2025 · intermediate
Firefly - an F# compiler goes native to CPU, GPU and beyond
This talk is an exploration of how F#s new Firefly compiler, part of the Fidelity Framework, creates a seamless bridge between high-performance computing and formal verification. Discover how Firefly's direct MLIR/LLVM compilation unlocks native efficiency while F* verification ensures mathematical correctness. Get a glimpse into how this groundbreaking open core initiative enables targeting verifiable AI systems to CPU, GPU and other accelerators "on the metal" - all while maintaining F#'s tooling vaunted reputation for providing guardrails to support developer productivity.