My graphics card would literally boil water. What do I do with it?

[Deoxim]

After a nightmarish RMA process that left me without a video card for a month, I received a Sparkle GTX 460 2GB card (which reports 1.5GB in the OS?). Now, I've had this card for two months, but I'm just getting around to asking what to do about my problem.

It creates lots of heat--more than I thought possible. After launching any demanding game, or about any 3d application without Vsync, it shoots up to 85c+ nearly instantly. After a short gaming session, I can alt tab and catch it at 97 or 99c. I don't think it's a heatsink issue, because it cools down almost as fast; see this chart: http://i.imgur.com/UPH7k.png .

It can play games as well as my last Sparkle GTX 460 1GB (which started artifacting), but when it stays at 99c for too long it starts crashing the drivers.
I don't believe it's overclocked, and it seems to create more heat than is physically possible. http://i.imgur.com/fRybh.jpg (see the UPS load.)

I don't want to RMA it, as I had trouble getting them to even answer the phone, and they might never return a card at all.
Is it even salvageable? Should I just buy a new one? Should I take it out because it's a fire hazard? Should I give it to a friend who has an even worse card? Should I donate it to science to aid in the design of more-than-ideal radiators?
I'm just perplexed. Have any of you even heard of something like this?

(P.S.: The Microcenter employee said they were out of stock of EVGA cards, and that Sparkle wasn't bad.
Remember this as one of those RMA horror stories; never buy Sparkle! They are not 'any good'!)

EDIT: I see people saying "85c is too hot!" and wonder what went wrong with mine. Buying another is not out of the question, I mainly wanted to know if I'm missing something.

 

[unksol]

check and make sure the gpu fan is ramping up with the temperature. unless your ambient temp is high or you have horrible airflow the card is probably defective. its perfectly possible for a GPU to hit 100°C if its cooling is not working properly

the lesson is more to never listen to a store employee trying to sell you something, kinda common sense

 

[JamesSneed]

That is some pretty high temps even for a GTX 460.

Just for kicks can you manually set the fan to 80-90% and see if the temps stay reasonable?

Have you tried pulling the heat sink and applying new thermal paste?

 

[Deoxim]

90% speed, or 4020 RPM, at 99c http://i.imgur.com/R8Dzs.png
I don't think it's airflow (with the case open and such), and ambient's only somewhat high. It's quite hot in Georgia now, but the log used in the graph was taken about a month and a half ago.
Also, Microcenter employees are different in my experience. But I guess they can't hire all good people.

I haven't messed with its heatsink. I'm assuming it's seated/pasted properly, with the speed at which it cools down, but I may be wrong.

 

[lycros]

My first guess would be the heatsink wasn't installed properly, either a bad application of thermal paste or not screwed on correctly, or maybe a combination of both.

 

[Deoxim]

I'll look into how to replace a GPU heatsink.
So is there any real danger to my other parts?
If I can give the card to a friend to use until it dies, I may do that, but not if it'll take anything else along with it.

 

[Antimatter79]

Jesus, dude. You must have went to the Microcenter up 85 N, the one in Marietta isn't stellar, but they are ok in there. Is there a possibility your mobo is defective just in that PCI-e slot? Have you tried using a different slot if you have one available? Make sure if you're using Nvidia control panel, MSI afterburner for clock/fan/voltage settings, or anything like that, be 100% sure you don't have overclocked or overvoltage settings applied. Put everything you can think of at default. Check your bios, make sure no voltage settings are out of place there. What about your PSU connectors feeding your card? You could meter those (if you have a meter) see if they're dropping too much voltage to your card, or if the card has bad parts on the board, it could be pulling to much current (which could also be metered but not as easily). Does your PSU smell burnt? Mobo smell burnt? If all those things check out, they must have a bad batch of cards with irregular contact surfaces that mate with the heat sinks, bad thermal paste, or something. Can you even hear if your card sounds like a Helicopter taking off when it gets hot, or is it quiet as a mouse? As far as damage to other parts, it really depends on what's causing it to overheat. But in general, unless the issue is electrical, I can't see it damaging anything else that's not touching it.