5870M Crossfire - Laptop Shuts Off When Crossfire is Enabled.

In my M17X R1 I have recently put in a second 5870M GPU, but I have run into a major problem. Whenever Crossfire is enabled and I run a game, within 30 seconds to 1 minute the laptop completely shuts down - whether it is just the menu or actually being in-game. With Crossfire disabled there are no issues whatsoever, only when Crossfire is enabled.

What I have done so far is try out various different AMD drivers, and have tested this issue with the original vBios and a few custom vBios.

I don't really have much time this week to tinker with this laptop to find the issue, so what do you guys think, what would be causing this problem? Thanks in advance!

 

Donno much about the R-1

What PSU do you have? Since this is an upgrade you could have power issues? For the R-2 you need 240W to make it run with crossfire and 5870M's.

Also could be bad cooling on your second card forcing a shutdown quickly due to GPU temp?

 

Yes, I have a 240W PSU. The power should be fine if the R2 originally shipped with 5870M Crossfire and i7 920... I have pretty much the same system except instead of the i7 I have the QX9300.

It might be cooling, the thermal pads that I used were pretty thick, so would that maybe cause the shutdown?

 

I agree, should not be the PSU

If thermal pads are to thick you could loose GPU contact with heatsink? What heatpaste you using?
Have you checked the temperatures on the GPU's? (GPU-z)
Donno if you are able to see anything since it shuts down so fast.

Perhaps some of the R-1 owners could chip in to help if its not cooling.

 

I suspect a Core overheat. Check temperatures using HWinfo and do it fast, you could potentially kill your card with repeated shutdowns IIRC.

 

After checking the temps with HWinfo the temps for the second card are about 10 degrees higher. I have turned off Crossfire for now and have bought 0.5mm thermal pads; I think that the thermal pads I used are 1.5mm. When I get the new thermal pads in I'll update if that stopped the shutdowns.

I'm using Coolermaster thermal paste.

Spoiler

 

This should not be a reason for a shutdown I'm afraid. What is more important to me is to know the load temperatures of your card. Please take some time to run a benchmark while running a thermal logging software like HWinfo to monitor card temps. There are about 4 sensors per card so you will have to be monitoring all of those temperatures. ANY temperatures over 100*C is trouble and needs to be looked after immediately. Many users don't check the vRAM temperatues and truth be told, they can go over 100*C on many occasions so that could be a place to worry about. Added to that, we also have a critical chip known as the PCH and the designers at Alienware put NO active cooling on it. This chip is responsible for all your I/O and display needs so it's a very critical chip that often goes overlooked.

So to sum up

1) Download HWinfo
2) Fire up a benchmark with crossfire engaged
3) Monitor vRAM temps, Core temps (GPU & CPU) and the PCH temps
4) Any temperatures exceeding 100*C is a problem

 

Well, I did a benchmark in 3D Mark, and it shut off again. After checking the saved log from HWinfo I really don't see anything abnormal, but I may have missed something.

Here is a link to the log file: Download benchmark.CSV from Sendspace.com - send big files the easy way

 

Everything seems normal there. Have you tried running it with a different power adaptor? it sounds like the current one is unable to keep up. Sorry but I'm starting to run out of ideas one by one, i'll try my best to help

 

Well if your thermalpads are a bit hard and thick your GPU chip might not even get close enough to the heatsink to be cooled. Your coolermaster should probably be good enough if you get enough pressure. (but not the best of pastes)
10 degrees for doing what? In idle with crossfire enabled? If so that is a LOT in idle.


A lot of not needed info in that file of yours, you can log with GPU temps only.

BTW if im not sorting entirely wrong your GPU2 diode hits the roof like h***.



This is a screendump of the last logged temperatures on both GPU's
But then again, I could be sorting it wrong.

Btw
It increases a bit fast during this test for my next question to be the problem but still, could your heatsink for the secondary card be bad? And is the fan running normally?
(HWinfo will tell you about the fan and you can try to run it manually with HWinfo and hear it spinning up)
Just since it seems your memory hits the roof to, and with thick pads, memory is the one that should have good contact i would guess.

Cant use the data for the fanspeed since its either 30% or 100% after 100degC-

 

Wow, huge misinformation on my part. I am really sorry. Yeah, your GPU2 Core is way too high

 

Word of advice though.
Dont make this card hit that temp again.
Idle temp will tell you if you are getting somewhere with your cooling.

15 degC difference in idle is WAY to much and the 41degC to 55 degC different temperatures should be eliminated before benching again.
It WILL kill your card.

 

Yep, I'll second what Trome said. I am really sorry I asked you to run a benchmark. At least we traced the problem down. Please keep us posted

 

The heatsink is new so I doubt that that's the problem. Well, after seeing these temps I will not be using the card until I get the thermal pads. Once I get them I will post here and let you know if this problem still persists. It will take about a month or so before they arrive, just to let you know.

 

Any updates on that GPU bud?

 

Hello, sorry for the late update, I have been extremely busy these past few weeks with work and such. Today I have finally got to applying new thermal pads and pasting the core with IC Diamond. I am glad to say that after a full run through of 3D Mark Firestrike and a few games, the laptop does not shut off anymore! Thanks so much for all your help guys, I really appreciate it. :thumbsup:

 

I'm so glad to hear we helped man. As a fellow notebook enthusiast there's no greater feeling than helping a fellow user sort out their problems which the official service personnel often fails to provide.

*Request thread closure?*