If the socket was not pressing the CPU tight enough and there is an unequal pressure brought to bear by the HSF on the CPU which is heavy on the bottom/south side of the CPU, light on the top side (I am thinking of the big tower type HSFs with vertical heat pipes), then it is possible that there might be "less contact" on the top part of the CPU. "Less contact" in this case might mean a small gap might even be forming towards the top part of the CPU/Pins compared to the bottom part which under pressure.
The scenario I am describing is the result of a combination of factors including, the tighter spacings (and tolerances) of the pins on the new CPUs, the "loose fitting" Foxcon sockets, and top heavy HSFs that overclockers are using to cool the CPU. Note that most of these arc burns appear on the pins on the upper end of the CPU.
Again I am just throwing out my 2 cents here based on observations of pictures I have seen on this thread and in the Anandtech article.
Bookmarks