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

It's a gruelling process man. I freak out every time I have to remove that bottom lid haha. Btw in the picture below (found on the internet), "Device ID" is a unique identifier?

 

Nope, it is not. Looks like its just a model number of sorts.

Source:
https://forums.laptopvideo2go.com/topic/31616-geforce-36879-msi-mobile/

Can anyone confirm?

 

It's card type (+subtype) and vendor, like my NVidia GTX680M from MSI is 10DE 11A0 1462 10BC (MSI is that 1462) (and Nvidia is 10DE).

 

Yeah you're right. On the other hand, I found this.
https://superuser.com/questions/469...rial-number-of-a-gpu-without-opening-the-case

Is this legit?

 

That just adds to my suspicions that it is a software/firmware fault since my problems were also solved with reinstalling vbios (0d version), bios and windows...

I have the samsung memory and no issues anymore. Before my card would clock up too high and get unstable, now despite the same vbios it is OK.

It is not summer or temperatures. My room temp is the same as before yet it doesn't go over 1900 or 2000 anymore (about 25-26 degrees celsius). Plus the laptop never goes above 75 degrees celsius on GPU (100% load) without coolerbooster (with it it's about 65-66 max). Any throttling happens above 85 or even 90 degrees I believe, but definitely not below 80. Plus if it were temperatures, then it would still boost like crazy shortly when you start putting load on it but that does not happen. IMHO it is magic how the clock speeds are regulated. It's just magic. And I do love how the cpu sits at 34 degrees when room temp is 26 lol (just browsing the internet... with 12 tabs ;x).

 

Try uninstalling dragon centre might help.

Sent from my SHIELD Tablet K1 using Tapatalk

 

Starcraft 2, instant crash.


Sent from my iPhone using Tapatalk

 

Oh, you've got a 1070?

Yeah, well, wasn't your laptop made in 2017? All the 2017 laptops 1070's are problem free.

Or, do you have a 2016 model?

 

October 2016.

 

Ewwww, well read up here, there's stuff to try... or RMA back to MSI for a new unit with a new 1070.

Since you are only getting this in Starcraft 2, and that POS is crashing for everyone, I still wouldn't assume you have a bad 1070...try some of the games others here are crashing in.

 

No other game crashes. Overwatch, Diablo 3, Heroes of the Storm, Doom 2016.

Running Heaven in basic mode to push high gpu clock. It is clocking at 1.83ghz right now in full screen mode. Gonna let it run and see if it blows. If it does, I am returning the laptop.

 

Darn, I hope it runs ok for you.

See if you can swap it for a 2017 model...let them know about the generic 1070 issues from 2016 models, and see if that get's you a bump up to a new one from 2017.

Where did you order from?

 

Reached 1.86ghz just now.

 

Microcenter. If I do return this laptop I will just get a Kaby Lake with GTX 1060.

 

Microcenter Store Managers are pretty cool... let him/her know of all your wasted time, finally arriving at the 1070 2016 glitch - and you might parlay this into a Kabylake 1070... or at least some other goodies - discounts, etc.

It's worth a shot

 

Called USA MSI. The guy on the phone acknowledged the issue and told me to return the laptop, hardware defect on 2016 GTX 1070 hahahahahahha.

They did not even try to argue. He knew about the issue. He knew about the chinese vbios. He told me to just return it.


Sent from my iPhone using Tapatalk

 

Cool, now you have an "official" story from MSI to tell Microcenter, and a good reason for them to bump you up to the 2017 Kabylake 1070 model

 

Is the correct spec 1900mhz? Hmmm I noticed Heaven shows max clock speed as 1911?? Is this why the 2016 cards are crashing? I flashed the chinese vbios now and Heaven displays 1645 as max.


Sent from my iPhone using Tapatalk

 

Yes, it's a Boost issue with that 1070 - not sure if it's the GPU die or the support parts that are failing.

There's supposed to be a beta vBIOS that brings that 1645 up to 18xx, under the failing point.

 

Oh where is that?


Sent from my iPhone using Tapatalk

 

Read the thread, someone put it up on a sharing site... maybe it's still there..

 

He's already using it, @hmscott . He's talking about the "chinese" Bios (it's not even chinese. It's simply "8A").

8A has automatic overclocking disabled (aka boost 2 clocks). Boost 1 clocks perform like normal.
I'm using it for the modded Pascal TDP because I can keep the clocks rock steady without temperature throttling at 1886 mhz (+250). The normal Bios will go higher but will temp throttle the boost 2 clocks starting at 45(?) C and this apparently can't be removed (yet).

 

So he has the right vBIOS, and only needs a +250 GPU core OC?

I thought I read it was a different vbios...

 

Uh.... shouldn't this crash? Are we sure this is a hardware issue and not just SUPER ****TY MSI Firmware code not handling some edge cases from the 1.0 revision of gtx 1070? Heaven was hitting 1911 MHZ for a while and when I took screenshot, clock went to 1822 MHZ. But regardless, it was hitting 1911 without ANY issues. This is using vbios 8A btw.

 

Did you change the vbios yet, or are you on the same vbios it came with?

It's possible you don't have the bad 1070, it's just POS SC2 crashing as per typical.

 

No, mine has issues. With the MSI bios, my **** crashes on heaven running on LOW because it goes pass 1900 MHZ. With the chinese BIOS, I can overclock it to 1911 NO ISSUE LOL.

 

Ok, so nothing changes, you are taking it back, right?

 

Not today even thouhh it is the last day. I will call microcenter to let me return it tomorrow.


Sent from my iPhone using Tapatalk

 

If today is your last day to return it based on your purchase date, I'd drive it over now, and not take any chances... they are cool, but they have to follow the return rules too.

 

I cannot go, got kids with me and the mrs. is out. I will try and explain that MSI mention it is a hardware issue.


Sent from my iPhone using Tapatalk

 

Worst case scenario I RMA lol. Got my XPS that is able to play all my games in low setting lol.


Sent from my iPhone using Tapatalk

 

I tried to explain this before but I honestly dont know if people just ignore me or what.
There are three basic clocks: base clocks, boost clocks and boost 2 clocks.
Base clocks are just that. e.g. 1443 mhz default, with +200 that's 1663 mhz (the steps are in +13).
There are 2 types of clocks beyond this:

boost clocks and boost 2 clocks.
Boost clocks are a fixed overclocking clock that is limited by TDP rating. If the card exceeds TDP, the power limit flag gets turned on (blue), and the card will throttle the boost clock down until it reaches TDP.
If TDP is never reached, boost clocks would always be at the maximum clocks.
I do NOT know if temperature affects the boost 1 clocks.

Boost 2 clocks are an automatic overclocking clock that, if not restricted below boost 1 max clocks from TDP, are limited by *TEMPERATURE* and *VOLTAGE*.
Boost 2 clocks I believe will start throttling at 42C and drop 13 mhz every 5C, and will try to rise up to the voltage limit for that clock. When the max clock is reached for the currente voltage, the VREL flag (voltage reliability) flag gets turned on, showing that the clocks cannot go any higher. The "temperature" flag will only occur (this throttles all clocks) on battery power, or 91C. The maximum boost 2 clock it will try to attain depends on the temperature step, and the voltage for that clock attempt. You should almost always have a VREL flag if in boost 2 clocks.

The 8A Bios disables boost 2 clocks, so no automatic overclocking will occur past the boost clocks. If automatic overclocking DOES occur, it means you flashed the 8A Bios and did NOT shut off the computer (meaning; AC power off, no fast shutdown).

The "bad" cards seem to have problems with a stable voltage in boost 2 mode. If this is not because of the "micron" vs "samsung" Ram issue, this is because of some bad part on the card itself, which must be replaced with a card with a good part. This cannot be fixed with a Bios, only limited.

The "good" cards will show VREL in boost 2 clocks and not try to run at an unstable voltage or clock (Unless you try to push the base and boost clocks too far).

 

I get that. Looks like there are tons of people using 8A by looking at the MSI forums since last year. I guess these people do not mind their cards not going to 1900 (boost 2).

 

My card never reaches 1900 anymore unless I overclock it in dragon center. With default settings (sport mode) it reaches about 1840-1860 or so and stops there because of VREL. I do have laptop made in late 2016, but not sure which revision GPU since my model is also a limited edition with HK processor (it's a GT72VR, not 73)... and I am not opening it up. It just works. I still think it is a firmware fault since others get their cards over 1900 MHz stable on 8A vbios overclocked, while they crash on official one even below 1900). Also my boost speeds were corrected with a reinstall (or just magic and praying did that??). I just wish MSI would finally dedicate some time to fixing this as there must be a way to do that...

 

It is not the clock killing it. It is the weird voltage on boost 2. Over cloaking past 1900mhz will not crash it on 8a because the voltage is stable. Also the gpu cannot stay more than a few secs in 1900mhz with official vbios.


Sent from my iPhone using Tapatalk

 

Been playing with 8A... looks like a +200 is the best on this laptop pushing it to 1835mhz. No freezing, stuttering, smooth af. The gpu is able to stay at that clock speed without throttling.


Sent from my iPhone using Tapatalk

 

No more crashing in SC2?

 

No crash in sc2 with this vbios at 1835 mhz LMAO. This is crazy.

 

Others that posted here tried the 8A vBIOS and still got crashes in games / benchmarks, and sent in their laptops for new 1070's or returned them for newly built laptops.

@Falkentyne are you suggesting they could have gotten them to work with some tuning and didn't need to RMA / return their 1070's?

What's different in the 2017 1070's ? Just the 8A vbios?

 

There is no 8A vbios for the new 2017 cards... it's not needed according to what others say. And it is not official for any model as far as I know. The 8A vbios is one someone pulled off of their laptop in China after sending it in with RMA to MSI. It just limits the card to the basic specs given by nvidia, disabling further boost based on voltage/power limit. That's all. It is not specific of any single card, it is a firmware that is not official and can be installed on gt72vr, gt62vr or gt73vr with gtx1070. It can be overclocked to reach higher speeds. Which you would probably know if you read this thread.

I would still not settle for a laptop that does not work as promised right out of the box, end of story. All that could have been said about this issue has pretty much been posted on this thread. The only thing you can get by posting more is getting out some frustration. Just return the thing if you aren't happy with it, the end. Or RMA it and get it back with limiting VBIOS anyway. Then argue for a new GPU. That is all there can be done if no amount of reinstalling/updating helps. Good luck.

 

Thanks for the reiteration of what we all know by reading the thread

The difference now is that @plee82 / @Falkentyne are seeing full clocks with OC, and aren't crashing any longer.

So, again, what's new with the 2017 1070 that fixes the "problem"?

Or, another way to say it is, what's the problem?

If they are getting full performance (benchmarks ok?) with their 1070 with 8A vbios + OC + no crashing, what's wrong with that?

I agree, I'd take it back and get a new 2017 1070 model like MSI recommended

 

Well... no one knows what the problems is, which is the problem. Even MSI doesn't know. It seems some older cards do not obey or recognize voltage limits correctly on the official VBIOS. No one knows why this happens. Going over power/voltage limits results in reboots. The card should by default detect voltage/power/temperature limits and adjust the speed according to it. Some cards are not capable of doing this. Whether it is firmware/software or hw fault is not clear. And I can guarantee you will not solve it. Sometimes reinstall of windows/vbios etc helps. Chinese VBIOS seems more stable for some people. No one knows why. You will NOT get an answer from this forum about what the problem is since not even MSI can sort it out obviously.

 

Just tried Doom and the stability at 1835mhz is amazing. I remember with the 0D vbios, the clock would not stay at that speed constantly.


Sent from my iPhone using Tapatalk

 

Hmmm lower temps too. This is probably the implicit gpu undervolt that happened when boost 2 is disabled and we overclock the gpu. The voltage stays the same but the clock speed is higher lol.


Sent from my iPhone using Tapatalk

 

I have a gt73vr with the crash issue. Am trying to decide to keep it or not. Has anyone got msi to install a new card via warranty or only the vbios bandaid?

 

New gtx 1070. Let me just say to just flash the 8A vbios. Boost 2 is useless in most situations. Clocks shoots up, voltage shoots up and throttles right away lol. It is good for your max value in hwinfo only. I do not think it stays there even for a full second and your fps will fluctuates when the throttle happens. With 8A and +200/+300 you get constant clock speeds that are much higher and stable than 0D or 3A. I should have just kept my old gtx 1070.


Sent from my iPhone using Tapatalk

 

This. The "unofficial" 8A vBios gives me the highest scores and best fps after a manual overclock. The fps is also more stable and constant because the clock speeds fluctuate very little. I never had the crashing issue on my version 1.0 Tornado F5 1070 but tested the 8A vbios and found it to be a better overclocked with more stable clock speeds.

 

Hi, i have too an GT72FR 6RE Dominator Pro with 1070 from france.
and i have some drop fps, on overwatch for the moment.

i have an K1608N too.

But i bought it on november, so i cant have a refund i think, but do you know if they can change the video card or repair it ?

Thanks

 

I think my MSI 1070's I've put in my clevo suffer from this, the way around is to use a MSI afterburner profile that locks the frequency/voltage changes of Turboboost 3.0, and just leaves the temperature-related clock reduction. 1950mhz/0.95V is stable and seems a good mix of speed, temps, and using the available 150W power limit with my modded vBIOS. The only problem is that the cards idle in the 50's using over 40W each when locked at that speed, leading to higher fan noise (and even worse battery life than I already have in a 17" stuck at full screen brightness, an XM and SLI GPUs lol); so I use hotkeys to switch back to the normal curve so they clock down at desktop