** 1070 laptop: GT73VR, GT62VR, GT72VR reboot/crash problem **

Gsync card?
Possibly. Try it.

 

If you are not successful in flashing the VBIOS you can download msi afterburner and open the voltage curve (ctrl+f) then flatline it beyond 0.9V. If that works you will know your card is suffering from the same issue as in this thread.

 

Yes Its a G-Sync Card.
Now my issue upgraded to new level. Even if i dont run any applications/games. Just fresh boot leave the desktop. Not even play video. For 15-20 min the clock is at 139 Mhz , 34-37 degree centigrade. then suddenly its shooting 1850 Mzh + and 80-90 degree centigrade.
any idea what caused this issue.
i could use all the help i can get. pls

 

Now my issue upgraded to new level. Even if i dont run any applications/games. Just fresh boot leave the desktop. Not even play video. For 15-20 min the clock is at 139 Mhz , 34-37 degree centigrade. then suddenly its shooting 1850 Mzh + and 80-90 degree centigrade.
any idea what caused this issue.
i could use all the help i can get. pls

 

I have MSI afterBurner. and i am able to open voltage curve (ctrl+f) but any guidance how to edit. any link to video or any demo ?

 

Impossible to be 80-90C on the desktop on a 115W video card.
Unless you are mining or something or maybe you have a virus that's putting the graphics under load.
Try repasting with a new thermal paste. Coolermaster Gel Maker Nano works well.

 

Starts to sound like a virus or something to me as well.... my 1070 would never even reach those temperatures under full load.

Just make the curve flat beyond 0.9V as I told you. But honestly your problem is at least partially totally different so not sure it would even help. You need to figure out your windows/rogue program/virus issue first before troubleshooting further.

 

Sounds to me like a miner is rubbing in background, you don't have a browser tab open do you?

Sent from my LLD-L31 using Tapatalk

 

Hi,
I dont Mine, and as part of resolving the issue, i have even recently Factory restored my laptop ( I am not using a desktop). All data on HDD and SSD are formatted. No chance of virus as well, i have been using and Internet Security antivirus program till before i factory resetted my Lappy. And may be i will be taking to a service center as last option as they will try to LOOT money. and If this is a software/Driver/bios issue which i could figure and fix myself i dont want to waste money to service centres. (any way i am not doing any advance tuning/tampering or something

 

No, I dont Mine, and as part of resolving the issue, i have even recently Factory restored my laptop ( I am not using a desktop). All data on HDD and SSD are formatted.
I just have basic softwares installed.

 

You might not have installed a miner on your system but a webpage that has been hijacked can run one remotely which is normally coinhive which uses both GPU and CPU. Look at my signature there is a thread of such as site doing what I am talking about.

Sent from my LLD-L31 using Tapatalk

 

Actually i am not able to figure why my AfterBurner is not showing voltage (in mV) in the monitering. its showing '0'. i turned on check box - "unlock voltage monitoring". So Frequency is in Y axis and Voltage in X axis.
So i do it flat horizontally or vertically ?

 

I checked and i dont have any browser running at that time. I checked the Task manager. I will check for any suspicious threads.
But in this time when my GPU hypers up. My CPU is perfectly stable. (monitoring through MSi after burner)
i will try With AirPlane mode ON. Do you have any ideas i can test or mining/virus.

 

You could run malware bytes etc.

Sent from my LLD-L31 using Tapatalk

 

Also, not sure how you install drivers and what drivers you are on. Try to install a newer/older version after cleaning with DDU first to see if you can get rid of the pop up errors.
It is normal for it not to monitor voltage by default. The curve will work regardless of whether monitoring is enabled or not. Look at the pic below for inspiration: but don't OC like I did - the pic is from my gt75 (your curve will be lower). I don't have the msi laptop anymore and don't feel like installing afterburner on the pc just to show you how it should look like. Just flatline it beyond 0.9V and don't drag any points before it upwards (yet).
But this will not solve your gpu usage spikes. If DDU-ing doesn't help, there must be some program going rogue despite your efforts - maybe something you normally use and would never suspect, something preinstalled by msi (I had some issues with some versions of SCM and dragon center related to high gpu usage), who knows. You can try looking in task manager. When you open task manager and click show more details, right click any of the top headers in processes (default should be cpu, memory, disk, network) - and also select gpu. If that doesn't reveal the culprit - I suggest to open nvidia control panel and on top click desktop -> display GPU activity icon. That way you will have a colorful icon in notification area and if you click that you see exactly which programs are using the gpu at that moment. When the gpu usage goes rogue, simply terminate them one by one in task manager (even SCM) to discover the culprit. You will know when you find the culprit cause the frequency will go back to idle. You can take it then from there.
PS. I would only mess with the curves etc once you get the drivers/rogue programs sorted out.

 

Hey bro,
I actually downgraded windows to 1607 build, which didnt have GPU monitering. And after hard trying not to damage GPU in process of upgrading to 1803 it took a while.
Now i am not facing any hyperups on non using of laptop. And the GPU is perfectly fine now. Still need to test few things.

 

And i also tried to set the volt-freq curve to control my GPU. but it only shifted the base and OC clock indicaters in MSI after burner indicater to lesser values than 1400 Mhz. but still GPU is running 1800+Mhz. (ps this is before upgarding to 1803 windows build) still testing new build

 

Could my laptop have this? I have a GE72MVR 7RG laptop, not the same as in OP, however sometime mine boosts to over 1900, I feel there is instability, games crash now and then, anyway to check if my GTX1070 is some sort of batch that is affected?

 

Well, if only the game crashes you do not have the same problem. It seems more like there is some sort of OC applied to your card causing the instability as I don't think it should be going over 1900 by default. In fact in demanding games using the gpu to 100% it should be around 1600MHz. I recommend to reinstall dragon center and SCM. Also reinstall nvidia drivers using ddu. If that doesn't help try reflashing the VBIOS (make sure to reboot after it's done) - just roll with the official one on the support page for the laptop for now. Then you can take it from there.
Anyway I don't think your gpu should be affected as this was specifically tied to the 1.0 revision MXM cards and yours is soldered to the motherboard. But ye, we can't know for sure what msi was up to when assembling these laptops.

 

Hello,
Just recently got the GT72VR (regret not knowing there are problems with this laptop, otherwise would've avoided it like the plague). First problem was the crashing issue everyone's been having here, but flashed the 8A VBIOS and everything's fine again. But suddenly I got another issue of blinking orange LED's and my laptop won't charge. Tried flashing new BIOS but that didn't help, as well as doing EC reset like 10x. If anyone could help me that would be appreciated. Cheers!

 

I hope for you that it is in warranty.... >< Anyway, if you look at the power adapter does the light go out at all when this happens? Is the cable loose when plugged into the laptop? You could try opening it up and seeing if there are any burnt spots on the connector itself or around it on the motherboard...
My cable was nice and snug on the gt72 but some people did not which resulted in the cable being shifted, not making perfect contact and lots of heat build up around it.... Not sure if it was also this model specifically but I do remember someone definitely having it on the gt72s and since it's the same mb....
The adapter itself could be dead too. Is the laptop new or second hand?

 

Hey there, thanks for swift response. Let me give further details:
-Got it used on eBay, 3 days ago. Looks new as hell despite used and got it ****ing cheap (1600 AUD).
-This problem didn't happen a few days ago, but could've happened after I flashed the vBIOS (but can't confirm this 100%)
- Adapter works fine, has green LED (also the adapter MUST work, because if I plug it the laptop IMMEDIATELY shuts down)
- Honestly don't give a crap if it's under warranty, I am in Australia and the warranty process could take weeks....would rather take it to a reputable 3rd party warranty shop.

 

Well... if (IF) worst case scenario the motherboard is faulty... you might have just thrown away that "cheap" 1600AUD. You already know the gpu is most likely faulty (ye it will probably work with 8A vbios) and the mb is a possible culprit at this point too... not much left of value aside from ssd's and ram which you can resell. Maybe screen. So yeah, I would inquire whether it is under warranty as if the mb were to be faulty, a repair can quickly run into hundreds of euro (or whatever currency) or even over thousand depending whether the whole thing needs to be replaced...Might well be more than what you spent on it.

If you are lucky the battery is the thing that is faulty. Your best bet is to open the laptop up, unplug the battery and try to run it without it and see if it behaves normally. You will lose some performance doing this though so it is in no way a solution. Well, at that point the mb can still be the culprit causing the battery to stop charging...

PS. you said you reset the EC. Have you tried reflashing it? Also, reflashing new bios sometimes doesn't reset everything. Unplug the laptop and hold the power button down for like 60 seconds to do a cmos reset (should work as it's the gt series).

 

Honestly, at this stage I'm kinda pessimistic...hopefully its just a battery problem but idk. It IS secondhand after all. And yes, I reflashed EC firmware, BIOS and did a CMOS reset. None worked.

 

So, I've been sending my Gt73vr to the warranty... They have changed the 1070 two times,but they have installed version 1.0 everytime.. What can I do? Can I ask for my money back? I've been without my laptop for two months already

 

Demand a revision 1.2 card.
Explain CLEARLY that he 1.0 cards have a bug.
Some of those people seem to have trouble understanding English. So you need to *explain* it to them.

 

Believe I've been trying to do that... I will call again on Monday. The problem is that I call to the service center in Spain and they send it to Portugal...I think I'm going to ask for my money back. Only problem, in case that i get it, is that nowadays I can only buy the gt75 with the 8750h. The i9 is 4000€...way more than the 2500€ i paid. I will keep you updated

 

This is the best solution
If you have an UNLOCKED processor, ask for the i9 8950HK.
Even it it comes with a 1070 instead of a 1080.
Trust me
Talk to them
Negotiate something.
Ask to pay the difference. At least you will get something which works.
You need to be assertive and stand your ground while also showing firm respect.
And be persistent.

 

I had the 8750H in the gt75 and trust me that little cpu was insanely better than the 6820HK I had in my gt72vr. In the gt75 it is not TDP limited so it performs nothing like it would in any other laptop. It ran 3.9GHz stable at all cores, with some undervolt it would remain around 80C even on stock paste running prime95 while pulling 90W or a little more constantly without droppping. I did not remove the iccmax stock limit though so it "only" ran at 3.7GHz on all cores with prime95, but this is something you can change easily and it will probably pull over 100W then, but you do need to repaste at that point if you wanna avoid approaching 90C.
Anyway, the cinebench score was over 1200. You can test your current cpu OC'ed to the max and see if you can even get close to that - I don't think so. Lemme assure you that the 8750H will be absolutely stunning even stock unless you need high single core OC or you get OCD-tendencies about not having enough control. But just looking at pure performance, I would take the 8750H over the 7th gen whatever (even desktop equivalents like 7700k just purely due to the extra cores).

 

So, talked to them again and they are taking it again tomorrow ... 4th time and two months already without my laptop. If they don't get it right this time, I'll ask for my money back.

Forgot to mention, which vbios should I install?

 

Just got my laptop back from msi, after having sent it 4 times during three months, but they finally did install the 1070 v1.2. I have to do some tests to see how well it Works, but I suppose it Will finally be fine.

Forgot to mention, which vbios should I install now?

 

No need to change the vbios. Just hope it's 'not 8A.

3A and 44 (last digits) are the good ones but I don't know how many other versions there are.

 

I'll check later which one I have. By the way, I'm trying to unlock the bios but when I'm on RU I'm not able to switch to the right page in the setup menu. I press control down and nothings happens, it stays on the same page... Any idea why? I may ask in the other post about the bios unlocking

 

I forgot. Arent you supposed to use page up/page down or control or shift+ page up/down?
Or you can just use the hotkey combination posted in the Laptops with Mux switch thread.
I'm too sleepy.

 

Yes, but it wasn´t working. Anyway, for my use, that key combination covers my purpose. Btw, I´ve just run a few fire strike tests...I´m not able to much my higher score and I don´t know why. Forunatedly, my clocks can go higher without the bug. I´ve been able to achieve 15266...but before I did achieved 15572...why is this happening when I have higher clocks now?
https://www.3dmark.com/fs/17153518
https://www.3dmark.com/fs/17153381
https://www.3dmark.com/fs/13965137

 

As far as I know, the 1.2 cards do have slightly less performance (several people noted that before you), but they are stable. So I wouldn't worry and just enjoy the laptop.
Higher clock does not necessarily mean higher performance... Anyway 300 points is not that high of a difference - kinda within margin of error imo, I assume you won't die from 1 fps difference or however little it is. It seemed to me as if either the 1.0 revision were bad batch of gpu's (unlikely) OR bad mxm pcbs (very likely to suffer from voltage or power delivery issues). There appear to be all sorts of voltage/power issues on these laptops as evidenced by the problems with bd prochot on the gt72vr ones... msi messed up somewhere on the design or got a batch of faulty voltage/power parts (whatever those were but only msi will ever know - if they cared enough at least).

 

It was the 1.013v voltage point that was broken on the 1.0 cards, but no one knows why it was broken. It was something on the card.

 

So it wasn't even the highest voltage point? That is totally messed up... it's like a mystery no one will ever solve. I am pretty positive the graphics chips themselves were good (the actual nvidia ones before msi soldered them onto their mxm modules).... I am so happy I have a pc now. Yay for my gpu manufacturer answering questions within 24 hours, accepting the card for rma due to coil whine without any hassle and a 1 week turnover (card shipped monday, replacement will arrive monday)... oh and did I forget the integrated gpu in the 8700k is pretty ok to use meanwhile? Or the 10 year warranty on my 1080ti? It's so hilariously easy to pull the nvidia one out, ship it off, have 0 driver issues even after windows loads the intel ones automatically... Why is it so easy on a pc? Why can't they make easily serviceable laptops? It is possible after all, especially if you were to open an eshop with proper spare parts...

 

Anyway, now that I can go farther than 1.013v, is there a way to lock the gpu clock? I've tried to lock it using msi afterburner, but the truth is that as soon as the gpu starts working, the clock lowers...

 

I don't know if it was 1.013 to 1.062v.
I only know that 1.013v crashed as soon as it reached it.
I will assume everything past 1.013v was busted too.

 

Nope. There is a 115W power limit. Almost everything will exceed it making it downclock.
You should be able to TDP mod that GPU to 151W. Then if you lock the clocks to exactly 1860 mhz, you will never downclock.
But this requires special hardware:

Skypro programmer
1.8v adapter.
Pomona 5250 IC clip
Male to female jumper cables (to connect the Pomona clip to the 1.8v adapter).

 

Another factor to keep in mind are temperatures - not sure how warm you run the gpu or how the msi gpu actually behaves when trying to lock the clock.
Anyway yeah, stock on 100% load it will stay around 1600MHz, even dipping into 1500's shortly due to power limit. These cards are badly limited... That is why I liked the 1080 in the gt75 as it pulled around 220W max, making it about as powerful as its stock desktop counterparts, easily running over 1800MHz all the time (in fact mine downclocked due to temperature in the msi laptop before reaching power limit most of the time during gaming). I swear that 1080 was just sweet in comparison to the crippled gtx 1070.

 

The pascal "throttle" you are talking about is the -13 mhz every 6C temp increase, starting at 38C.
In my experience, if you lock a lower clock speed than the maximum (you do this by locking a lower voltage step manually example, 0.881v, which is what the 8a vbios limits you to already), and you have a TDP mod (>150W) in place, the card will maintain those clocks without dropping. If it does drop, it will only drop 13 mhz once.

 

Yep I noticed my evga 1080ti behaved the same way ^^ Totally normal. Might watercool it in the future to get awesome temps and stuff which is unfortunately something impossible to do in a laptop... Was thinking of investing in one of the ekwb watercooling systems in a few years to push the hw to its limits before replacing it. So far my watercooling experience with a corsair aio was bad (it leaked lol, thank god nothing was hit as I was fast seeing the tube coming off the pump - should never happen....). To be fair, I did find quite a few aio's leaking and ruining entire systems on youtube afterwards... nothing as dramatic happening in the custom loop department

 

Ok, that's far beyond my knowledge. Is it easy to do? Did anyone do it with good results? How much of an improvement can I get? Thank you very much !

 

In fact Falkentyne did it to his gt73vr and it seems to be running well. You might need a bigger power adapter probably (330W) to avoid any potential issues. But it would un-cripple your 1070 and make it actually close to the desktop one. When you look at official nvidia specs, a 1070 is supposed to draw up to 150W even in the founders edition version official specs. Yours pulls max 120W now. For comparison, a founders edition 1080 pulls max 180W by default while the one in the msi laptop goes to 220W. That is why I said the 1080 versions are sweet while the 1070 ones are crippled. They are below nvidia official specs, while the 1080 one surpasses nvidia official specs. What Falkentyne did was to essentially bring the 1070 up to spec/little above the official nvidia specs, which is how it should have been to begin with. Honestly, imo it is not necessary to do this at all if you are happy with the performance as it is now. I am pretty sure that for 90% of the people out there the risks would outweigh the potential gains (I know I would mess something up for sure).

 

Thank you! I´ve been reading a lot and I will read trhough the pascal tpt post. After that I will ask falkentyne, but the mod looks promising

 

If he has a Titan laptop, he can change the value in EC RAM register E3 from 10 to 11 (or just increase the value in there by 1) with RW Everything, and that will change the power ID to 330W.
This also will stop all battery boost drain with a stock 1070 as well. Then he can just mod it to 151W-170W, then MSI afterburner TDP 100% will be 151W and 112% will be 170W.
The power ID has to be changed every windows boot. Failure to change the ID to 330W will cause >10% per hour battery drain if you get close to 230W total system usage, and will power throttle the CPU if you exceed 230W.

Buying the 330W power adapter isn't a bad choice. The 230W power supply can handle 151W from the GPU, but it will run quite hot if the entire system is at full load and overclocked. 170W will risk tripping the adapter, as it seems to trip if more than 280W is pulled from the wall (>245W to the entire system).

for people with GT72VR's trying the GTX 1070 TDP mod, it's almost not worth it. I don't know if a 330W power ID is 'hidden' in the EC, but you can test that without doing the mod, by opening the EC RAM in RW Everything, increasing the power ID in register "E3" by 1, then try playing a 3D game with prime95 running. If you get >10% battery drain an hour, then you have a 230W power cap. If you get no battery drain, then you have the 330W power ID, even though no 330W PSU exists for that power plug.

Then GT72VR users should be safe going up to 150W, but not higher.

 

He said a few pages back he has a gt73vr. I'd still rather invest in the 330w power supply just simply because a 230w one has a higher chance of suddenly dying with some melted plastic if it gets too hot for extended periods of time. I don't really understand why they chose the smaller adapter anyway if there is a 330w one available. It can't even be a power saving trend as the adapter will only pull as much as is demanded by the system... Is it just a money saving strategy for the manufacturer?

 

Thank you! I have a MSI Gt73vr 6820hk and a 1070. I´m reading the other post, but i´ve only read 30 pages, lots to finish. Anyway, looks like you were able to surpase the 230w limit, but having to do it on every boot...I´ll hate that, anyway, if its worth...How much performance boost is possible with fans on auto , having decent noise values and temperatures? I don´t want to have a jet at home, hahaha.