Speed and Performance
Harp has a wide array of architectural improvements that contribute to high efficiency gains.
Harp is an HPC-Cloud convergence framework for Big Data. Our overarching goal is to automate ML as a service for both ease of use and scalability. Harp is designed to cover a full range of data-intensive computation from pleasingly parallel to machine learning and simulation.
Harp has a wide array of architectural improvements that contribute to high efficiency gains.
Harp is a plugin built into Apache Hadoop and hence no code change of Hadoop is required for migration.
Easily debug and identify the issues, allowing faster iteration during development and easy to use.
Harp is highly scalable in the ability to run on large amount of training data and model parameters. It can run on both cloud and HPC platforms.