Is my 680M in my m17x r4 dead? And what can i do besides buy a new card.

I have been playing overwatch like crazy the past 2 weeks on my m17x r4 with 680M. Unfortunately since yesterday i am getting BSODs anytime i try to even access the dedicated 680M GPU. Starting any game or even starting the nvidia control panel and going to the 3D settings which renders the nvidia logo will cause a BSOD. I can't even downlock the GPU, everytime i try in nvidia inspector, it BSODs.

I have tried reinstalling the driver, upgrading the driver, downgrading the driver, starting in A/C power both with and without battery, starting in battery only, shutting it down and leaving it for about 12 hours and nothing seems to fix this.

Here is the imgur album of my BSODs (getting different numbers each time): http://imgur.com/a/wwRZp .

If this is indeed graphics card failure, is there anyway i can get this repaired? I know i'm out of warranty, but does dell do repairs for payment? Unfortunately I'm a student at the near end of my studies and i'm flat broke so buying a new card is out of my budget [Around 200$].

Thanks in advance for any assistance you can offer me!

 

Where are you from? A friend has a spare 780m chip, but it wont come much cheaper than this.

 

Phoenix, AZ, USA. How much is he asking for it? I don't want to run into another dell card which fails on me .

But like i said, ideally i want to repair my card and save some money if possible. Also i don't have much technical skills so replacing the card would be tough!

Also BTW do you know if those errors show in BSOD are confirmed dead card?

 

Well I was talking about repairing, swapping your dead 680m chip with a 780m one. As I said, it's costly, the chip alone is quite expensive, then comes the work involved, the shipping (which is not at all feasible, since you are on the other side of the world). If you were in Europe, at least the shipping wouldn't been a problem and duty-free as well. You'll be better off getting a new GPU, or save some more for a new machine. Sorry.

Pretty much dead.

 

Have you tried cleaning the edge connector and reseating it? It's very unusual for the 680m to fail but of course it could have done.

A common fault I have seen with my R4 and others is the card to appear to have completely failed (will not POST) but a clean and reseat works wonders.

You can remove the card and heatsink as a complete assembly (after removing the fan) so no need to get involved with pads and paste .

Fingers crossed .

 

What is the edge connector? Where would i find it?

By reseating you mean completely taking out the GPU/heatsink/fan and all the wires attached to it and then plugging them all back in and putting everything in place? I could try that, never done more then re-pasting my CPU in this but i'll give it a shot! Any special tools i need before i start?

Also i don't think it's a complete failure, because a couple of times when i went to nvidia panel the driver crashed and did get recovered by windows, and then the nvidia logo was rendered (but then a few seconds/minutes after that it, it did BSOD).

 

Yes, remove the GPU and heatsink. It is connected along one edge and simply needs two retaining screws removed. It then lifts at around 15degrees opposite to the connector and you can then lift and pull it out. If you have ever changed a memory chip it's a similar type of edge connector.

Seven screws in total IIRC, two for the lower case, three for the fan and two for the card. Takes less than 5 minutes (when you know where it is!). You should find some teardown videos if you hunt around the alienware support. I quickly found this one on youtube:

From 2.15min onward BUT, you do not need to remove the four screws to remove the heatsink, just undo the two retaining screws and wiggle it out in one piece. Don't panic about a wriststrap, just make sure you discharge any static from yourself on the chassis and don't go shuffling along the carpet haha. Clean the edge connector with some alcohol and blow out the socket - a can of compressed air is very useful for this. You will also probably see a whole bunch of fluff in the heatsink. If this is really bad the card could even be overheating, although your experiences with the card don't immediately indicate this.

It is amazingly easy as they were designed to be stripped down - even the original AW documentation had instructions on how to replace a GPU! No special tools just one very good watchmakers cross-head screwdriver (sometimes known as glasses screwdrivers). Just do not attempt to use an ill fitting screwdriver it should fit nice and snug. Biggest problem I see is people butchering the screws with screwdrivers too big for the job.

Good luck. Hopefully this is it but otherwise you now know how to replace it .

 

Did you get to try it?

 

I have a GTX 680M I would consider parting with...

 

Yeah i did but to no avail sadly. And it took a considerable amount of force (literally had to grab the heatsink and pull it out, wiggling was getting nowhere) to get the card out of the socket and it had this wierd square of yellow sticky stuff underneath the card on the motherboard. Here are some pictures of the card and the sticky stuff: http://imgur.com/a/53Bv8 . I didn't have any alcohol and just used a hand pump for the air and was still getting the same BSOD after putting it back in.

EDIT: Just got a new type of BSOD which said "DRIVER_POWER_STATE_FAILURE". Maybe this indicates it is a connector problem?

I could try booting it in dedicated graphics mode with FN + F7. But i'm afraid if i do this and the card is actually failing my computer won't boot and i won't be able to switch back to integrated graphics.

Yeah but for how much and are you in the USA?

 

Afraid you can't use just dedicated. You said earlier you could run the 3D logo test. This means you have a 120hz display and the dedicated graphics is disabled. Fn+F7 will just give some error about already being optimized (IIRC!).

That pad under the card is nothing critical, just to protect the MB components. You can get alcohol pads very cheap from a pharmacy. Get injection site swabs (diabetic supplies), these are perfect for cleaning AND you will need them if you have to change the card.

If a clean and reseat does not help (give it one more go) than the likelyhood is the card has indeed failed - sorry

 

I have a Dell OEM 780M for sale, 307 USD shipped to you USPS priority.

 

Just a little update, i recently noticed the gpu was actually working on low power mode so i tried underclocking it but i'm having a slight issue. The core clock won't budge for some reason but the memory clock is going down. I found the memory clock to be stable at -300 MHz (It works anywhere below -100 MHz, but anywhere other then -300MHz there are lots of artifacts). I'm using MSI Afterburner 4.2 and i cant get the core clock to move no matter what i do, it stays at 719. Any idea how to fix this? I'm currently on driver 331.65 (the latest dell/alienware supplied driver).

EDIT: After a restart, it seems sometimes the core clock is also not moving from the base 324 MHz, even when being utilized in games. The memory clock is moving from its base up to the maximum but the core clock is not moving from the base.

EDIT2: After some fiddling with nvidia system tools,i was able to change the core clock but it has no bearing on the artifacts at all. Seems only the memory clock affects it. Even at -300 MHz I'm getting noticeable artifacts on games like overwatch, but its barely even noticable on games like dota 2.

@MickyD1234

 

seems like its finally on its way out. I'd call 4 years of service out of any high performance graphics card a life well spent

 

probably shorter in the mxm because of higher operating temperatures

 

seems like its finally on its way out. I'd call 4 years of service out of any high performance graphics card a life well spent

 

Figured as much, sad . I have to keep this gpu running until at least december when i can afford to upgrade, so underclocking it will have to do for now @.@.

 

Gotta agree - myself, I consider two years 'normal', after that a bonus...

 

just my $0.02
i used to get similar BSODs when i hadnt cleaned up for a while(my area is quite dusty)
Just after boot, i would start hwinfo and msi afterburner to see temps
idle temps would be 65-70C
and would bsod as soon as i started anything(chrome, games etc) max temp would hit 95+(anything above 75 is high for gpu IMO)

after cleaning and repasting, idle temps went back to 40-50C
no more BSODs

 

A little update/new complication.

So i disabled the graphics card and have been recently using the integrated intel hd 4000 graphics for the past 3 months. Last weekend all of a sudden my screen kind of fizzled out and got a black screen. I tried restarting, no avail. Computer boots, everything works perfect but the screen is just black. When i run the dell diagnostic test holding FN the monitor shows the red blue green white colors so it's not a screen issue. I also have plugged it into an external monitor and it was working perfectly fine. I also used the old heating the graphics card trick which allowed me to boot the laptop in dedicated only graphics mode (only on severely under-clocked speeds) and it worked fine, but trying to boot it in integrated/shared mode launches a black screen.

Is this related to the gtx 680 or did my intel hd 4000 graphics also die? Talk about bad luck -_-'. (I thought that it wasn't possible for integrated graphics to die without the processor dieing too)

I really need this laptop to last until december, any help would be appreciated!