# Comparative benchmarking of cloud computing vendors with High Performance Linpack

## TL;DR

This paper benchmarks various cloud computing providers using the High Performance Linpack to evaluate their performance for high-performance computing tasks, comparing single-node and distributed scenarios, and assessing their viability for HPC applications.

## Contribution

It offers a comprehensive comparison of cloud providers' HPC performance using Linpack, including hyper-threading effects and core-level performance, providing a ranking framework.

## Key findings

- Cloud providers' performance varies significantly.
- Distributed memory calculations scale with up to 32 nodes.
- Cloud HPC performance approaches traditional supercomputers in some cases.

## Abstract

We present a comparative analysis of the maximum performance achieved by the Linpack benchmark on compute intensive hardware publicly available from multiple cloud providers. We study both performance within a single compute node, and speedup for distributed memory calculations with up to 32 nodes or at least 512 computing cores. We distinguish between hyper-threaded and non-hyper-threaded scenarios and estimate the performance per single computing core. We also compare results with a traditional supercomputing system for reference. Our findings provide a way to rank the cloud providers and demonstrate the viability of the cloud for high performance computing applications.