The one roll for both method would do as Vroom said above. . .you'd be chasing that high DB monster around for a while before you hit with an 90+ roll. . .and kill it.
running HARP, I have noticed that odds wise the results heavily cluster to "Miss" or "Max result for attack". . .I'm not saying they miss or max, but that in 100 attacks, Miss or Max will be far more common than any other result, with the rest of the results evenly distributed between miss and max, and you will likely see every result at least once in 100 attacks. . .With RM's tables, miss will again be the most common result, but then the distribution from there wildly varies. . .this is due to the number of possible results:
One weapon, assuming any armor, and lowballing by saying only half of all hits cause crits in RM:
~20,000 RM2 AL 1989
~10,000 2003 AL
~900 RMX
95 Martial Law
19 HARP Core
Assuming only one AT
~1,000 RM2 AL
~500 2003 AL
~275 RMX
95 martial law
19 HARP Core
One could say the hits damage makes no difference, only the crit matters, then you get:
95 RM2
95 2003 AL
95 Martial Law
19 RMX
19 HARP Core
Though the existance of things like I criticals in RM2/c and RMSS, which resolve as EDB, would give:
6,859 RM2
6,859 2003 AL
95 Martial Law
19 RMX
19 HARP Core
But if you don't like seeing the same result over and over, something like Martial Law (in which you use the 1s die to get the location) is radically more diverse than HARP core. . .and if you ignore hits, is more diverse than RMX. . .and ignoring spells comes in a tie with RM's method. . .
Vroom, this conversation knocked loose some dust in my brain. . .there was a one roll RM method I'd tried in the past that worked fairly well. . .inverted die. . .so a 31 attack roll is a 13 crit. . .a 79 is a 97 crit, etc etc. . .avoids the "miss or kill" issue of using only one roll. (That would put RM on a complexity par with Martial Law, assuming the crit tables are in a spread across from the attack table).