AW18 - 880M SLi Temperature Discrepancy

I am wondering if someone can give some thoughts on the following.

The system is running 880M SLi under windows 8.1, and both cards are using the exact same thermal paste and setup. Ducts are clean. All appears fine.

Yet, for reasons that I can't pinpoint, GPU 1 seems to heat up to 93 degrees and then consistently throttle down in order to avoid going in excess while GPU2 generally levels off at 83-86 degrees, even at full load. I've attached an image that demonstrates the point and I can assure you, this is a consistent case across a wide range of titles. Even with games that don't stress the GPUs as much, GPU 1's temperature keeps climbing gradually before hitting the 90 mark while GPU 2 -- under the same load -- will level off at around 80 degrees. This just doesn't seem right.

Paste was reapplied recently just to confirm if this may have been a one-off problem with the prior paste application, but despite this, the very same issue persists.

I don't know what else to suspect other than the heatsink or the thermal pads on the memory modules, but even that suspicion is unsubstantiated.

Any input would be greatly appreciated.

Thanks.

 

Sounds like either a paste issue, a contact issue with the hotter card (HS not screwed on properly) or an airflow issue caused by something like a cable getting in the way or under-performing fan on the one side.

 

Thanks for the response. It does read a lot like an issue with improper paste application and-or poor contact between the heat sink and the card. However, I can assure you that this isn't the case -- the card is making appropriate contact with the heatsink, along with a relatively ideal amount of paste. I had even gone back after the last reapplication to be certain of the matter. The same can be said about the heatsink's air ducts being obstruction free and while I can't be certain about the GPU fan running at its best, it does sound like it's running fine. This is why I'm rather curious about this issue, which in most other circumstances, would be considered rather trivial.

I could have passed this along to Dell to handle, but I wanted to get some input on here because it was something I was curious about. I guess I'll give the paste routine another go and see how things turn out and if it doesn't help, I'll get in touch with technical support.

 

My guess is one of your 880Ms is a lemming.. It must be having extra voltage leak hence heat generation.. I think your safest bet is to get Dell to replace that card only.. The other one which is fine, keep at any cost!

Sent from my LG-H811 using Tapatalk

 

Thanks for the response.

I didn't really give voltage leakage any consideration, but now that you mention it, the issue seems plausible.

I'll work on making sure that Dell only replace GPU 1. Of course, if they insist or press further for the other GPU to be replaced, and if that results in more problems, then I'm going to make sure that they take full responsibility for the matter.

 

I'm wondering if something goofy is happening with the fan not working correctly. If you force the fans to run full blast using HWiNFO64 does it still get that hot, or only when allowing the EC fan tables to function automatically?

 

Thanks for the response, Mr. Fox.

The issue persists even when I use HWiNFO64 to run all the fans on full blast. Until and unless the fan itself has a blatant fault with its mechanism and isn't physically able to run as fast as it normally should, I can't say that this is the problem.

 

Swap the cards.

If the problem goes away, it was a paste issue.

If the problem still occurs on the same slot then it is a fan/airflow issue.

If the problem moves with the card to the new slot, it is the card.

 

That's definitely worth a shot when I do the repaste. Thanks.

 

I have an update on this matter.

I swapped the cards around and reapplied -- to a near exact distribution -- the same paste on both cards. Nothing else was changed and the heatsinks were properly reseated making relatively equal contact on both cards.

Once again, the GPU in slot 1 seems to continue heading upwards until it eventually hits the 93 degree mark while the GPU in slot 2 tends to level off around 5-8 degrees earlier. It is safe to say that the issue is not with the individual card itself, but something pertaining to that particular slot for GPU 1 and the other components involved (heat sink, fan, etcetera).

I also used HWiNFO64 to keep the fans running at maximum, but this merely delayed the inevitable. Although I do have one additional question on this topic. Why does the manual control in HWiNFO64 have three different RPM limits for the fans? I could understand the CPU fan having a different limit, but the GPU fans ought to be the same, yet what I see is 3700 max for Fan 1, 3500 max for Fan 2, and 3300 max for Fan 3. Is this normal?

Also, while swapping the cards, I took some photos of the heatsinks because I noticed an inconsistency in the structure of the one that sits on GPU 1. The tips of the heat pipes on the GPU 1 heatsink -- as they sit at the centre of the heatsink -- seem to have very deep indentations, rather akin to potholes. Whereas the CPU and GPU 2 heatsinks don't have such an indentation. I can't be sure if this is of any consequence, but could this be part of the problem? I have attached a comparison image for direct reference.

 

Get them to replace that heatsink... That doesn't look correct.. This should be the explanation for the issues!

 

Yeah, I thought so, but felt compelled to bring it up here since I have seen photographs of other Alienware 18 heatsinks on the internet, and some of them seem to have the same indentations.

 

That's too many indentations IMO.. Get it replaced, no harm done... If it doesn't work out, get a new 880M.. One of the good thing about the warranty!

Sent from my LG-H811 using Tapatalk