0 views collected around this technical thread.
This article explains the principles of Profile‑Guided Optimization, outlines the three‑step workflow, demonstrates practical PGO implementations for Go and C++ with code examples and devirtualization techniques, and presents benchmark results showing significant latency reduction and throughput improvement.
Profile‑Guided Optimization (PGO) collects runtime profiling data to recompile programs for higher performance, reducing branch mispredictions and improving code layout; Go gained built‑in PGO in 1.21 with typical 5 % gains, while C++ sees 15‑18 % QPS improvements and devirtualization benefits, and future work aims at deeper block ordering and register allocation.
This technical report details the background, hardware configuration, database setup, tuning results, and step‑by‑step optimization procedures—including BIOS, OS, network, container NUMA binding, MySQL CRC32 patching, and Go PGO tuning—performed to improve JED performance on ARM compared with Intel.
Go 1.21 has been officially released, introducing toolchain improvements such as GA‑ready profile‑guided optimization, new built‑in functions, several standard library packages, performance boosts, and experimental support for the WebAssembly System Interface (WASI).
This article explains ByteDance's large‑scale adoption of profile‑guided optimization (PGO), covering its principles, instrumentation and sampling methods, the automated platform built for data collection and compilation, and the resulting performance gains across dozens of critical services.
The article details OMAX’s end‑to‑end platform for large‑scale C/C++ service compilation, covering optimization flags, profile‑guided and link‑time techniques, Facebook BOLT post‑link tuning, and real‑world results that cut CPU use, latency and deployment time while shrinking binary size.
By generating an order file that places frequently executed code contiguously, the authors reduce iOS app cold‑launch page faults by about 15%, achieving roughly a 10% launch‑time improvement without code changes, using runtime hooking and static analysis to build the symbol ordering.