https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=256594
--- Comment #17 from Mark Johnston <[email protected]> --- (In reply to Andriy Gapon from comment #16) Yes, I've observed this too when measuring TCP ping-pong latency over loopback - latency is higher if the core was recently idle, I guess because autonomous CPPC gradually ramps up the awoken core's performance level. Running a background loop on the core results in lower latency. Unfortunately we don't have any driver (yet) to control CPPC on modern AMD systems. -- You are receiving this mail because: You are the assignee for the bug.
