The plot thickens, trouble with a brand new M17x R3 580m

Yesterday I posted a thread because I thought I was having trouble removing the throttling. Turns out the voltage throttle is removed after I flash the vBios, the interesting thing is that the computer still crashes (thus why I thought it was voltage throttling).

What is really interesting is that it ONLY crashes when I have the laptop elevated. I knew that it would really help, so right out of the box I set it up with lots of air space. The problem is, when it has lots of air space it will BSOD BEFORE it hits 77 in furmark, it is always the nvkflt.sys file that is blamed for the BSOD. Even on stock settings with the lower voltage modified vBios it will crash before 77. I can even vary when it will crash by putting it closer or farther away from the desk. Way off the desk it crashes at 71, closer it crashes at 74.

However, I take it and set it flat on the desk and then it doesn't crash at all. It goes up to 77, it throttles, it goes back up, etc. Basically, I seem to have a laptop that hates being cold, or something.

Anyway, just wondering if anyone has heard of this before, I have a bunch of .dmp files that I need to read but I will have to try to get the software to do it. If anyone is good at reading .dmp files I will gladly post them here. I really feel like this is power related, it is like my machine cannot handle full power all the time, because if I keep it cool it crashes, but if I let it heat up and throttle it doesn't. I guess it might be possible this is a bios or driver issue, but sadly I suspect it is hardware.

If you guys have any advice let me know, I would rather try some fix and have it work than get into tech support with a 24 hour old laptop. I have to admit, it is a bit disappointing, but I am going to test actual games today and see how they behave, maybe Furmark is just too stressful? Wishful thinking, I know.

 

you running at default voltage or 92?

Also, what driver version are you running?

The fact that your card crashes when the laptop is elevated tells me that perhaps you need to look at the heatsink and how it is sitting on the GPU die, or maybe a bad connection on the mxm 3.0b slot. Whether or not the laptop is flat should have no baring on stability. Havin the laptop flat will increase your temps 7-8c though. A repaste is probably required too

 

Default voltage, everything stock, the only difference is the modded bios that makes the two P states the same to avoid the 63 degree throttle.

 

ok, then physically look at how the heatsink sits on the gpu die...and reseat the card cause it could be a bad connection on the MXM slot

 

I have tried several drivers, I tried the stock ones on the machine, the drivers direct from Nvidia, and the Dell verde package from their site. Right now I have it running a clean install of the Dell Verde package with the newest drivers.

Any pictures of how it should look? I opened it yesterday to swap my SSD into the middle slot for faster speed and everything looked fine, but I probably don't know what to look for. I would be happy to open it and fiddle around (but I want to be careful not to void my warranty, I have 4 years accidental, so please let me know if something you suggest would void it).

If I take it apart am I required to repaste? I don't have any paste right now, but I feel like the seating thing is very likely, as it gets hot it expands and when elevated maybe that causes the card to lose connection. This also makes sense because sometimes the very next reboot after a crash my card is not found, but then another reboot after that (so the card has cooled) it shows up again.

 

Here's a guide that should give you very useful hints on how to take everything apart:

http://forum.notebookreview.com/alienware-m17x/561599-m17x-r3-cpu-gpu-re-pasting-guide-w-pics.html

The heatsink should be flat and snug on the GPU die. Check for bending. If the screws are overtightened, it could cause flex and bending of the heatsink which is not good. If you do take it apart, you should repaste with a quality paste because the stock paste is built for durability, and NOT best temps

And reseating sounds like a good idea.

HOwever, if you're underwarranty, having the dell tech over might be a good idea too

 

I probably will not take it all the way apart, I don't have paste right now anyway. What paste would you recommend so I could order some?

I will probably take it apart enough to see if the screws are overtight and if just reseating fixes it, because a tech is probably at least 4 days away and I would like to use my new $3000 paper weight.

 

Sorry for the bad luck man, my new R3 580M works perfectly as I hoped since I got it weeks ago.

I'm using the beta 290.53 and they're amazing for BF3, slightly disappointing for GTA4 but that's a CPU dependant game I read somewhere.

I suggest just getting a tech out with a new GPU and see if they can fix it easily.

 

Yup gonna have to be a new GPU. Like I mentioned in my other post it has been disappearing after a crash, well now it disappeared and won't come back. On the phone to Alienware right now, they offered a brand new system, but I would rather not wait so we are gonna try the tech with a new GPU/Motherboard approach.

Anyone know if I had thermal paste when he came if he would use it? Maybe I get a better paste job outta of it, though it is pretty frustrating to wait a month, get the computer, and have the GPU die a day later. I understand this isn't really Dells fault, it lasted through lots of stressing yesterday while I was trying to figure out the problem, so it is probably not something they could catch in a quick factory test.