I suspect that "roll - calculate - roll a crit if needed" would end up being as slow or slower than "roll - look up result - roll crit if needed" in actual use, based on my experiences lookup tables save time over calculations on the fly, unless you can't find the #$@% table. . .
It does also tend to drastically reduce the result spread.
The two table method is the number of rows on the attack table that don't have a critical + (The number of rows that do have a critical x the number of rows on the critical table)
That's X + (Y x Z) for easy talk sake.
Having a multiplication in there means the reduction of variation of removing X is limited, but pulling Y or Z out of play causes the number of potential results to crash. . .
There's a thread around here somewhere where we did the math on it, and the contrast is dramatic. . .and anecdotal comments from people who have played 2 roll RM and any of the lower variation versions like HARP, RMX or the Combat Companion rules have a lot of comments in the line of "I keep seeing the same result over and over and over again."
Some of the HARP combat variations offer more variety, but in the end, the 2 roll method in RM offers the highest level of variability.
Some people love the lower variation results, so likely it's a YMMV thing, what suits your personal preferences.