Caleb Winston is a co-creator of Banyan and the creator of Emu, a GPGPU framework for Rust. In the past, he has worked on GPU-accelerated data science at NVIDIA and virtualization for wet labs at the Molecular Information Systems Lab at the University of Washington, Seattle. His primary interests are in making computing at both massive scale and the edge accessible for anyone to easily program.
Running Julia code at scale can be quite difficult. First, running at scale typically requires rewriting a significant amount of code to use libraries such as MPI.jl, Dagger.jl, Spark.jl, etc. Second, it requires developing a strategy for deploying a cluster either on-premise or in the cloud. In our talk, we discuss how the Banyan suite of libraries enables instant offload of existing code to multi-node clusters in the cloud with maximal utilization of RAM and CPU cache.