Writing Optimized Solana Programs
summary
Claims concrete CU and size wins: an Anchor example drops from 649 to 281 CU by smaller discriminators and manual zero-copy deserialization; a Pinocchio implementation runs ~109–108 CU; hand-written BPF assembly hits ~105 then 104 CU when skipping checks. A tiny deployed binary reaches ~928 bytes (<1 KB) after a custom linker, and the live-coded assembly logs instruction data using R0/R1/R2 hacks. Tools cited: Pinocchio, zero copy, custom linker.
AI-written summary · about
chapters (5)
- 0:00The three pillars of Solana
- 2:00Balancing optimization
- 3:02Practical optimization hacks
- 8:02Framework benchmarks
- 9:53Assembly and final results
related talks
55mCode Size Compiler Optimizations and Techniques for Embedded Systems
01h 00mKilling C++ Serialization Overhead & Complexity
01h 00mJust-in-Time Compilation
38mOptimized String Processing in RISC-V: How Toolchain Improvements Can Boost Performance
32mSmall is beautiful: Techniques to minimise memory footprint
01h 05mFast and small C++ - When efficiency matters
17mThe Pod: IBRLing Solana w/Quasar
33mA Deep Dive into Binaries