S³ School 2026 - Profile and improve Code Optimization - Karl Kosack скачать в хорошем качестве

S³ School 2026 - Profile and improve Code Optimization - Karl Kosack

How do you make Python code faster (without breaking everything)? 🔧📊 In this S³ School 2026 lecture, Karl Kosack walks through practical strategies to optimize code the right way: identify bottlenecks, improve algorithms, leverage NumPy vectorization, and (when needed) use JIT compilation with Numba—while keeping an eye on memory and correctness. What you’ll learn When optimization is worth it (and when it’s not) Moving from Python loops → NumPy vectorization Benchmarking properly (repeatable timing, comparing variants) Using Numba JIT: speedups, compilation overhead, caching, pitfalls Parallelization considerations (threads, data hazards) Why performance can become a memory bottleneck (and how to notice it) Links & resources Slides & materials (S3 2026 Lectures): https://s3-school.github.io/s3-2026-l... School resources page: https://indico.in2p3.fr/event/36319/p... GitHub organization: https://github.com/s3-school About S³ School S³ School (Sustainable Scientific Software School) is a training event focused on research software, best practices, and sustainable development workflows. CTA : If this helped you, like, subscribe, and share this lecture with someone who cares about performance and research software. Tags: S3 School, S3School, Sustainable Scientific Software, Research Software Engineering, RSE, Code Optimization, Performance Engineering, Profiling, Python Performance, Python Optimization, Numpy, NumPy Vectorization, Numba, JIT Compilation, Benchmarking, timeit, Memory Profiling, Bottleneck, CPU Cache, Parallel Computing, Multithreading, Scientific Python, High Performance Computing, HPC, Software Sustainability, Reproducible Science