Is my M18x R2 7970M CF dead?

Hello, Been a long time stalker , first time poster. So last night I was playing some BF4 on my M18x when the machine just crashed. After that, the machine no turns on, lights come on, all the fans spin up, then after a couple seconds the GPU fans turn off. There are no error beeps, nothing. Removed the memory, turned on the machine, I did get memory error code. But again once memory was re-installed and powered up nothing happened. Unplugged Cmos battery, waited and plugged back in, I got a Cmos battery error beep code, then the machine turned off.

After that the machine fired up, got it into widows repair (as windows would not boot), but I noticed blue bars on the screen, Windows ran repair utility then the machine crashed out. Now its back to the turning on but nothing happening mode. Part of me thinks my Primary GPU died, as the memory error codes/cmos codes from before lead me to believe the Mobo is ok. I have tried everything short of taking the machine completely apart, secondary monitors, swapping ram, anything my PC experience lead me to do. Has anyone come across the problem before? Am I correct to think it could be the graphics card?

Alienware M18x
Core i7-3630QM
7970m Crossfire

here is a picture from when the machine did boot up that one time.

 

How long have you been using those 7970m?

 

I've had the machine about 3 years now.

 

It does sound like a GPU died on you. You should be proud of them, they lasted a lot of time with you. Try to boot with the secondary GPU on the primary GPU place.

 

Is 3 years long for these cards? is this a common problem with the 7970m?

 

Yes, the 7970m had a very bad life time. Almost everybody that was using them had to switch to nvidia because of this. There are some lucky users, like you, that got good life time from them. Just do some search in this forum section and you will see how many died faster than yours did (If they are dead).

 

You are lucky to get 3 years out from the 7970m, i got 2 set of 7970ms, the first set died around 1 year and the second set which im using is showing signs of dying. You might as well do a gpu upgrade on your m18x r2.

 

what would you guys recommend for a gpu upgrade that priced decently?

 

Your best upgrade is a 780m sli setup. The only thing that you have to keep in mind is since Alienware never supported this GPU for our M18x you will need to mod the driver so that it can install. You can download a modded driver from here: Mr. Fox's GeForce 345.20 Desktop Driver Mod for Alienware and Clevo Mobile GPUs Those are the best ones that you could get. 980m users are having a lot of trouble making them work with our machines and I haven't seen any happy 880m user.

If you don't want to mess with modded drivers you best upgrade is 680m sli.

 

Well, so far, I've taken apart the laptop, cleaned both GPU's and applied new past after cleaning up the jizz shot that dell left on the Cores, attached monitor and the laptop fired right up. Took alittle to get windows loading again, but I am now testing the gpu's on furmark, 99% load crossfire hold at 75 Celsius. Hopefully this was all cause from a cable coming loose or something.

Edit: after 20 minutes on furmark temps are hovering around 69 Celsius.

 

Im a happy 880M user. With new stock vBIOS that I had on the cards (from AW18) they never throttles.

And 680M SLI is a lot slower than 780M so don't even consider that. Drivers are pre-modded on the forum and you don't need to mess with modding unless you want some specific driver installed.

And really, don't buy AMD GPUs for laptop. I had two 6990M and I will never do that mistake again.

 

I didn't say that 680m was equal or better than 780m. There are users who don't like to use modded things, that's why I suggested it.

 

Do not use Furmark, it is a power hog that utilizes cards in a very strange fashion that is not indictive of real world stability. I've hard cards that passed furmark fine, but crashed during 3dmark. Furmark only heats up/stresses the VRMs of your cards, so avoid. I use OCCT GPU testing.

 

Yeah I know but there are pre-modded drivers on the forum that doesn't require any modding skills to install so he shouldn't worry about that.

 

After running 2 20 minutes test with furmark I loaded up Battlefield 4, played full round at ultra settings, and everything seems to be fine now, going to conduct more test to see if everything is "stable".

 

Plus FurMark is OpenGL while majority of games are Direct3D. That said, OCCT GPU test is DX11 but it's just as inaccurate as FurMark in determining stability. Heaven and Valley are much better, but actual games are best.

 

I really tend to question the rationale of stress testing under most circumstances. If things are working correctly there is no reason to, and if they are not it is not going to take very long to find out. What we are seeking does not require "stress" and there is no need to create it artificially. You wouldn't (hopefully) rebuild an engine for your vehicle and throw a cinder block on the accelerator pedal to see if it blows up, and there is no reason to do that same thing, conceptually, to a CPU or GPU.

I agree with @octiceps here about games being a good test of stability. If it is going to be unstable or have thermal management problems, you will find out pretty quick just playing games. In fact, some things seem stable in benchmarks because they are brief, but start to unravel 15 to 20 minutes into a frag-fest that isn't even close to exhausting system resources.

 

The stress testing is to create a set of variables that will not change. Playing a game is to infrequent in terms of GPU/Memory usage for full on reliable testing or to pin point a problem. I wanted to test the cards, as they where designed to be used at 100% capacity, to test thermal and GPU capacity, now to your point though, testing should be done with a barrage of programs, as I have done and have yet to find an issue pop back up , plus being that I just re-pasted the GPU's I wanted to see how the temps were holding up.

 

Yes, I understand there can be a time and place for stress testing to troubleshoot a problem. I agree with you completely on that. My comments were more the opposite... when things are working correctly, there is no reason to. Stress testing for no reason is not recommended. Some people actually do that. Some have burned up perfectly good video cards stress testing systems for extended periods of time that are working properly. I feel bad for people when that happens to them, so I don't recommend it unless there is a specific purpose to it, and then only for short periods of time with careful monitoring. Letting stuff run at full load overnight and crazy things like that can end very badly. Gaming and running popular benchmarks is strenuous enough for most purposes, and allows you to have some overclocking fun without destroying things.