Scott McMillan
Software Engineering Institute
Publications by Scott McMillan
-
PHITE: Portable High-Performance Inference at the Tactical Edge
November 11, 2022 • Presentation
Scott McMillan
This project applies performance engineering processes to the analysis of existing open source ML frameworks for embedded systems, to inform the development and optimization of a portable software library.
read -
Preview of PHITE: Portable High-Performance Inference at the Tactical Edge
November 07, 2022 • Video
Scott McMillan
This short video provides an introduction to a research topic presented at the SEI Research Review 2022.
watch -
Spiral/AIML: Resource-Constrained Co-Optimization for High-Performance, Data-Intensive Computing
November 07, 2021 • Presentation
Scott McMillan
Spiral AI/ML is an SEI project to automate code generation for data-intensive computations while simultaneously optimizing for targeted hardware.
read -
Spiral/AIML: Resource-Constrained Co-Optimization for High-Performance, Data-Intensive Computing
November 04, 2021 • Video
Scott McMillan
This short video provides an introduction to a research topic presented at the SEI Research Review 2021.
watch -
Poster - Co-Optimization for High-Performance Data-Intensive Computing in Resource-Constrained Environments
November 03, 2020 • Poster
Scott McMillan
Spiral AI/ML helps developers build high-performance applications on leading-edge hardware architectures faster and cheaper, speeding new capabilities to serve national and tactical needs.
read -
Spiral/AIML: Resource-Constrained Co-Optimization for High-Performance, Data-Intensive Computing
November 11, 2019 • Video
Scott McMillanFranz Franchetti (Carnegie Mellon University)
Watch SEI Principal Investigator, Dr. Scott McMillan, and research collaborator, CMU ECE Professor Franz Franchetti, discuss a community research effort to develop tools to reduce the prohibitive cost of implementing and re-implementing AI/ML software on
watch -
Spiral/AIML: Frontiers of Graph Processing in Linear Algebra
October 28, 2019 • Poster
Scott McMillanFranz Franchetti (Carnegie Mellon University)
This poster describes research to use a linear algebraic approach to graph algorithms
read -
Spiral/AIML: Co-optimization for High-Performance, Data-Intensive Computing in Resource Constrained Environments
October 28, 2019 • Presentation
Scott McMillanFranz Franchetti (Carnegie Mellon University)
Data-intensive computing is pervasive. This presentation provides an update on research to allow platform developers to realize high-performance AI/ML applications on leading-edge hardware architectures faster and cheaper.
read -
Spiral/AIML: Frontiers of Graph Processing in Linear Algebra
October 28, 2019 • Poster
Scott McMillanFranz Franchetti (Carnegie Mellon University)
This poster represents research to extend the use of linear algebra beyond simple graph traversal.
read -
Building a COTS Benchmark Baseline for Graph Analytics
October 23, 2018 • Poster
Scott McMillan
This poster describes research aimed at building a benchmark baseline based on commercial off-the-shelf (COTS) field-programmable gate array (FPGA) hardware.
read -
Automatic Code Generation for Graph Algorithms
October 23, 2018 • Poster
Scott McMillanFranz Franchetti (Carnegie Mellon University)
This poster describes automated code generation of high-performance libraries of graph algorithms, tuned for different hardware architectures.
read -
Measuring Performance of Big Learning Workloads
October 30, 2017 • Presentation
Scott McMillan
Presentation on research to build a performance measurement workbench with tools to measure and report performance of large-scale ML platforms
read -
Automated Code Generation for High-Performance Graph Libraries
October 30, 2017 • Presentation
Scott McMillan
Presentation on research into graph analytics
read -
Measuring Performance of Big Learning Workloads
October 30, 2017 • Poster
Scott McMillan
Poster on research to build a performance measurement workbench with tools to measure and report performance of large-scale ML platforms
read -
Automated Code Generation for High-Performance, Future-Compatible Graph Libraries
October 30, 2017 • Poster
Scott McMillan
Poster on research into graph analytics
read -
Automated Code Generation for High-Performance, Future-Compatible Graph Libraries (2017)
October 27, 2017 • Poster
Scott McMillan
Poster for research project on graph analytics
read -
GraphBLAS: A Programming Specification for Graph Analysis
November 01, 2016 • Presentation
Scott McMillan
Describes work in graph analysis, an important and pervasive areas for the DoD
read -
GraphBLAS
October 18, 2016 • Poster
Scott McMillan
A Programming Specification for Graph Analysis
read -
Mathematical Foundations of the GraphBLAS
September 15, 2016 • Conference Paper
Jeremy Kepner (MIT Lincoln Laboratory)Peter Aaltonen (Indiana University)David Bader (Georgia Institute of Technology)
This paper introduces the mathematics of the GraphBLAS, which is being developed to bring matrix-based graph algorithms to the broadest possible audience.
read -
Design and Implementation of the GraphBLAS Template Library (GBTL)
July 11, 2016 • Presentation
Scott McMillanSamantha MisurdaMarcin Zalewski (Indiana University)
The design of the GraphBLAS Template Library separates graph algorithm development from performance tuning for heterogeneous high-performance computing architectures.
read -
GBTL-CUDA: Graph Algorithms and Primitives for GPUs
May 23, 2016 • Conference Paper
Peter Zhang (Indiana University)Marcin Zalewski (Indiana University)Andrew Lumsdaine (Indiana University)
In this paper we present our initial implementation of GraphBLAS primitives for graphics processing unit (GPU) systems called GraphBLAS Template Library (GBTL).
read -
Dynamic Parallelism for Simple and Efficient GPU Graph Algorithms
December 09, 2015 • Conference Paper
Peter Zhang (Indiana University)Eric Holk (Indiana University)John Matty
Presented at the 2015 Supercomputing Conference, this paper shows that dynamic parallelism enables relatively high-performance graph algorithms for GPUs.
read -
Graph Algorithms on Future Architectures Poster (SEI 2015 Research Review)
October 22, 2015 • Poster
Scott McMillan
Delves into whether primitives and operations can be defined to separate graph analytic application development and complexity of underlying hardware concern
read -
Graph Algorithms on Future Architectures
October 16, 2015 • Presentation
Scott McMillan
Delves into whether primitives and operations can be defined to separate graph analytic application development and complexity of underlying hardware concern
read -
Toward Speed and Simplicity: Creating a Software Library for Graph Analytics
August 27, 2015 • Podcast
Scott McMillanEric Werner
In this podcast, Scott McMillan and Eric Werner of the SEI's Emerging Technology Center discuss work to create a software library for graph analytics that would take advantage of more powerful heterogeneous supercomputers.
learn more -
Patterns and Practices for Future Architectures
August 15, 2014 • Technical Note
Eric WernerScott McMillanJonathan Chu
This report discusses best practices and patterns that will make high-performance graph analytics on new and emerging architectures more accessible to users.
read