According to a reliable guy at HFR this CPU has 16% lower
TDP at 5ghz/1.552V than an usual 5ghz capable 8350 that would
be overclocked with theses settings.
So much for thoses who pretend that it s an ocked 8350.....
I don't quite understand what you are trying to get at here.
This guy, and I'll take your word for it, is comparing one chip to another and observing a difference in power consumption.
How is that news? How is that proof of anything related to your last sentence?
Power consumption is leakage dependent as well as dynamic power dependent.
Dynamic power will be identical for an 8350 OC'ed to 5GHz and an FX-9590 clocked to 5GHz (provided they are operating at the same voltage and temperature). There are not layout differences, no xtor topology differences, no mask set differences, between the FX-8350 and FX-9590.
However the static leakage may very well be quite different between the two specific processors this guy tested, as it is different for any two processor period.
Leakage at the device level is all about mask alignment at the various litho steps during the creation of the different layers of the IC. Some wafers get better alignment (simply a random occurrence) and as such the chips on those wafers will have markedly superior insulation between conductors (wires, vias, contacts, and gates).
Those chips, be they binned as 8320's or 9590's, are going to have the same dynamic power consumption but far better (lower) static leakage in comparison to chips from other wafers in which the alignment was less than optimal (but still within spec and thus were not reworked at litho).
All that the cited results tell us is that AMD is binning the 5GHz chips for having better electrical properties versus those of the FX-8350's...and who isn't expecting that to be the case? Of course AMD is going to bin out the better silicon for the higher clocked SKU, that is what binning exists to accomplish.