I did a test (Tiger) with 1050MHz on gpu's, got a result.
I then upped the clock to 1085MHz, and got a better result witch I uploaded to boards.
When I found it on the board, the clock was 1050MHz on the board, witch is lower than what I accualy used.
Allmark should do a new reading when benchmark is started imo, so we don't confuse ppl with high scoores on missleadingly low clock imo.
Other than that, the nev version seem more stable on my setup, keep up the good work