Clevo P775TM1 GPU started crashing

So, about month ago I got my P775TM1 with GTX 1080 and i7 8700K. Everything was smooth. Games ran fine, and then I even undervolted GPU. Temps still hit 90 on demanding stuff, but it was working stable. At least until this week.

Now issues started. I noticed Stalker (yes, that 2007 game) stopped running stable and started crashing. I started testing stuff. I noticed that PUBG crashes as soon as temps hit 81 degrees (in one case it even did at 74). So I decided to try with no undervolt. And... Stalker decided to crash the entire laptop at 90 degrees. PUBG didn't, strangely.

I tried reinstalling drivers for GPU (including old driver), and for chipset, however it gave me no result.

Is there anything I could do? I can request warranty, however I am in different country than the shop where I bought it from, so it'll be a hassle. I would like to try to prevent the issue on my own if I can.
I have more testing to do, but I have no time to do so today anymore - and I decided to attempt looking for help asap, as during this week a family member should be available to help me with moving laptop to home country.

Thanks in advance, and sorry in case it happens to be some (really strange) false positive, but I prefer to make sure of possibilities fast to reduce potential courier costs.

 

Try downclocking the CPU and GPU separately to see if you get more of a reaction from one or the other.

 

Oh for crying out loud RMA it. Do not try to tinker with it, you'll waste more time in the end. Running components on high heat as you do will damage stuff in the end, so better get parts swapped out before trying to fix it and end with a chunk of metal right after the warranty.

 

I believe you should try running games well optimized like BF1, csgo, dota, and if*still* , this happens, then RMA is your last option.
Note pubg isnt well optimized, cant say anything abt stalker!

 

Clevo 1080 failures are suprisingly often - this is the third damaged 1080 I heard about since Pascals release. Fortunately all under warranty period.
Funny fact - instead of allow fans to running at 100%, Clevo decides to keep maximum speed at 80%.

 

Well, hasnt @Prema mod bios max fan speed fixed that

 

On a forum where people go to ask about issues seeing 3 failures in 18 months is actually remarkably good.

 

Seriously? Gpu failure on any cost arent good. Be it straight out of box or after 3 years.
Anyways what would be cost of a gtx 1080 replcement approx?

 

Define failures please. Are we talking about hardware failure or just some general term where someone has something going on and doesn't know whats going on despite being an easy software fix for example.

Also don't forget, most people RMA their product because they do not accept failure at that pricepoint and refuse to have to fiddle with it. It will be much more interesting once the hardware is around 3+ years old, since then people are more desperate trying to fix their system.

 

Yeah, good call sir,
What were the failures? VRMs, VRAMs, complete gpu burned?

 

I won't tinker with it physically for sure. Just hoped it could be dealt with software wise or something, as losing laptop for at least 2 weeks will be horrible. I have old Clevo still, but I still would prefer avoid having to get a replacement if possible, which I think is easy to understand.

The thing is that last week both PUBG and Stalker ran completely fine, undervolted or not. And Stalker is an old game, too. I will surely still do some testing today once I get home (maybe my laptop got bipolar... I mean, hope dies last!), but still, I'd not be as much confused if it was happening before.

 

Did you do any updates? Nvidiadrivers or win10updates stir up crap like this everytime, try reverting to *some of the old drivers and see if this happens again!

 

This is more than enough reason to mod it or RMA it.

 

last system update was Friday last week. Everything was fine during the weekend.
Not sure about Nvidia drivers, but reverting to older one was one of the first things I did yesterday when I noticed something is wrong. No luck, though.

 

Well, did you try contacting your reseller? What was his reply?

 

Not yet, as I've came across the issue late evening yesterday, and decided to rest and take some time to find a solution, or get a miracle before I talk to them about warranty.
If the issue still persists today, I will contact them, since my time runs out. Unless they agree to pay for inter-country courier, but I doubt they would.

 

Test - PUBG stock voltage curve - gets to 90 degrees after 5 minutes. Then Bluescreen.

Considering it ran fine before, I think there is no point testing further.

Is it possible that paste is at fault? I highly doubt, as I ordered it with Gelid, and it's been only a month. I am going to email reseller regardless.

 

Could very well be a subpar paste job, or the heatsink's contact isn't optimal.

 

I'm commenting on the anecdotal evidence and you guys need to realise that there will be a failure rate.

We could test them for 6 months of butn in and there would still be a failure rate.

The job of resellers is to help reduce that number where possible and fix it when it WILL happen to some people.

There is not one area of electronics (even space/military) that wont have an attrition rate.

 

It's not like I speak spanish or chinese, You're being very vague. We simply want to know what you mean with your statement. Are you talking about hardware failures? General failures? and what amount of notebooks are we talking off. I mean 3 failures is extremely high if we talk 3 notebooks as an example.

 

I do have to remember you guys do need this explaining every so often.

Percentages when dealing in higher volumes simply do not let you be 100% reliable, even at 99.9% reliability you will get multiple failed devices especially when you are shipping a complete machine.

I am not going to start going into breakdown of exact numbers on this and I think you will find most companies will not.

 

Hi,
I had the same problem on a P775DM3 with a 1070 GTX. GPU skyrocketed to 90°C within 5 min in games.
Did that trick:
http://forum.notebookreview.com/thr...cooling-mod-for-p775dm2-p775dm3.803626/page-4
And now it does not go above 65°C with automatic fan profile (80% max).
It turns out the GPU heatsink does not make proper contact with the GPU die. The small paperclip eventually adds and evens the contact betwxeen the HS and the GPU die.
You definitely should try this before RMA your whole laptop.

 

That's interesting. Unfortunately I already sent my laptop to home country. But it's a good thing to know if it's needed later on.

 

Well, i use copper shim rather then the paperclip mod. But a friend of mine @Falkentyne pointed out something wrong on the coppershim mod(cant thank this guy more), the coppershim actually elevated the heatsink of my old msi, making contact between heatsink and vrms low, so i had to add some thermal pads(2 layers to be exact).
See to this before doing that paperclip modd too!

 

Oh you changed your name!
Wait how many laptops do you guys have? o.o

 

Lets count
1. 2nd hand msi 16l13(dead)
2. Msi ge62(sister) (alive)
3. Clevo p150(dead)
4. Clevo p775dm3g(dead)
5. Clevo p870km1(dad)(almost dead)
6. Clevo p775tm1g(me)(at repair centre)(consider dead)
7. Touchmate laptop(only laptop of min which is healthy)
8. Msi gt83vr(brother)(sold on ebay)
I am like manager of laptops and mobilephones in my house. Consider all of these laptop to be "co-owned" by me

Edit, first brakets are rhe real owners and second brackets are the LAPTOP'S condition.

 

Why so many dead taptops?
Bad weather or accidents?

 

Accidents and my cursed hands!
Most have them died either by dropping or liquid damage!

 

Liquid damage? I would have thought after once incident it would not happen again.

 

1. Dropped
2. (Alive)
3. MB failure
4..dropped
5. Spilled over orange juice
6. Scorch marks over MB( no fault from me.
7.(alive)
8.(was working, but waa too heavy and big)

 

You call three hardware failures surprisingly often LOL
You do stand up show? Where can we find you

 

He is not being vague at all. Someone mentioned this being the 3rd Clevo 1080 that he's heard fail since Pascal's release in the mobile market. Meaker then responded by saying "3 failures in 18 months is remarkably good", which is very true in the context of how many units were sold in the past 18 months. The exact number is impossible to extrapolate without gathering data from each retailers/ODM, but you get the picture that a lot were sold in 18 months, and out of the many sold, hes only seen a few fail on this forum. Why it failed is completely irrelevant to the subject, when the context is about the number of failures that he has seen.

Now, from your post, it's hard to determine whether or not you want RMA's for petty reasons to be considered failures. If you do want them to be classified as failures, that would indeed impact the "failure" rate, but it's still impossible to give that exact statistic without knowing how many units were sold with these Clevo 1080's. If you then wanted to know why each of those 1080's failed, you would then need additional information about the environment they were used in, how hot they got on average, whether or not people overclocked even higher with subpar cooling solutions, and if any negligence took place when servicing the component. That's an awful lot of variables to account for when trying to get specific data.

All things considered, if poprostujakub only saw 3 Clevo 1080 failures on this forum in the 8 months he's had his account here, I would still consider that a win.

 

3 Clevo 1080 failures in such short time after release, and not even two 1070, 1060, 980M or other GPU (except those, where user set power limit on 180-200% of default level, like me with my old 980M. Beware of Furmark!). If you have too much money to spend for post-warranty repairs - go ahead. Temps above 80°C on that large die are critical for reliability and Clevo engineers knows this for sure. But this is their problem only for two years - I'm curious, how many used Clevo 1080 will survive long enaugh to be still usable.
We can see a lot of 680M, 780M, 880M and 980M GPU's on secondary market and only very few AMD cards. And I thinks, that Clevo 1080 will end like AMD cards - cheap, problematic and atractive only for baking fans.

 

Yeah. That's a point to noted, only gtx1080 failures, not even single 1070 or 1060!

 

You are way over thinking this with no information.

There has been no spike in a particular cards failure rate that I am aware of.

 

So, @TehGM did ya do rma? Or figured out problem by yourself?