3 ms·
Why would you use this over the profile guided optimizations that gcc and clang already offer, that use the same data? What kind of perf benefits can you expec
by fefe23 7y ago
Why would you use this over the profile guided optimizations that gcc and clang already offer, that use the same data?
What kind of perf benefits can you expect?
What kind of application have you tested this with?
- ahupp 7y agoFrom the paper: "For datacenter applications, BOLT achieves up to 7.0% performance speedups on top of profile-guided function reordering and LTO." This is used for large binaries, like hhvm. There's a lot more details here: https://research.fb.com/publications/bolt-a-practical-binary-optimizer-for-data-centers-and-beyond/ https://research.fb.com/publications/bolt-a-practical-binary...