R2 5870m Temps question!

My cards were overheating so I re-applied thermal paste on both (and the CPU for good measure) and replaced the thermal pads. I was careful to do everything properly.

My slave card hasn't improved (maybe even running hotter at idle) but my master card... is like magic... and hardly heats up at all? This screenshot is after 8 minutes or so of Call of Duty 4.

Any idea what's happening? I've already redone the slave card once, so I don't think it's just a bad pasting/padding job. Maybe the slave fan isn't spinning up right? Also, in case this is relevant: setting the GPU fan speeds manually works for the master card but not the slave.

 

Well COD4 will hardly stress out a 5870M. After playing DIRT 3 and Starcraft 2 on 1920x1200 high, my 5870M CF were 82C+ without a repaste job.

 

It sounds like you are having slave fan issues. If you have manually set it to max the RPM should scale up to the 4200-4400 RPM range.

It's important to verify this by looking under the "Alienware M17x-R2 EC" section in HWinfo like the pic below.

Note: I only have a single GPU1 but your's will show your slave fan GPU2 as well
.
Also important to note that the GPU fan speed reading under each of the individual GPU sections is not valid on our systems

 

That's why I was worried when one card hit 100C while playing it.

Thanks for the tip on the GPU fan speed - I was wondering about that. I only see one GPU under "Alienware M17x-R2 EC", however (see attachment). Does this mean the cards or HWinfo or something is configured incorrectly? By the way, I made sure to disable ULPS - if that matters.

Thanks for the help!

 

Remount the GPU and report back. That looks like a poor contact job.

 

I doubt it's your card as it's being detected.
I'd suggest pulling the card (you don't need to remove the heat plate from the card) and making sure the fan is free of obstruction, unplug it's cable and plug back in.

Then before putting your card back in, plug in your power supply and turn it on while watching the fan.

All 3 fans "should" briefly spin up at max speed as soon as it powers up.

After verifying, pull the power supply, and make sure you properly reseat the GPU.

Check it in HWinfo again. HWinfo should detect and monitor the fan by default unless you've changed the program's configure to disable monitoring.

 

I'll do both if necessary when I get a chance. Thanks!

 

An update:

I turned the computer over while it was running to check the fans and found that the slave fan was not spinning (certainly explains the temps). So, I opened it up, re-seated the card and fan, and now the fan works. It does, however, make a new grinding noise which worries me. Perhaps it simply needs replacement (unfortunate, since I'm out of warranty). Interestingly, the grinding sound subsides as the fan goes to 100% speed and sounds more normal (the grinding does actually get a bit quieter, it isn't simply drowned out). This is a much more pleasant sound than the grounding at idle. Is it safe to keep it at 100% until I figure out what to do about replacing it?

HWinfo now makes more sense and reports two fans and I control both. See the attachment.

 

I just replaced my secondary GPU fan yesterday, so I know the feeling. I'd definitely get a new fan and replace it. It's not horribly expensive if you are out of warranty (I picked one up on E-bay for $37 shipped).

Mine would spin for a few minutes and then completely turn off (or slow down) and my temps would sky-rocket and the system would power off. Needless to say I was lucky the card didn't fry.

I wouldn't risk it, get a secondary fan while you have the chance and then if things get worse you can replace it. Really simple thing to do, and you'll feel much better.

Good luck!

 

By the way: I can only get my fans to go to 100% by manually forcing them there in HWinfo. How can I get them to work properly without running that program?

 

Out of curiosity, what AMD driver level are you running?