I have a 680 SC SIG 2 and have had it for 3 yrs now.
 
I have noticed my active temps go from sub 60 deg to now sub 70 deg and have a performance decrease and crashing driver services when ever my card gets above 65 deg.   I play EVE online and planetside 2 @ max settings running 5760x1080. 
Is it time to redo the paste, update my drivers, or do I need to bounce for a new card?

Update to driver 353.38 and see what happens.

Sajin, Thanks, while the problem didn't completely go away, it made the high temp effect less damaging to gaming so I could run at slightly higher settings.
 
Zuhl, yeah, Im going to order parts but whatever advice you can give (parts etc) would be much appreciated.

Zuhl, Thank you so much. I ordered the MX-4 today. I'll get back to you next week sometime after Ive did everything. Very helpful! Thx man
 
Ken
 
 
 

I don't think redoing the tim is going to help here. 65-70C is nothing.

I saw a guy use a (new) latex glove to smear it over as evenly as possible. It seemed to have more control of how much would 'leak' off the edges, but at the risk of the inner part may be inconsistently applied. >shrug<

I use the edge of a credit card every time.  It is a fairly even surface and has worked on 13 cards so far.  I am just going to continue doing it.  the latex glove/sandwich bag sounds like a good idea as well. 
 
Another thing you can do while in there is replace the thermal pads, purchasable through Performance PC's.  That may help with overall temps, as it will help actively cool the VRM and Memory on board quicker, and allow slightly better overall board cooling.  3 years is a good lifespan for thermal material.. mine usually makes it 3 months before I am tearing it apart to clean it anyway... I am impatient, and don't want to wait for it to dry.

Pea method.. Don't do that to your GPU.  There is no guarantee it will cover the edges which can kill the core, since there is absolutely no way to verify it 100% covers everything.  The Pea method is for CPUs, which have an IHS to spread the stock thermal paste, which we have seen is often ineffective thanks to Intel using thermal paste.  If you spread the paste yourself, at least you can verify 100% coverage like you 100% should.  Pea size method can result in this, which will causes crashes and eventually kill your card.  The thermal monitoring is in the center of the die.. the edge was overheating and the card crashed.  I had three do this. This one came from the factory this way, and the other two were while installing Kraken AIO's.  I will always use the pea method on the CPU, but NEVER on a GPU, which is FAR more sensitive to temps and is NOT measured per core like CPUs.