GCC 3.4/3.3 Optimization Flags
I need to optimize some CPU-Intensive Real-Time software I have running, but I have a few optimization flags missing, those would be :
-falign-functions
-falign-labels
-falign-loops
-falign-jumps
I have the correct optimization flags for my Athlon-XP which are :
-falign-functions=16
-falign-labels=1
-falign-loops=16
-falign-jumps=16
wrc1944 wrote :
"The =16 values set things to corelate precisely with the athlon-xp's design
parameters, which I would think are pretty important. This is said to allow
the on-die cache to be fully utilized and filled with 16 byte segments."
it's true enough to get my doom3 performance up by about 3 FPS.
however I need to know the same stuff (if available) for the other processors
but I can't find any similar data for the Pentium 3 and the Pentium-MMX.
I need to know their architecture-specific optimization parametrs to tell gcc to compile with.
|