Supporting HPC Workloads on a Common Platform

Many organizations are expanding their definitions of high performance computing (HPC) beyond simulation and modeling workloads to include artificial intelligence (AI) and analytics. Intel is working on a number of fronts to facilitate that transition, such as contributing optimizations for Intel® architecture to open source projects, co-engineering with other members of the ecosystem, and developing reference architecture solutions. This solution brief also introduces capabilities for running TensorFlow* and Apache Spark* on existing HPC systems using standard batch schedulers.