Chasm is getting faster and faster with each relea...

# webassembly

[Charlie Tapping]

Chasm is getting faster and faster with each release, the latest sees around 30-40% improvement across all workloads. Internally Chasm now transforms wasm bytecode to an intermediate representation, this allows us to reshape the bytecode in a way that makes it preferable for interpretation but maintains the exact same wasm semantics. This is really important as wasm was never designed for interpretation, it was designed for staged emulation in a VM like V8 and frankly its quite awkward to interpret it in place. This release includes a bytecode fusion optimisation pass which is responsible for the performance jump, it’s on by default but can be turned off through the RuntimeConfig if you encounter any issues. There’s some breaking changes so please check the release notes here: https://github.com/CharlieTap/chasm/releases/tag/0.9.6

Next Up

• Add an optimisation pass to rewrite control flow, this is a major bottle neck in chasm currently.
• Add a plugin to generate kotlin interfaces from wasm binaries