12 replies

Optimizing a bubble sort implementation in C for an x86-64 architecture

Hello , I'm working on optimizing a bubble sort implementation in

for an

x86-64

x86-64

architecture specifically targeting an Intel Core i7 processor using

GCC 11.2

GCC 11.2

. I noticed that the

-O3

-O3

flag resulted in slower performance compared to

-O2

-O2

when sorting large arrays of integers 1 million elements in this case.

Here are the timings, average of multiple runs:

Solution

Compile your code with profiling enabled using

Copy code
gcc -pg -O3 -o sort sort.c
./sort 1000000
gprof sort gmon.out > analysis.txt

Copy code
gcc -pg -O3 -o sort sort.c
./sort 1000000
gprof sort gmon.out > analysis.txt

For

perf

perf

, use

perf record -g ./sort 1000000

perf record -g ./sort 1000000

Generate a report to get full detail of where time is being spent on your program

perf report

perf report