Work

OpenRAND: A Performance Portable, Reproducible Random Number Generation Library for Parallel Computations

Year: 2023

Type: preprint

Abstract: We introduce OpenRAND, a C++17 library aimed at facilitating reproducible scientific research through the generation of statistically robust and yet replicable random numbers. OpenRAND accommodates single and multi-threaded applications on CPUs and GPUs and offers a simplified, user-friendly API that complies with the C++ standard's random number engine interface. It is portable: it functions seamlessly as a lightweight, header-only library, making it adaptable to a wide spectrum of software and hardware platforms. It is statistically robust: a suite of built-in tests ensures no pattern exists within single or multiple streams. Despite the simplicity and portability, it is remarkably performant-matching and sometimes even outperforming native libraries by a significant margin. Our tests, including a Brownian walk simulation, affirm its reproducibility and highlight its computational efficiency, outperforming CUDA's cuRAND by up to 1.8 times.

Source: arXiv (Cornell University)

DOI: 10.48550/arxiv.2310.19925

Authors: Khan, Shihab Shahriar, Palmer, Bryce, Edelmaierd, Christopher, Aktulga, Hasan Metin

Institution: Michigan State University

Language: English

Topic: Parallel Computing and Optimization Techniques

Subfield: Hardware and Architecture

Field: Computer Science

Domain: Physical Sciences

Open Access status: green

Funders: National Science Foundation, Michigan State University, College of Engineering, Michigan State University, Office of Advanced Cyberinfrastructure

Awards: G2580342688, OAC Core: Small: Efficient and scalable tools for design and analysis of active matter systems

OpenAlex ID: W4388183295

Cites:

Cited by:

Related to: 10