First of all thanks for the reply.
Let's go off topic for a moment...
I cannot see any sensible reason for using such a small PSU when the price and running costs are not going to differ substantially. I would recommend swapping the power supply out for a larger one (300W ish) of good quality from a manufacturer like Seasonic, CWT, Enermax, etc.
Well, I can give you 2:
1)Space. If you are not familiar with that specific case please google it. There is no way to fit a "normal" psu in there and if i wanted a full size pc (which i already own anyway, 2600K, GTX580, AX850 and so forth) i would have definitely not gone this route. Space is vital to me.
2)Power consumption. Get a 300W with 80% efficiency and stick it to a pc with average consumption of 25W. Now you have a psu that operates at 10% of it's nominal load. Guess how efficient it is at this level (and calculate how many watts are being consumed by the psu itself). Now do the same for the maximum consumption under the heaviest load, say 55W. Again, we are in the 20% of the nominal wattage territory. Guess again how efficient is the same psu at this level. So again, if i wanted an average pc i wouldn't care about power consumption too much. Still, this pc is for very specific usage like downloading for days etc etc.
Btw, consumption is another reason i don't prefer the latest firmwares. Compared to the F2 they consume 3-4W more at least when idle with no apparent reason (=honestly i can't say the difference between the firmware versions, never had instabilities with F2 or other issues).
Back on topic...
Even though you seem assured that it is not a power supply issue I would still be incredibly surprised if that is not where your problem lies. Everything points to an under powered scenario. Even Gigabyte recommend a 200W PSU for this motherboard. I would agree that is probably overkill but better to over than under estimate.
Page 22 of the manual ( http://download.gigabyte.eu/FileList/Manual/mb_manual_ga-h61n-usb3_e.pdf )
To meet expansion requirements, it is recommended that a power supply that can withstand high power consumption be used (200W or greater). If a power supply is used that does not provide the required power, the result can lead to an unstable or unbootable system.
Gigabyte's' recommendation could be for the company to get it's back covered because the company doesn't know how exactly you are going to use it (you could stick a G530 or you could stick a 2700K in there, with or without an extra vga, one or more storage devices and so forth). I know that the G530 has a TDP of 65W but in practice that's far from reality if you fine tune your system(=find the optimal Vcore for a specific frequency, see PS1). Furthermore there are plenty of people in the net that have pcs with the same configuration like me (same case+psu, same cpu) and they are fine with it...
These symptoms are only when sometimes (NOT all the time) switching the pc on. I did a little test the other day, running Linx AVX, Furmark and HD Sentinel self test (on the hdd) all at the same time to get the absolute maximum power consumption of my system. Never exceeded the 52W. So is it possible to require more than 52W the moment you are switching it on? The kill-o-watt (while like most of the consumer instruments is a bit "slow" to track peak wattage) has never showed more than 40W (and that's peak, just for a second then it goes back down to the 20ies range) when posting.
I have 25 years of experience with pcs, half of these are professionally. I think i know the symptoms of a cold boot or/and inadequate psu and that's not it.
If it was the psu then i'd either dealing with cold boot symptoms (which is not the case) or i would dealing with several instabilities under full load (reboots, shutdowns and lot's of BSODs especially under heavy stress testing and again, that's not the case).
Now, since everything have been tested or at least have been measured in my system and since this phenomenon didn't bothered me for several months i had a new theory which has nothing to do with the psu, firmware version and so forth.
Troubleshooting took me almost a month (started testing it late July, that's why i didn't post earlier) and i verified that my issue has to do with number of memory sticks/motherboard slots (the memory has been tested several hours via mem86, they are clean and they are in the ram compatibility list for this mobo).
2 sticks of ram will give me these phenomenon at least within the first week (randomly, no patterns found, could be day #1, #5 and so forth).
1 stick of ram and everything runs fine for more than two weeks.
And if anybody thinks it's still the PSU, each stick of ram (Kingston 2Gb running at 1066Mhz) consumes 1-2W.
Also when my pc gets in this infinite "switch on-->no post-->switch off-->switch on" infinite loop, when i finally manage to get it to post properly it reports that some settings in bios have changed (while i haven't changed anything) etc. It's not the battery, that have been tested too.
Anyway, i'll try to RMA the board today.
Thanks again for your reply and for your time.
PS1: keep in mind that my Vcore is set to be 0.840V that's why the consumption is that low. Yes stability has been tested since day 1, multiple runs of 12 hours Prime95 26.6 (27.7 wasn't around at this time), couple dozens of Linx AVX runs (although the cpu doesn't support AVX), several hours of Prime95 27.7 (couple months ago) etc etc.
PS2: anybody who reports the same probs, please give the full configuration of your pc, report each component otherwise there's no way to anybody to find a link or a pattern.
PS3: to the ones that may doubt about the stability after the Vcore fine tuning. Running Vccre at auto doesn't make things any better or worse, tried that as well.