The talk discusses running MPI applications on a Unikernel, specifically the Toro Unikernel, an application-oriented Unikernel that offers an API for efficiently deploying parallel applications. The Unikernel has three building blocks: per-core memory, a cooperative scheduler, and core-to-core communication based on the Buezo specification. The talk also includes a benchmark of the current implementation using MPI barrier microbenchmark.