You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
So I recently read about BOLT, which seems like an interesting concept. It could have significant performance gains for certain applications, for example their 'test case' was running it against clang compiled with PGO & LTO, with a healthy performance improvement. Could something like this be implemented?
The text was updated successfully, but these errors were encountered:
Post-link optimizing sounds interesting indeed (especially if need on-the-fly optimization for a specific task without recompiling every times), but has same problem as any profile-based optimizations, it's hard to automate for a general use-case unless the package already provides some mean to collect perf data in a relevant way (aka pgo ebuilds).
So I recently read about BOLT, which seems like an interesting concept. It could have significant performance gains for certain applications, for example their 'test case' was running it against clang compiled with PGO & LTO, with a healthy performance improvement. Could something like this be implemented?
The text was updated successfully, but these errors were encountered: