Результаты acovea

Вот получил такие результаты:

generation 20 complete, average fitness: 2.90007

Acovea completed its analysis at 2007 Aug 20 22:57:01

Optimistic options:

                                  -fgcse  (2.419)
                       -fschedule-insns2  (1.662)
                       -fstrict-aliasing  (1.51)
                              -ftree-pre  (1.712)
              -maccumulate-outgoing-args  (1.611)

Pessimistic options:

           -fno-guess-branch-probability  (-2.173)
                      -fno-if-conversion  (-2.224)
                      -fno-loop-optimize  (-1.769)
                        -floop-optimize2  (-1.87)
                           -fno-tree-lrs  (-1.669)
                            -fno-tree-ch  (-2.224)
                             -fforce-mem  (-2.022)
                            -mfpmath=387  (-2.123)
                            -mfpmath=sse  (-1.769)
                        -mfpmath=sse,387  (-1.618)

Acovea's Best-of-the-Best:
gcc -lrt -lm -std=gnu99 -O1 -march=opteron -fno-merge-constants -fno-thread-jumps -fno-cprop-registers -fno-if-conversion2 -fno-delayed-branch -fno-tree-ccp -fno-tree-ter -fno-move-loop-invariants -fcrossjumping -fgcse -fexpensive-optimizations -fstrength-reduce -frerun-loop-opt -fcaller-saves -fforce-addr -fschedule-insns -fschedule-insns2 -fstrict-aliasing -fdelete-null-pointer-checks -freorder-blocks -fthread-jumps -fgcse-lm -fsched-interblock -fsched-spec -falign-functions -falign-loops -ftree-pre -fno-inline -ftracer -funroll-all-loops -fbranch-target-load-optimize2 -fmodulo-sched -fgcse-las -ftree-loop-im -fvariable-expansion-in-unroller -mno-push-args -minline-all-stringops -o /tmp/ACOVEA2AB92FB4 /usr/share/libacovea/benchmarks/huffbench.c

Acovea's Common Options:
gcc -lrt -lm -std=gnu99 -O1 -march=opteron -fno-tree-ccp -fno-tree-ter -fcrossjumping -fgcse -fexpensive-optimizations -fcaller-saves -fforce-addr -fschedule-insns2 -fstrict-aliasing -freorder-blocks -fthread-jumps -ftree-pre -funroll-all-loops -ftree-loop-im -fvariable-expansion-in-unroller -mno-push-args -o /tmp/ACOVEAEC59C9A7 /usr/share/libacovea/benchmarks/huffbench.c

-O1:
gcc -lrt -lm -std=gnu99 -O1 -march=opteron -o /tmp/ACOVEAD05C904A /usr/share/libacovea/benchmarks/huffbench.c

-O2:
gcc -lrt -lm -std=gnu99 -O2 -march=opteron -o /tmp/ACOVEA62DD1876 /usr/share/libacovea/benchmarks/huffbench.c

-O3:
gcc -lrt -lm -std=gnu99 -O3 -march=opteron -o /tmp/ACOVEAEE113350 /usr/share/libacovea/benchmarks/huffbench.c

-O3 -ffast-math:
gcc -lrt -lm -std=gnu99 -O3 -march=opteron -ffast-math -o /tmp/ACOVEA4BD6D278 /usr/share/libacovea/benchmarks/huffbench.c

-Os:
gcc -lrt -lm -std=gnu99 -Os -march=opteron -o /tmp/ACOVEAF8CAEBC0 /usr/share/libacovea/benchmarks/huffbench.c


A relative graph of fitnesses:

     Acovea's Best-of-the-Best: ****************************                          (2.48365)
       Acovea's Common Options: *****************************                         (2.56999)
                           -O1: *********************************                     (2.89893)
                           -O2: ********************************                      (2.85918)
                           -O3: ******************************                        (2.6833)
               -O3 -ffast-math: *******************************                       (2.71792)
                           -Os: **************************************************    (4.36609)

Acovea is done.

Вопрос такой, как их интерпритировать, т.е. , к примеру, какой из -Ox 
лучше, тот у которого больше значение или у которого меньше?! 

Доку читал, но чё то не понятно..

Заранее спасибо!!!