Help with MSI 16L13 (Eurocom Tornado F5)

Hello,

I have tried asking around and searching forums but unfortunately, no one has responded to my issue.
And even reps from Eurocom can't explain what is happening.

So, it first started with games completely crashing to the desktop. It would hang and then just close the game. I wouldn't get any notification about Nvidia driver errors or recoveries.

But this only happened in certain games. For example here is a list of games it would crash on:

Titanfall 2
Space Engineers (After 5 or 10 minutes)
The Division (After 5 or 10 minutes)
Metal Gear Solid V: The Phantom Pain (It varies, sometimes I can play for hours. Sometimes it crashes after 10 minutes)
Watch_Dogs 2 ( Instantly or 10 minutes after launching)
Dying light (within 5 to 10 minutes after starting)
Battlefield 1 (after about 30 minutes of play)

Games that I could play for 6 plus hours with no issues:

Just cause 3
GTA V
Fallout 4
Deus Ex: Mankind Divided
Ghost recon Wildlands
Sniper Elite 4
Wolfenstein the old blood
Portal 2

After this issue became clearly a major problem and limited the functionality of my laptop I reached out to Eurocom. But unfortunately, my laptop is no longer under warranty so I cannot RMA it or send it in without it costing me an arm and a leg.

The only thing they had me do was flash the vBios to one they recommended and it only made the issue worse. Instead of crashing to the desktop, my whole system locks up, and power cycles and reboots.

I have tried re-installing windows twice, tried older and new drivers, I lost the old vBios file so I can't downgrade, I have tried installing the OS on a 2.5 inch SSD, and on different M.2 Drives with no change in the issue.

GPU never goes past 76 Degrees in temperature as I use fan boost when gaming.

Please, any help would be appreciated.

I am willing to do anything to get my beast working properly.

Specs:
Core i5 7500
GTX 1070 8GB (Desktop)
32GB DDR4 3000mhz
256gb WD black M.2 OS
512 GB A-Data M.2 New games
1 TB WD Black HDD for old games
120hz 1080p LCD
230watt power adapter

 

I just happened to notice your post in the Owner's Lounge for your system a few minutes ago and I could not find where anyone responded to you. I looked on your profile and it appears your request for help was ignored. I apologize that nobody replied to your request for assistance. See quote below.

NVMe always runs warmer than SATA SSD. You need to get some coolers for them to help with that. Just be sure the cooler is not too thick to fit under the cover. DO NOT use a thermal pad between the motherboard and NVMe drives, as this makes them run hotter.

I am puzzled by the second thing you mentioned. You should not be using an AMD RAID driver and if the BIOS is set for AHCI mode the drives should not be showing as RAID SCSI devices in Device Manager. That is definitely not correct and I am wondering if something goofed up there is causing the NVMe drives to run abnormally hot. I never had an issue with them overheating to the point that they disable themselves until cool down in the Tornado F5. When you installed Windows from scratch (clean install) did that problem go away, or is it still showing up that way?

Are you still having an issue with overheating NVMe SSDs, or is that sorted now? If not, that could cause the problems with the games crashing if there are read/write errors from overheating.

 

I would say this seems some RAM or, most likely, VRAM error. I don't know how to find what's happening. Maybe there is a way to disable some RAM or VRAM modules to find if disabling one the error goes out... But I'm talking too much without enough knowledge...

EDIT: You can start monitoring the ram&vram usages and see if there is some weir behavior. If its ram related you are lucky because is easy to remove/replace. Maybe I'm going nuts and this is not even hw related!

Sent from my SM-G900F using Tapatalk

 

If you want to go back to the stock MSI 1080 vBIOS, you will find a copy of it in this archive from my Google Drive: https://drive.google.com/file/d/0Bwdqi25LDwZyVUhJdDhKNWFfSEE/view?usp=sharing

There are 3 vBIOS files in the package. All are stock and will flash using NVFLASH. The Clevo vBIOS and the stock vBIOS used for EVOC machines with 1080 will run hotter than the original vBIOS used by MSI because the Clevo and EVOC firmware are for a different GTX 1080 GPU and will allow your GTX 1080 to draw more watts than the original vBIOS provided by MSI.

 

Hello Mr. Fox,

I was hoping you'd see this since you seem to be very knowledgeable about these laptops.

About the NVMe Drives, it turns out the 128 GB A-Data XPG M.2 I purchased was faulty, although A-Data XPG M.2 SSD's have been shown to run extremely hot due to the cheap method of manufacturing. I ended up returning the 128 and keeping the 512 GB M.2 from A-Data for game storage and eventually purchasing the WD Black 256gb M.2 as it had better performance and didn't run as hot. After removing the 128 out of the combination the AMD raid Driver and RAID SCSI device issues went away after doing a fresh install onto the WD M.2

They no longer disable themselves and show up properly in device manager. I did purchase a heat-sink for the A-Data 512 just in case and it has shown no issues since,

 

Sorry Mr. Fox,

I forgot to mention in my post I am running the GTX 1070. If you have a Vbios for this card that I can try flashing to, that'd be more than greatly appreciated.

 

I do not have a GTX 1070 vBIOS for you, but maybe someone with a stock 1070 in the 16L13 will share theirs with you. Sorry, I missed that GPU in your specs. My bad.

Awesome. I am glad to hear that problem is fixed now. I wondered if one of the NVMe drives was defective.

 

No worries Mr. Fox, thanks for the replies. Hopefully, someone will see this with the stock vBios for the 1070 in the 16L13.

Do you know of a V-Ram testing application by chance? Because I can run furmark for hours without an issue, I can run the unigine benchmark: Supercomposition benchmark and get the correct scores. But only in certain games as mentioned it crashes.

 

Hello Txordi,

Thank you for the reply, I have tried new ram, and even tested ram modules in different DIMM's etc. Unfortently that did not resolve my issue. I wish there was a way to disable or limit V-ram available to test this.

 

Oh I edited my post too late hehe. It's nice ghat both of us thought the same, but what a pitty that the ram replacement did not solve the issue.

Sent from my SM-G900F using Tapatalk

 

That is really puzzling. Running Furmark for hours is not necessarily a safe thing for your GPU, so I would not recommend doing that any more.

I posted this (below) in the 16L13 Owner's Lounge to see if going back to the stock vBIOS helps. Hopefully, someone can share theirs with you. I am a bit concerned that your GTX 1070 has a problem of some kind, but it's weird that it only happens with certain games. It could be a bad vRAM chip that only causes a problem when it is utilized.

 

I have not tested them, but one of these utilities might work: https://www.raymond.cc/blog/having-problems-with-video-card-stress-test-its-memory/

Download Video Memory Stress Test

Downwnload OCCT 3.1.0

 

Yap, that's what I initially thought. What a nice answers Mr. Fox!

Sent from my SM-G900F using Tapatalk

 

Also @Mr. Fox

No worries, I have kept an eye on the ram and V-ram. V-ram usage I have pushed over 6 GB with no issue. I forgot to mention that I am able to play the witcher 3 maxed out completely for hours with no issue.

Do any of you think that this can be CPU related? I have maxed out my CPU usage and ram multiple benchmarks. But CPU stress tests don't cause anything to happen. But I wonder if its an issue with the PCI-e lanes or something that communicates with the CPU and GPU.

Also should mention that there are never ram dumps or crash dumps of any kind done. I always check after a crash and use the app "who crashed" and nothing is reported, unfortunately.

 

Quick question, bro. Your 1070 is MXM installed in the laptop and not a desktop GTX 1070 in an eGPU, is that correct? I am asking for clarification because you have "Desktop" in the specs relating to the GPU.

 

I have used Video Memory stress test, but I couldn't ever get it to work.
I will try the other links posted and post an update with results.

 

Yes that is correct, Eurocom Labels it as the desktop 1070 though. I have also tried both the desktop drivers and notebook drivers with same issues happening.

EDIT: the only difference is that mine has the SLI bridge connector removed @Mr. Fox

Here is an image you posted of the exact 1070 I have in my notebook:

 

Woah... At this point I have no clue... I have never went so deep in finding hardware related issues, so I don't know the specific error behavior of these components... I think that you have to heed Mr Fox and try to get your gpu vbios stock to begin with

Sent from my SM-G900F using Tapatalk

 

OK, thanks. Just wanted to be sure. If you were using a desktop 1070 with an eGPU, those can be very unpredictable.

It is possible the CPU or an issue with the MXM slot is the problem, but not likely. This could be a challenging issue to resolve without access to another GPU and CPU to test with. Sometimes swapping out parts and using the process of elimination is the only way. But, if you don't have spare parts available for testing, which hardly any end user does, that is not going to work. I am hoping someone will share their stock 16L13 1070 vBIOS with you and that might change things for the better. If I had to take a wild guess, it would be that the GPU has a problem, or the vBIOS you have on it is creating a problem. Hopefully, the latter and a simple flash will resolve it. GPUs are more prone to failure than an Intel CPU, but anything is possible.

 

@Mr. Fox

I just tried running OCCT and I got this Error: Your GPU does not support cuda, you need a Geforce 8 or higher card. Mem test disabled"

I will keep my fingers crossed someone that comes across this post with the vBios I need. But yeah I am not sure why I am getting that error.

I am unfortunately limited with parts and funds. I can't afford to take a dive on a new 1070 and have it end up being a motherboard issue and needing to replace the laptop itself.

 

OCCT has an artifact tester built in for VRAM. I take NO responsibility if you blow up your card using it.
*edit* I saw your error. What version of OCCT are you using? I'm using the one that came out last year with NO problems with a TDP modded GTX 1070 MXM.

 

I just tried launching the application, but I got the error as mentioned

 

Sorry Didn't see your edit until after, I am using the link Mr. Fox gave me. It downloaded Version 3.1.0

 

@Izen

What the hell?
Mine is 4.5.1.
And that came out last year.
What's Mr Fox doing linking you to one that came out 10 years ago? LOL

Here you go!

http://www.ocbase.com/index.php/download
@Mr. Fox

 

LOL... That was the link in the article with the list of 5 vRAM test tools. Did not know it pointed to such an old version. Sorry about that.

 

@Mr. Fox Just wanted an excuse to use the butt fumble, haha, all good!

 

@Mr. Fox @Falkentyne

Hahaha, it happens. But that's a good reason to use the but fumble lol.

And no worries, I'll download the latest one and do some testing once I get back from work. I'll let you both know the results.

Thanks again for your advice! I hope to figure out what's wrong with this beast.

 

@Mr. Fox @Falkentyne

So I ran occt for 15 minutes on CPU and RAM, and 15 minutes on GPU, no crashes, not spikes in voltages or voltage drops. at 1080p, render complexity at 7, and I maxed out the limit for testing V-ram of 2 GB. No issues, no spikes or performance issues. Temperatures were low and stable. (CPU and GPU both stayed around 55 to 60 degrees)

Could this be a motherboard issue? (I hope to god its only a motherboard or software issue as the GPU is not cheap to replace as I could get a used barebones for half the cost of a simular model from eurocom on Ebay)

 

Well, the results of the OCCT test is encouraging, but may not be conclusive.

Have you already received a stock 1070 vBIOS yet for testing? (I have to looked back in the thread yet to see if that happened.) It could be a number of things, but I would try that first if you have not already.

 

No, unfortunately. I am not sure where I would even look for one. Or who to contact for one.

Do you think I should try re-flashing the vBios Eurocom gave me? I still have that as they emailed it to me.

 

@Izen please use imgur to host pictures. Photobucket and imageshack are useless now. Everyone uses IMGUR now.

 

If what Eurocom gave you is a stock vBIOS, yes it is worth a shot. Perhaps @Falkentyne has his stock MSI 1070 vBIOS saved and you could try that one as well.

 

@Izen i'm really dumb today. I really should have read the original post you made.

Do you have MSI Afterburner installed?
If you do, set an average core overclock let's say +150 (most cards are ok with this), then a +500 mhz RAM overclock.
(haven't seen a samsung RAM card that can't do this).

then, press CONTROL F and bring up the frequency curve.
Now, click the 0.950v voltage point on the graph after setting up your overclock. Then press "L"
Then press apply.

Now play your games and tell me if you get any crashes.

 

It's from when I joined the site years ago, I will have to edit my signature. It was a signature photo for Alienware M17x r1 owners. I will fix it.

Unfortently no, it is a vBios image that ended up making things worse but claimed it would solve my issues. I was thinking maybe a bad flash.

I was using ASUS GPU tweek to downclock my card. I will definitely give that a try and get back to you!!

 

@Izen No need to use Asus GPU tweak. MSI Afterburner does everything you need anyway.
Ok try what I suggested, test multiple games and post back. To save time, post back as soon as you get one crash.

 

So I didn't do it with the overclock as I just wanted that extra level of precaution. But my god. I locked it at the 950 like you said to, and it played every game I have tossed at it without crashing!

Should I try with the overclock? Is there a way to set this to load each time windows starts?

 

Great news! Might as well try it with the overclock.

Yes. See the yellow circle? Activate "Startup" and it will do that for you.

Just be aware that if the GPU is overclocked too far and unstable at the time of shutdown, it will re-apply the same unstable overclock at Windows logon. Better to save it to a profile. Then you can select it from the tray icon whenever you want it applied.

 

Perfect, Thanks so much, everyone.

But can I be explained by chance why this works? Is this a fault in the vBios for voltage control? Or is the voltage set by vBios set too high? Or is my card just unable to handle the voltage used by the GPU boost when it clocks it self up?

EDIT: I will do the overclock tomorrow and report what happens, time flies when playing games. Need to be up for work in 5 hours lol.

 

Pretty much. Each GPU is different (same for CPUs) so it may be that your 1070 is one that doesn't play nice with the vBIOS defaults. Likewise, most Gen 6, 7 and 8 Intel CPUs receive way too much stock voltage and need to be undervolted to run correctly with stock clocks or a modest overclock. I think they do that because chip bin quality is random, inconsistent and sometimes not good enough to run stable with lower voltage. Rather than fixing the underlying bin quality problem by culling crappy chips to keep them out of circulation, they manage things based on their lowest common denominators and let everyone else deal with the fallout of high temps from unnecessarily high voltage. Unfortunately, handling things based on the lowest common denominator seems to be status quo for the industry now. It applies to other things as well. I guess that affords them the freedom to sell their junk with no accountability for the samples of garbage that fared poorly in the silicon lottery.

 

The first versions of the MSI GTX 1070 MXM video card (Revision 1.0) had flawed hardware (what was flawed? No one knows) that caused the card to crash when it hit the 1.013v voltage point. The card would not deliver the proper voltage, and either the driver or the entire laptop would crash. It was a crash far worse than just overclocking the card too far type of crash (on fully working cards). It's a crash that could even make the laptop fail to reboot properly or to boot into VGA mode or something (crashes like if you overclocked the RAM too far--really bad crashes). This was a very serious and eventually well known issue with users with GT62VR's and GT72VR's with 6820HK and 6700HQ processors, and some GT73VR with 6820HK processors. There were forum threads created here for that, which took awhile to grab up steam, and there were some unnecessary flame wars (and I'll admit I was one of the flamers at first), but eventually proof was posted that this was a direct hardware flaw. There was discussion about this on the official MSI forums also.

MSI refused to admit there was a flaw, and instead did band-aid fixes by supplying a "8A" vbios that prevented the card from using any voltage higher than 0.881v. Boost clocks still worked, but the card would not go higher than 0.881v. This was ok for some users with 115W cards, because in newer games and benchmarks, there was no difference in performance between the original bios on a fully working (no bugged) card, and the 8A vbios, because the power limit flag would stop the card from using the higher voltage and clocks anyway. But in older games, where the card could boost up to 1.0v and higher and 1900+ mhz (games where you always saw the crashes on bad cards), there was a noticeable performance drop with the .8A vbios.

Most Kaby Lake laptops (laptops made in 2017) came with the fixed revision cards. I was unaware that 16L13 users received the broken cards. I thought the form factor of the 16L13 1070 was different than the GT73VR 1070, but in fact it may be the GT62VR 1070 that has a different form factor than the 73VR 1070. But I don't know. Regardless, the form factor is irrelevant, if the same defective circuitry exists on both card layouts.

There was another vbios which disabled ALL boost clocks completely, a 'failsafe' vbios which MSI put on some people's laptops, even when they asked them NOT to.

This entire debacle, combined with the trade-in "Scam" debacle, caused MANY, MANY People to hate MSI. BGA vs LGA wars notwithstanding.

Eventually, late last year, MSI started actually replacing the 1.0 buggy cards with 1.2 revision cards, but ONLY if people actually asked them (well, demanded) that they did it. Since by now the cat was out of the bag and MSI knew full well that people knew the 1.0 cards were just defective.

That's basically it.

What I suggest is contact Tom Ho (tomh AT MSI DOT COM) and request VERY firmly and plainly if you can get a RMA replacement purchase of a GTX 1070 revision 1.2 (i don't know about revision 1.1 or if that is the highest revision for the barebones) card for the MSI 16L13. You need to explain VERY plainly that you have the defective 1.0 card with the broken 1.013v crash voltage step and you want a fixed card shipped from MSI, and see if there is a way to get a cross shipment under warranty, without having to send in the laptop. If Tom cannot help you, then you need to CALL MSI on the phone and ask for an escalated supervisor request for an RMA upgrade for a defective part. See if you can work out a way to get the new card shipped to you without having to send the entire laptop back, and then you ship the old card back, to avoid any downtime. This is assuming you know how to re-paste with thermal paste and repad with Arctic 1mm thermal pads.

If you are uncomfortable with that, then you would have to ship the laptop back to MSI, and God knows how long that will take. I've seen MSI completely butcher people's laptops, so I would fight VERY VERY hard to get the card itself shipped to you and set up some sort of cross shipment. It might be VERY hard but good luck.

I've attached the "8a" Vbios here. I DO NOT know if this will work on the 16L13. This is a GSYNC vBios, and I do not know if you have a gsync card or if this bios is whitelisted for your laptop. This vbios works for GT62VR's, GT72VR's and GT73VR's with 1070's and gsync, but not for non gsync boards. I don't think a "0.881v" vbios exists for non gsync models.

If this doesn't work, you have to lock the voltage like i suggested.
But I think you should just get that card RMA *UPGRADED* at MSI's cost (minus shippping). But you're going to have to be VERY, VERY stern and pushy. They can do it if they want to. But they usually do NOT want to. It's too much trouble for them because it breaks their "system" and makes them have to admit that "they messed up."

 

That's honestly pretty pathetic for a massive company to do. But the world's F**ked so there isn't much we can do. But thank you for the explanation. I kinda figured it had something to do with the voltage but I didn't know afterburner would give me that level of control. I always tried to downclock my card thinking it would help.

Wow, thank you so much for the information. I do have revision 1.0 in my 16L13, so I will certainly give this a try with the email to Tom first and then phone call if it doesn't work. I would love to have my beast functioning 100% again with no issues.

I also have the Eurocom Sky x4e2, but that 1070 failed on me with in months of owning it and I never bothered to rma it as I have had my Tornado F5 since the gpu failed in the Sky.

 

Hopefully, that was still under warranty and @Eurocom Support can take care of it for you. If it is under warranty, you should do it before it expires.

 

Any chance you can take a picture of the 1070 card without removing the heatsink? (unless you know how to re-paste and can repaste without an inconvenience).

I want to see the form factor. But if the heatsink is blocking too much.....need to see the heat plate position relative to the MXM slot and see if that "tab" is on the card (the infamous 1070 tab which made the MSI 1070 (at least the GT73VR version) hard to fit in other laptops even though it would work in a Clevo P870 with some modification.

 

This is the model I have of the gpu. But it doesn't have the SLI connector.

 

Unfortunately I no longer have warranty on this model as both laptops are no longer under warranty at this point. My tornado F5 just ran out of warranty in January this year.

 

Man, I'm sorry. That is a bummer.

 

Ok that's the exact same 1070 used in the GT73VR.
Try to contact MSI. Go through Tom first.
Make sure he understands that you know PRECISELY the problem and that the vbios band aid fix is not acceptable.

If that fails, go through MSI support. You're going to have to fight the battle of the bulge for this.
Let them know that you want a REVISION 1.2 card that is compatible with the GT73VR and 16L13. Make sure to specify whether your laptop has gsync or not. I don't know anything about whether a non gsync vbios can be used on a gsync card or if your laptop was not designed for gsync originally, what would happen if you installed a Gsync card (most likely gsync would simply not work and it's business as usual. I think the PROBLEMS happen when the card itself is a non gsync revision and you put a gsync vbios on it).

Good luck. I want to hear some good news.

 

So I realised at work we have a reflow and reball automated machine.

On my sky X4 I reflowed the 1070 and it has been running Watch Dogs 2 completely maxed out ultra for 6 hours strait no crashes no artifacts or clock fluctuations. It seems as if I just fixed one of my beasts.

I'll be shooting an email shortly to Tom at msi as Falkentyne recommended to get the 1070 replaced in my 16L13

I will let you both know as soon as I get a reply! Thank you both so much for your help. And yes that is correct a G Sync card will not function on a 16L13 that did not come with a G-sync display because they flash the motherboard with a G-sync enabled bios.

 

@Izen What card did you have to re-flow and why?

I believe the gsync card will function fine, but gsync will simply not be available.
The problems were when users (this was discussed on the Clevo section) were flashing a Gsync bios into a non Gsync card (there are differences in the PCI ID and probably some Ngreedia resistors that determine if a card supports a feature or not), even though the CHIP is the same. It's all DRM, folks. Gsync= Adaptive Sync + DRM.
Free sync= Adaptive Sync+ FREEDOM.

FRREEDOOM !!!!!!!!!!!!!!!!!!!!!!!

 

I have a Eurocom Sky X4E2 with a GTX 1070 (The clevo one that requires external power and is a massive card) The GPU failed 3 months into owning it but I received my tornado F5 shortly after it happened and didn't bother touching my Sky X4 and honestly forgot about it sitting in my closet. The GPU was giving me lines and crashing constantly on the screen. I just started working at my job 2 months ago now and discovered they have a reflow/reball machine and asked around and was given permission to use it on my break. Took the GPU with me to work and reflowed it and it has been working fine for 8 hours now without a single issue. Benchmarks are where they should be, temps are good, clocks are stable and even with a 6gb Vram load it handles it fine.