the value of a performance oracle

This article examines the performance of tail-calling versus switch-based interpreters, specifically Matt Keeter's Raven VM implemented in Rust. The author replicates Matt's experiments, testing native compilation, Wasmtime, and Wastrel runtimes. Results show tail-calling interpreters outperform switch-based ones natively, but underperform in Wasmtime due to JIT limitations. However, Wastrel demonstrates that Wasm can achieve competitive performance, suggesting the issue is not inherent to WebAssembly but rather an implementation accident. The article delves into assembly analysis to explain performance differences.