I’ve got a hellacious project due and finals all next week, but this was just too much fun to pass up. In any case, compiler optimization increases compile time, and anything that gives me more time to sword fight on $1000 office chairs is worth a little R&D.
I’m working on writing this cut-down MIPS processor simulator for my Computer Organization class at UCF. I googled “word alignments” to help me better understand the most efficient calculations for converting the Byte Aligned Program Counter address to the Word Aligned Memory array when I ran across an interesting article showing that the mere *order* of variable declarations in a C program can affect the amount of memory used by that program.
The article explained the situation very well, and it makes sense why this issue would happen, but I was surprised that the compiler wouldn’t try to optimize situations like this by re-ordering a set of concurrent variable declarations of alternating data types.
In any case, I continued to hack away at my project when I began to think about whether or not gcc translates multiplication and division operations where one of the operands is a power of 2 into simple
. . . → Read More: gcc Optimizations for Arithmetic Operations using Bit Shifts
