System-wide sampled record perf
Sample whole machine at 99Hz for 30s, then print collapsed report; 99Hz dodges lockstep aliasing
perf record -F 99 -a -g -- sleep 30 && perf report --stdio
more perf
all 17 commands →
Flame graph pipeline
perf script | stackcollapse-perf.pl | flamegraph.pl > flame.svg
Detailed counter stats
perf stat -d -d -d -- ./app
Scheduler latency analysis
perf sched record -- sleep 10 && perf sched latency --sort max
Syscall trace via perf
perf trace -s -p <pid>
Dynamic probe on function
perf probe --add 'tcp_sendmsg size' && perf record -e probe:tcp_sendmsg -a -- sleep 5