Ravi Reddy Manumachu's Research

I am an assistant professor in the School of Computer Science at University College Dublin (UCD) in Ireland. I earned my B.Tech degree from the Indian Institute of Technology (I.I.T), Madras, in 1997 and completed my PhD in Computer Science, focusing on high-performance heterogeneous computing, at UCD in 2005. My primary research interests include heterogeneous parallel computing and energy-efficient computing.

My research focuses on developing innovative models, algorithms, and tools aimed at optimizing performance and energy usage for applications running on modern extreme-scale and highly heterogeneous computing platforms, such as cloud environments, grid systems, and supercomputers.

I am a co-author of various works, including functional performance and energy models of computation, a theoretical framework for energy predictive models of computing, and analyses of energy proportionality in modern server processors. Additionally, I have developed accurate and reliable linear energy predictive models, as well as model-based methods and data partitioning algorithms designed for bi-objective optimization of performance and energy on heterogeneous computing platforms.

Furthermore, I contributed to the creation of Heterogeneous MPI, an extension of the Message-Passing Interface (MPI) for heterogeneous clusters, and Heterogeneous ScaLAPACK, a linear algebra package tailored for use in heterogeneous clusters. My work has been published extensively in top journals and conferences related to parallel and distributed computing, as well as energy science.

I am the director of the M.Sc. program in Computer Science through Negotiated Learning and the postdoctoral coordinator in the School of Computer Science (ucdmscnl.com). Additionally, I serve as a member of my School’s Equality, Diversity, and Inclusion (EDI) committee, where I focus on developing and implementing action plans for the ATHENA SWAN initiative.

I have 11+ years of industrial experience in top software development companies (IONA Technologies, Ansys, Siemens Research), working in diverse software fields such as distributed middleware, high-performance web services, parallel computational fluid dynamics, and IBM mainframe processing.

Academic Profiles
UCD
Google Scholar
ResearchGate
DBLP


Ph.D Thesis
HMPI: A Message-Passing Library for Heterogeneous Networks of Computers
School of Computer Science, University College Dublin, Dublin, pp. 456, 06/2005
pdf

Softwares

OpenH: A Tool for Programming Portable Parallel Applications on Heterogeneous Hybrid Servers, 2024
OpenH

libedm: Energy modelling of data movement and communications in modern heterogeneous hybrid platforms, 2023.
libedm

EPPACK: Energy proportionality toolkit for servers and data centers, 2022
EPPACK

PAREPOPT: Parallel ALgorithms for Energy and Performance Optimization on modern HPC platforms, 2021
PAREPOPT

hcllimb: Model-based Performance Optimization of Scientific Applications using Parallel Computing Load Balancing and Imbalancing Methods for Multicore Processors, 2018
HCLLIMB

libhclooc: Software Library Facilitating Out-of-core Implementations of Accelerator Kernels on Hybrid Computing Platforms, 2018
LIBHCLOOC

PARALEPH: Parallel ALgorithms for Energy and Performance using functional performance models for Homogeneous Multicore Clusters, 2017
PARALEPH

HCLWattsUp: Tool for Reliable and Accurate System-Level Energy Measurements Using Power Meters, 2016
HCLWattsUp

HeteroScaLAPACK: A Linear Algebra Library for Heterogeneous Networks of Computers, 2010
HeteroScaLAPACK

HeteroMPI: An extension of MPI for high performance heterogeneous computing, 2004
HeteroMPI