Archanan provides a cloud-based development environment for massively parallel and HPC software. We enable single-click deployment of parallel codes into thousands of compute nodes for debugging, profiling and visualizing memory maps. Archanan bridges the gap between the software engineering process on laptops or workstations and the deployment on multi-million dollar supercomputers to gather results.
It is common for large-scale software developers to write their code on workstations, equipped with 10-20 cores. The problem is that this code is meant to run on supercomputers with up to hundreds of thousands of cores. Once it becomes time to test scaling, developers migrate their code to a supercomputer and finish debugging there. Around a third of a supercomputer's utilization can be attributed to debugging workloads. This means that many compute cycles become unavailable for production workloads to run. By offloading debug jobs to the cloud, we enable computing centers to grant research groups more resources for more time. Additionally, our platform gives developers the opportunity to write their code on large-scale systems from the start. By providing state-of-the-art debuggers, profilers, and memory map visualizations, we help developers create better quality code more efficiently.
Other platforms which offer solutions for HPC in the cloud focus only on running production workloads. Archanan leverages the cloud to create a complete development suite and environment for properly testing large-scale codes before deployment on supercomputers.Download the One-Pager