This might take a fair bit of testing, but I just took it apart and put different paste in. (Akasa stuff that came with my Alpenphonn Mount Doom Heatsink I believe) As previously I put a small amount of Arctic Silver 5 in. I went for the less is more approach of cpu's as it's rather conductive to electricity, so didn't want spillages.
Having been re-inspired to have a go at self remedy I decided to put more of this paste on than the arctic silver.
After a run on 3dmark (where I got 101c on gpu 1), I didn't go over 91c this time, however gpu2, which I don't see go over 72c actually hit 76c at one point. I will just keep testing etc over the next few days to a week and see if the paste sets in and improves. Still the results seem a little baffling and may take a bit more testing to see where my temps consistently lie.
One thing I did notice was gpu1 temps are dropping quicker when the it's no longer under heavy load, more akin to that of GPU2. The gap is still a bit big, but if I can start getting a grasp on whether gpu1 is controllable, then I will muster up the courage to build my first water cooling system for it.
Between the general fear of my first attempt at water cooling and fear that I may be throwing hundreds of pounds to cool a card that may be faulty. It's something I have serious reservations about. So I will work to get the card to be more manageable and if I achieve that, then I will throw it under water and see where that gets me. Dammit, 7990's been selling like hot cakes, some smart cookie should bring out an aftermarket air cooler for this bahemouth! (Not going to happen I know
)