To get a call graph of the benchmarks with a percentage runtime breakdown:
gcc -g mips.c valgrind --tool=callgrind ./a.out kcachegrind callgrind.out.606