Alienware 17 GTX 780M Temperatures?

Hey everyone,

I recently repasted my GPU with TX-4, installed the T|I modified vBIOS, and ran Furmark. I undervolted the GPU from 1.0 to .90V and the temps still hit 87C, sometimes 88 or 89C. What the $%*^ is going on? This is ridiculous and I know I did a fine pasting job, too. Any other users experience really high temps with this GPU?

Thanks

 

I had the same problem after I repasted. My problem was that the thermal pads got in the way of the heatsink connecting properly to the GPU. I saw 90 C running the Heaven benchmark. After I readjusted one or two thermal pads and repasted again it's now fine and I haven't seen temps above 73 C.

Maybe you have the same issue?

 

Could you show us a picture of what we should be seeing?

 

That could be the issue, also the heatsink could need a replacement.

 

Could you send a picture perhaps or in more detail explain what you see and what actions you took to correct it?

I don't see how the heatsink could need a replacement? The thermal pads are all on there and it has no flaws in its design as far as I can tell. However, I can easily reach 90C at stock clocks and 1.0 stock voltage, which is ludicrous.

The heatsink is lying fine over the chip. I don't see any issues.

 

You need to use a better paste and use more of it. You said you used a tic-tac (mint) size amount and that's not enough with TX-4. That much may work with IC Diamond, but TX-4 doesn't spread well. Don't be afraid to be a little extra on there (obviously don't cake it on).

Try using more, but not too much more. If that doesn't work, switch to IC Diamond and put a little more paste on it. If that still doesn't work, it could be hardware related (heat sink, fan, etc.) or software related (vBIOS or driver).

 

I'll try a different paste. Temperature issue cannot be due to vBIOS for everyone else would have problems. I've reflashed a few times and still the same issue so I can assume that it's not a factor. As with drivers I have the latest ones. No performance issues so it seems like a non-factor as well. The only thing I can think of is insufficient/bad paste.

What would make the heatsink faulty? I don't see why/how it would be, as it looks fine.

 

The heat sink could have a weak spot or something, which wouldn't be noticeable to the naked eye. It's rare, but still a possibility. Like I said above, there's a good chance it's just the paste or paste job. I'd say there's a 95% chance of that. Just give it another shot using a different paste.

 

Take pics of the thermal pad layout and htsnk, probably poor contact GPU die/heatsink also make sure that the vents/fan are clean.

 

I see. Well idle temps would be an interesting bit of data too, and I seem to hover around 46-50C. Any word on that? It seems like I have good idle temps but bad load temps.

I'll try that tonight. Perhaps I should take off the pads from the heatsink and put it on the card itself?

 

Thermal pad placement, "fill the gap" black metal part/580M heatsink.

 

You're saying I should follow that model?

 

Yep, since you're using the 6970M heatsink (75W) consider removing the c-clips (protection) be careful not to add excessive pressure, if temps are still high get the 580m heatsink (100W) try sending an email at [email protected] with your service tag# and phone number (insist if there's no answer), 580m htsnk $25 P/N V1K2G.

 

You're using a 6970M heat sink for the 780M? I didn't know that. That could be the problem. You need to get a 680M heat sink off eBay.

 

The 580m htsnk works fine and ebay prices are ridiculous $100 (Dell $25) 680m P/N 9CNMG (R4) you could try, but using the R3 service tag# probably won't work.
3DMark11, room temperature 25C (580m heatsink) :hi2:

OC 1097/6180 1.075v

 

No no no, in my title I explicitly outlined "Alienware 17," as it was a replacement for the R3 in my sig. My sig now reflects the new AW 17 model.

I have an AW 17 with a 780M using stock everything. Sorry for the confusion.

 

I'm tempted the say that the problem is that you're using Furmark with T|I's vBios.

The default vBios automatically down-clocks the 780m with Furmark.
T|I's vBios lets the chip run at full speed. I even think there's a warning about how dangerous Furmark is with their vBios on their website.

The temps during an hour long session of Crysis. TR, Bioshock or Metro 2033 would be more relevant. If they don't go over 75 C, then they are probably good enough.

 

Missed that never use Furmark, even more with mobile GPUs...run a demanding game like Voltran suggested, use HWiNFO or GPU-Z to monitor temps, report back.