MSI GTX580 Twin Frozr II High Temps [Resolved]

I finished my first build about a fortnight ago. Specs are in my sig. Nothing overclocked. I am worried about my GPU temperatures. I don't know my ambient in the room while doing the tests but I'm in Ireland in October and the window is open. The ambient is no more than high teens/20ºc

I was hitting 91ºc during the BF3 beta with everything maxed out at 1920x1200. The card idles at around 35ºc.

I was hitting over 100ºc with some tests in Furmark and 3Dmark2011 and got them down to the low 90's when I ran with the side panel off and a 120mm fan aimed at the card and changed the afterburner fan profile (40%@40ºc etc)

I figured, "this ain't right", and was going to get straight to the nitty gritty in this thread and ask about RMA'ing or re-doing the TIM with IC7 etc. However, then I read threads where some say the temps are normal, some say they are not. For all I know all they guys saying 90ºc is insane for a 580 Twin Frozr II who get 65-70ºc with theirs are running the 720p Furmark test with god knows what settings while I am running the 1080p test so no wonder my temps are high.

So can someone tell me what tests to run at what resolution and with what settings so that I am definately working from the same baseline as everyone else for my temperature comparisons.

I know I have work to do in modding my Corsair 600T case to improve airflow for these internally venting Video cards but am I right in thinking I can put case airflow issues aside for the moment as a cause of high GPU temps when I am running these tests with the side panel off with a 120mm blowing directly on the card and still getting what look like abnormal temps.
Edited by Calibos - 10/31/11 at 3:56am

Both fans are running and I was getting high 90's and 100ºc with default afterburner profile side panel on and no extra fan aimed at card. Side panel off, 120mm fan aimed at card and afterburner fan profile set to 40%-40º, 60%-60ºc, 100%-80ºc etc I dropped the temps to 91ºc for the 1080p benchmark on furmark.

Why thermal pads??

There is paste on it at the moment. In fact I can see more paste squeezed out on one side of the GPU/Heatsink interface than I would have thought was normal to put on a card in total in the first place if you get me. ie. if I scraped this squeezed out excess off, I am sure I could make a blob bigger than the size of a pea with it? Is a pea not the recommended amount for a GPU?

Again, can I confirm we are all on the same page with regard to the testing parametres. After making a custom afterburner fan profile and adding a 120mm fan blowing on the cards, with the furmark 1080p test I am getting 91ºc and with the 720p test I am getting mid to high 70's.

My seemingly normal Idle temperatures of about 35ºc with an ambient < 20ºc are what has me wondering if talk of RMA'ing is premature. Thats why I really really need to ensure I am testing with the same settings/parametres as the rest of you guys. So can someone please tell me what benchmark with what settings at what res you guys are using so that we are comparing like with like.