Wow I can't believe I just read through the whole thread.
On the topic of the 93o
C reading - I'm sorry maarten12100, but I don't see how I am not to trust it and I can't for the life of me decide how you figured out it was glitched. Usually sensors have a specific hardware address and way of reporting that doesn't vary between architectures (if it does at all) hence why GPU-Z is able to perceive the reading. And it is far more likely that if the sensors weren't read (correctly) their output would be -128.0 / 0 / 128.0 or a value of similar nature, depending on the code of GPU-Z.
After all, this is a stock cooler with the single fan running at 43%. And this isn't surprising when you think about the actual die size and the heat intensity on it. Even assuming core power draw equal to the one of a Titan, the heat will be dissipated through a smaller area.
What I said and I still stand by is that this stock cooler isn't reliable for overclocking.
And a couple of pages ago I believe the nonsense came because of numbers in the chart like these:
7990: 211.5% (1200, guru3d.com)
Now, correct me if I'm wrong, but if a single 280X is 1/3 the performance of a 7990... This is like saying that a 7970 is 1/3 the performance of two 7970s paired together. Let alone:
a) 280X tested on 1080p while the 7990 drives a 1200p display;
b) clock speeds of the GPUs in the 7990.Edited by fragamemnon - 10/16/13 at 4:10am