When did STI 33.8 bug originate?

I was just looking back through the GOL tests. It seems that the the STI did well on 33.8 in the January 2006 testing (1496' and 1204', 3rd and 6th in the ranking for the two samples), but in later GOL tests it began to exhibit extreme suckiness on this frequency. Anyone know which firmware was tested in Jan. 2006? One unit belonged to Roy and the other to csteph.

Update: csteph has Rev A1 M7 listed in his profile, I presume that's the one he submitted to GOL in 2006? It's definitely an early Sti.