Hi all.
Come across this post last night when i was having the same issue. Before i get stuck in hardcore to fix something i always look on a google search to see if others had the same problem, it might just save me a heap load of time.
This is me giving some back to hopefully help others.
I fixed my reboot loop last night, as odd as this is going to sound it worked for me.
I had almost given up and was considereing firing up the test gear to find out what was going on when i had a thought.
I disconected the HDD front panel LED and the case Reset button. It worked and fired up, 24 hours later and 4 restarts/power downs and no problems what so ever.
I am using Hynix memory and Xeon 2620's so my error is not going to be relevant to ES(Engineering Sample) Xeons.
You should not be using these ES chips for proper use anyway. They can corrupt your data due to design and or manufacture faults. They are only used as a sample during production so they dont require everything to work 100%
SuperMicro dont list the current(I) or even Voltage ratings for case LED's nor do any other manufacturer that i have seen so it is hard to buy replacement or moded parts for a case. That said i love the bios post output on these supermicro boards. it can really speed up diagnosing a fault.
So yea, i fixed my problem by disconecting the HDD LED and Reset switch. It always seemed to be running much brighter than other LED's so something was wrong. It is hard to test for a intermitent switch failure as you kind of need it to fail to test it so when it works you think its ok then a month later it gives you the run around. My reset switch currently tests ok but seen as it all works while both reset and HDD led are disconected I am happy to leave it at that.
Hope it helped someone else, stupid and simple i know but that is what seems my fault was.