Python Performance at Scale - Making Python Faster at Instagram

Explore techniques for improving Python performance at scale in this PyCon US talk. Dive into the architecture of large-scale Python web applications and discover optimization opportunities. Learn about targeted improvements to the CPython runtime, including inline caching, JIT compilation, and leveraging type annotations. Examine both proven and experimental techniques that have led to significant performance gains, with real-world examples from Instagram's production environment. Gain insights into common hotspots in the Python runtime and strategies for reducing overhead, ultimately achieving 20-30% improvement in production workloads and up to 7x improvement on benchmarks.

Intro
Fork Improvements
bytecodes
custom jet
performance results