Gprofng is a modern sampling-based call-graph profiler (in the lineage of GNU gprof) for native Linux applications. It collects periodic program-counter samples to build statistical call graphs, letting you pinpoint where code spends time with far less overhead than heavy instrumentation.

Gprofng is characterized by three key components:

• Sampling-based collection: periodic samples reduce runtime overhead while revealing hotspots.
• Call-graph generation: aggregates caller/callee relationships to prioritize optimization targets.
• Integration with tooling: works alongside build toolchains and other profilers for analysis (symbol resolution, flame graphs).

Profilers like gprofng let developers find performance bottlenecks and quantify the impact of optimizations. On Arm servers (e.g., Ampere Altra), profiling is essential to tune compiler flags and threading for best performance-per-watt on the target architecture.

• perf Linux profiler: kernel perf wiki
• GNU gprof docs
• Performance tuning on Arm servers
• Flame graphs: profiling visualization