I should point out don't expect anything anywhere near what some of the guys posted on the first few pages...as those are the early early Venices that were hitting 3+ghz on air at <1.4vcore
man I'm still blown away by what the hype and results that were around the venice back in april
I honestly can't tell you the reason why, however since this trend also occured in the Winchester core
I suspect that rather than risk producing fewer, more accurate cores...AMD "overshoots" their target at early production stages which basically means they produce a 3000+ in the same fashion as they normally would a 3500+ and then underclock it...that way they have higher yields and have fewer underperforming cores...and the 3500+ may be produced as if it were a 3800+ etc.
Basically what I'm saying is a 3000+ in April may have infact been an underclocked 3500+...and over time AMD is able to produce more consistant yields by perfecting their production process and thus doesn't have to overshoot anymore