications. The line on the GPU and X method on the CPU give me excellent coverage, it has great conductivity and lasts a long time with zero cure time. But IMO these pastes seem to all work within a few degrees of each other. What I've found is the pastes don't all seem to last. PK3 and ICD seem to last for at least a year or more.

Now I'm getting the delidd and CLU itch. I won't dare touch the GPU with it but under the IHS and a reseal seems harmless and very beneficial.

 

I'm not sure on the CFM and so far found nothing. Seems the same basic part number is used on a bunch of MSI notebooks and only the outer housing/screw tab placement seems to vary. Unfortunately, I think it is a low CFM fan. Sadly, almost all laptops have stupid low CFM fans... probably because too people whine and moan about loud fans on laptops. I wish these put out as much air as the violent suckers on the P870DM3... I just love those crazy monster fans. I find it very satisfying to hear them blasting gobs of air through the radiators. I don't know what CFM they are, but I am pretty sure they are way more CFM than these.

I haven't heard anything more about the F7. Should be interesting. It's going to be tough for anything to be better than the P870DM3... the bar has been set really high with that bad boy.

Yeah, just stick a small square piece of Scotch tape over the gold contacts under the IHS near the die and you're good to go with the CLU. If your heat sink fit is good, using CLU between the die and IHS will make you never want to use PK3 or ICD ever again. CLU performance totally eclipses the non-metal pastes, but if the heat sink fit isn't good liquid metal paste isn't effective at all. Using a 30mm x 30mm x 0.5mm copper shim on top of the IHS with CLU on all the mating surfaces worked extremely well on my P870DM3. HIDevolution is testing that and finding equally excellent results. I forgot to ask @Donald@HIDevolution if they are using IC Diamond or CLU on the shim, but just adding the shim will knock off anywhere between 10 to 20°C.

 

Great stuff. Please keep us posted.

Have you tried putting the mxm card in another slot and/or another system (if available)?

I really think it's a software issue though of some kind...that error code is usually resolved with a new driver install, but I'm thinking that another so

 

As I reading your reply a crazy thought came over my mind... how interesting would it be to incorporate a bladeless fan technology into these machines like:



That could move some serious air while curing the db's...

As for the copper shims, I've tried both LM and traditional paste and have found that LM worked pretty well since there is more mounting pressure with the shim being present.

Not that it matters, but with copper having a rating of 401w m/k, if you factor in the thermal conductivity loss from stacking (factor of 2) and with 2 contact points (IHS-> SHIM <-HS) it would still exceed the value of the LM's rating so there's a lot of head room there where the efficiency is not affected.

ie...the bottle neck is in the thermal paste / LM itself if any, which makes the use of copper shims great when needed as it provides a great medium even with the law of conductivity at play.

 

@Mr. Fox 40.9 to 65 CFM

 

Well, that's more than I thought. Where did you find the specs? And, thanks.

 

There is definitely something funky with the batches you guys are using. I havent had any such issues since the past year i have been using Kryonaut. Same with my friends systems that i have re pasted.

Yeap, on my DM1's 980 the last 2 times i took it apart , all i did was to scoop and move the paste to the center of the die, and temps are exactly the same as before.

 

Right now it's speculation because I look at the fan they were talking about it it's slightly different. And I couldn't go back and edit my prior post.
So I'm using this one as a base till i get an answer back on the exact one.
https://www.aliexpress.com/item/New...-GX660-GT680-GT683-GT60-GT70/32789543714.html

 

@bloodhawk
All I did was set the laptop flat and it ran more or less to the middle.

 

LOL ill try that with the DM3, once it get delivered this Tuesday. Provided i get time after commuting from Santa Monica to NoHo @ 7PM xD

 

I have to shoot past there later tonight. Chatsworth. I do not like driving during business hours out there. Total 4 hours driving time. (or Sit in traffic time that is)

 

Dunno though, I'd still take a minimum 24h burn in period on full system load with any thermal paste.

It's Z170, CM is for mobile.

Have you guys check contact between the heatsink and the CPU?

AFAIK the 7700K is even thinner than the 6700K, the 16L13 and P870DMx heatsink is designed for the thicker Skylake 6700K PCB and on top of that the factory pressure / QC is not that great (ie: Fox using shim with CLU on the 16L13).

Kryonaut works great for 8 months without any repaste on my Hotwell-E that's spewing over 190w on 4.3GHz.

 

It's on the specs on HID's site and confirmed by the Donald himself. I read C236 is similar to Z170.

http://ark.intel.com/products/90594/Intel-C236-Chipset

 

I add more info http://www.tomshardware.com/news/intel-c232-c236-chipsets-xeon,31051.html

 

@Mobius 1 is right. The Tornado F5 has a Z170 chipset. Straight from Eurocom... http://www.eurocom.com/ec/configure(2,384,0)TornadoF5

 

See cpu-z mainboard

 

I did specify EVOC 16L
But interesting, in that I thought that the Tornado and the EVOC are the exact same machine. Maybe @Donald@HIDevolution can clarify.

 

Are you sure that's "bladeless" ? There are blades inside the base, though.. Asymmetrical ones. Which is why it's so quiet.. Like Apple uses in all their 2012+ Macbooks. Which is why they are virtually completely silent up to 4000 rpms.. I wish more makers would use real engineered, thought out fans for laptops. Speaking of which, I went through about 6 laptops from Best Buy recently trying to find the "perfect" small 2-in-1 laptop but all except for the completely fanless Dell XPS 13 2-in-1 were all plagued with high-pitched fan noise immediately causing me to return them all.. Surprised I haven't been banned from Best Buy at this rate but appears i'm still good.

If you need a small, yet powerful laptop.. Seems Apple is still the only maker out there doing it right - and you have to pay in excess of 2k+ to get a fast one :/ Oh well - seems i'll have to go that route for my next small on the go machine. The Dell XPS 2-in-1 fanless design was almost perfect, but the CABC brightness issue ultimately had me returning it.

 

Ah yes, I should have clarified. What I had meant by bladeless was more so the technology to move a lot of air while keeping it almost silent even at high cfm's. You're right, with the technology that exist already there should be more focus on this area...because it can be done.

 

It's pure laziness on the manufacturers part.. and or companies like MSI, Asus, etc. are limited by the fans that are already made by the fan-maker companies. It really doesn't matter what I think though, I would bet more than 80% of people do not care/notice loud fan sounds but it annoys the crap out of me when I value a quiet machine.. Or one with fan profiles that turn off the fan under normal load. I had a Razer Blade 14" with 1060 for two days that went back because the fans were always running (both GPU/CPU) and for no reason at all should they be running at 40% speed idling. I blame the EC bios devs on that one.. It's overkill and completely nonsense. How they can take a fabulous machine and in my eyes, is completely ruined by it's fan noise/profiles.

That being said, with how seriously anal I am about fan noise - the Eurocom F5 (and the other versions) is a very acceptable sounding machine that doesn't really get noticeable until something very demanding is running.. It's done well and perfectly acceptable. The small laptop market has a long way to come in terms of fans.. it's just bad.. all bad.. Except for maybe the Dell XPS 13 (non-2in1) which was perfect when I had mine.. Fan didn't turn on until 80c or so and it was barely noticeable when running and still managed to lower temps tremendously well. And as soon as the load was gone, the fan would turn back off nearly instantly. That would be the only machine i've owned that matched Apple's fans.

 

Okay, so after exhaustive testing over the past week, I conclude that some GTX 1070N GPUs may not perfectly stable when Vcore exceeds 1.031v, core frequency is high, and VRAM usage is high. This issue was so difficult to diagnose because most people think to run stress tests (Furmark, 3DMark, etc.) to check stability. However, these stress tests actually cause the Vcore and Core Frequency to drop, due to the power throttling. For example, when running Firestrike the (GTX 1070N) Frequency will drop to ~1700 MHz and Vcore will hover around 0.900v. Thus, the symptoms will not reveal themselves in normal stress tests.

I was able to reproduce the crashing periodically (~1 crash every 3 hours) using MSI Kombuster's VRAM burner, but this would take forever to diagnose. Even better, I found CoD: Black Ops 3's "Sand Castle" level to be a more consistent stress test. By capping the game's framerate to 90 FPS, you can effectively create the "perfect storm" of high Vcore, Frequency, and VRAM usage.

The built-in power restriction is the most likely culprit for instability by forcing Vcore to drop even though the card wants to run at high frequency. The simplest solution is to undervolt. You can't control Vcore directly (thanks to GPU Boost 3.0), but you CAN actually control it indirectly using a Voltage/Frequency curve editor.

One of the quarks of GPU Boost 3.0 is that it will refuse to step down frequency with increasing voltage. In this way, you can cap the voltage of the GPU! For example, if the GPU normally boosts to 1850 MHz at 1.031v, setting the frequency at 1.043v below 1850 MHz will force the GPU to stay at 1.031v.

When set like this, the GPU Vcore will not exceed 1.031v (where the dip starts):


You might ask, what about performance? Aren't you limiting the GPU by underclocking/undervolting? Well, actually, due to the power limits put in place by Nvidia, performance will not change. This is because under heavy load, the GPU has already throttled itself well below 1.031v. Under light load, high frequency is not needed. Final results:

No crashing, and happy

 

Yeah I'm still curious to know too.

 

If you have to do this, it means your GPU is faulty.

A GPU should be able to boost on STOCK CLOCKS no matter what and not crash.

Demand a replacement.

 

I hope that @woodzstack will not encounter those issues from the laptop I will be buying from him.

 

Heh........ this is very very interesting.....

@woodzstack

 

I'm thinking its more temp related and or memory related. When GTA 5 would crash on me it had to do with the ram.
And if you cards are not below 42C then for get run

What is the temperature at?

 

Grrrrrr. Edit not working....

I'm thinking its more temp related and or memory related. When GTA 5 would crash on me it had to do with the ram.
And if your cards are not below 42C then forget running anything optimal after that point.

 

How come?

 

I meant to quote keith. Misquote, but I am gonna test something....
I had a bunch of issues on my laptop with games crashing, gonna test if this is a solution to the issue. There seems to be a lot of factors at play here.

 

I agree to a certain extent. I think I lost the silicon lottery, and received a lemon GPU. However, I wanted to avoid an RMA if at all possible as it would cost time, money, and stress. Not to mention, the conditions for crashing as so specific, it is highly unlikely the technician will be able to, or put in the effort to reproduce it. There is also no guarantee that a replacement GPU would solve the issue. I can still reach the same boost clocks as everyone else, I just have to keep the GPU undervolted.

You are more than likely in the clear. I would bet only a handful of people in the world have run into the same issue I have. First, you have to lose the silicone lottery, and secondly the "stars have to align," so to speak, in order to create the perfect storm for a crash.

 

I hope my post is able to help you!

However, finding an application that is able to consistently crash is key, in order to narrow down factors. You never know if you might be encountering a separate issue which would send you on a wild goose chase.

The GPU core never exceeds 75 C. Sometimes it would even crash in the main menu of BO3 right when I launch the game, when GPU temperature was in the 60's. I had previously suspected it was temperature related, so I purposely turned down my fan speed and tortured my GPU using Furmark. It ran at 85 C with no crash.

 

Yup, temps are critical for Pascal. Here's a personal best with the Tornado F5... more to follow as time permits.

 

This question may have already been asked, but in case it has not, have you tried making sure the power management mode in NVIDIA Control Panel is set to Prefer Maximum Performance? Optimal and Adaptive can sometimes be unstable, especially with Pascal and its low voltage, high clocks routine.