FlameGraph Visualization Tool

FlameGraph is a visualization tool for profiling software performance, created by Brendan Gregg (a renowned performance engineer formerly at Netflix and Intel). With over 17,000 stars as of 2026, FlameGraph has revolutionized how developers analyze and optimize performance by popularizing the flame graph visualization technique. A flame graph displays the call stack of a program as a set of stacked bars, where the x-axis represents the number of samples (proportional to time spent) and the y-axis represents the call stack depth, making it immediately clear which functions consume the most CPU time. Key features include: stack collapse (the stackcollapse tools converting raw profiler output from various formats including DTrace, perf, eBPF/bcc, Instruments, XCode, VTune, and gdb into the folded stack format used by flamegraph.pl), flame graph generation (the flamegraph.pl Perl script producing interactive SVG flame graphs with color-coded stacks, hover tooltips showing function names and sample counts, and click-to-zoom functionality), differential flame graphs (comparing two profiles side by side using difffolded.pl and flamegraphdiff to identify performance regressions or improvements between code versions), multiple color palettes (hot palette for CPU profiles, red palette for memory allocations, blue palette for system calls, and green palette for I/O wait time), CPU profiling via perf (collecting stack traces at fixed intervals using Linux perf_events for CPU profiling), off-CPU profiling (visualizing time threads spend waiting rather than executing, using eBPF for lock contention and I/O wait analysis), memory profiling (tracking allocation sites and heap growth via USDT probes and eBPF), language-specific profiling (Java via async-profiler and JFR, Python via py-spy, Node.js via perf and --perf-prof, and Ruby via rbspy), and microstate accounting.