Julia: Making dynamic programs run fast
Abstract: Julia is a general purpose language that utilizes type inference combined with a Just-In-Time (JIT) compiler, built upon LLVM, to achieve C-like performance while retaining the dynamic semantics of interactive languages like MATLAB, Python, and R. This talk will briefly introduce Julia, explore how it is possible for a dynamic language to be fast and to how Julia is making use of LLVM. We will focus on interacting with the compiler infrastructure and explore how to Julia targets accelerators like NVIDIA GPUs.
Bio: Valentin Churavy is a PhD student at CSAIL. His research at the JuliaLab focuses on heterogeneous distributed computing for high-performance computing.
He is interested in human-compiler interaction and how to help people write fast code in a high-level dynamic language.