Blogs

• Flame Graphs
• How to read icicle and flame graphs
• Sampling vs Tracing
• C++ performance tools

Books

• System Performance Enterprise and the Cloud
• BPF Performance tools

Continuous Profiling

• parca
• parca-agent - based always-on profiler auto-discovering targets in Kubernetes and systemd, zero code changes or restarts needed! Supports multiple languages: C/C++, Rust, Go, Python, Ruby, Java, etc.

CPU specific

• Intel GPA
• ARM profiling
• Intel Vtune

C# specific

• Unity profiler

C++ specific

• Callgrind
• Tracy

General utilites

• https://ebpf.io/
• Dtrace
• gprof
• Mac OS instruments
• htop
• google perf tools
• Heaptrack
• jemalloc
• Renderdoc
• Magic Trace
• pprof
• Samply
• ETW
• collectl
• Windows Perf Analyzer

GPU specific

• NVIDIA visual profiler
• NVIDIA tools
• NVIDIA NCU
• pynvml - smi` for your code with deeper level instrumentation
• Ingero - based GPU causal observability agent. Traces CUDA Runtime/Driver APIs via uprobes and host kernel events via tracepoints. Builds causal chains explaining GPU latency with full Python-to-CUDA stack traces.
• GPU View
• ROC profiler
• Omniperf

Java specific

• Java visual VM

Mobile specific

• iOS
• Android

Python specific

• Pycharm profiler
• TAU
• viztracer
• psutil
• pyinstrument
• pycallgraph
• py-spy
• line profiler
• palanteer
• gprof2dot
• snakeviz
• scalene
• pprofile
• austin-python - level very low overhead time & memory profiler with web & terminal UI
• py-perf - overhead, sampling CPU profiler for Python implemented using eBPF.
• oracletrace
• yappi

PyTorch specific

• Pytorch profiler
• PyTorch memory profiler

Ruby specific

• rbspy
• rbperf - overhead sampling profiler and tracer for Ruby implemented in BPF
• vernier

Talks

• scalene talk
• Common performance anti-patterns
• Why performance matters

Understanding code structure

• pyreverse
• pdb
• IntelliJ UML Class diagrams

Web specific

• Chrome profiler