Loop and task parallelization on heterogeneous cloud clusters using map-reduce for scientific workloads

• Grant number: 17/21339-7
• Support Opportunities: Scholarships abroad - Research Internship - Post-doctor
• Start date: December 01, 2017
• End date: November 30, 2018
• Field of knowledge: Physical Sciences and Mathematics - Computer Science - Computing Methodologies and Techniques
• Principal Investigator: Guido Costa Souza de Araújo
• Grantee: Hervé Yviquel
• Supervisor: Xavier Martorell
• Host Institution: Instituto de Computação (IC). Universidade Estadual de Campinas (UNICAMP). Campinas , SP, Brazil
• Institution abroad: Barcelona Supercomputing Center (BSC), Spain
• Associated to the scholarship: 14/25694-8 - Loop and task parallelization using cloud clusters map-reduce for scientific workloads, BP.PD

Abstract

In this project, we aim to tackle the complexities of programming modern cloud clusters containing heterogeneous components such as CPUs and GPUs. Indeed, modern scientific applications need to exploit all type of parallel programming pattern to reach the performance goals. Computation offloading is a programming model in which program fragments (e.g. hot loops) are annotated so that their execution is performed in dedicated hardware or accelerator devices available. Although offloading has been extensively used to move computation to GPUs, through directive-based annotation standards like OpenMP, offloading computation to very large computer clusters in the cloud can become a complex and cumbersome task. It typically requires mixing programming models (e.g. OpenMP and MPI) and languages (e.g. C/C++ and Scala), dealing with various access control mechanisms from different clouds (e.g. AWS and Azure), and integrating all this into a single application. To this end, we developed a framework to allow the use of the cloud as a computation offloading device. It integrates OpenMP directives, cloud based map-reduce Spark nodes and remote communication management such that the cloud appears to the programmer as yet another device available in its local computer. This project aims to extend the execution model of cloud offloading as well as OpenMP directives in order to increasing the processing power of the cloud device by performing part of the computation on the GPUs available in the cluster. (AU)

Scientific publications

(References retrieved automatically from Web of Science and SciELO through information on FAPESP grants and their corresponding numbers as mentioned in the publications by the authors)

YVIQUEL, HERVE; CRUZ, LAURO; ARAUJO, GUIDO. Cluster Programming using the OpenMP Accelerator Model. ACM TRANSACTIONS ON ARCHITECTURE AND CODE OPTIMIZATION, v. 15, n. 3, OCT 2018. (14/25694-8, 17/21339-7)

MORTATTI, MATHEUS; YVIQUEL, HERVE; ARAUJO, GUIDO; IEEE. Automatic Ray-Tracer Cloud Offloading in OpenMP. 2018 30TH INTERNATIONAL SYMPOSIUM ON COMPUTER ARCHITECTURE AND HIGH PERFORMANCE COMPUTING (SBAC-PAD 2018), v. N/A, p. 8-pg., 2018-01-01. (17/21339-7, 14/25694-8)