3 ms·
Compiling for your target is good, not using the PLT for mem* functions helps a ton, feedback-directed optimization and LTO and/or post-link optimization helps
by jeffbee 5y ago
Compiling for your target is good, not using the PLT for mem* functions helps a ton, feedback-directed optimization and LTO and/or post-link optimization helps the most. For CPU-bound workloads of course.