Protobuf is an extremely popular binary data interchange format. This session dives into hyperpb, a Protobuf parser for Go that uses every trick in the book, and some new ones, to achieve the highest throughout possible. We’ll talk about table-driven parsing, the parser paradigm used by hyperpb, and hyperpb’s innovations on the concept. We’ll also take a look at the high-level architecture of hyperpb, including its compiler, parser VM, and layout optimizations.
Miguel is a compiler engineer with experience in language design, performance optimization, and standard library implementation. He has worked on high-performance Protobuf C++ and Rust at Google, and is obsessed with getting the most out of hardware OEMs' parts. He also thinks that performance should be extremely accessible and frequently writes on the topic. Miguel is currently at Buf where, unsurprisingly, he focuses on Protobuf performance, but also high-quality developer tooling (including Buf's Protobuf compiler frontend). Outside of software, Miguel is an experienced digital illustrator and painter.
