Optimization and mathematical modeling in computer architecture /

Bibliographic Details
Main Authors:	Nowatzki, Tony (Author), Estan, Cristian (Author), Ferris, Michael C. (Author), Sankaralingam, Karthikeyan (Author), Vaish, Nilay (Author), Wood, David Allen (Author)
Format:	eBook
Language:	English
Published:	San Rafael, California (1537 Fourth Street, San Rafael, CA 94901 USA) : Morgan & Claypool, 2014.
Series:	Synthesis lectures in computer architecture ; # 26.
Subjects:	Computer architecture > Mathematical models. Integer programming. Linear programming. Mathematical optimization. GAMS General Algebraic Modeling System ILP Instruction Set Customization Integer Linear Programming MILP Mathematical Modeling Mixed Integer Linear Programming Optimization Resource Allocation Scheduling Spatial Architectures Tiled Architectures Electronic books.
Online Access:	Connect to the full text of this electronic book

Table of Contents:

1. Introduction
1.1 Why this book?
1.1.1 Evolution of mathematical theories and algorithms
1.1.2 Maturity of solvers and modeling systems
1.1.3 Complexity of computer systems
1.2 Who is this book for?
1.3 What is this book about?
1.3.1 Mathematical modeling
1.3.2 Optimization as a modeling technique
1.3.3 The essential primitives of MILP
1.3.4 Illustrative examples
1.3.5 Benefits of modeling and MILP
1.4 What this book is not about
1.5 Book overview
1.6 Code provided with this book
2. An overview of optimization
2.1 Overview of optimization
2.2 Models for optimization
2.2.1 Linear programming
2.2.2 Convex programming
2.2.3 Network flow problems
2.2.4 Mixed integer linear programming
2.2.5 Mixed integer nonlinear programs
2.3 Modeling problems as MILP
2.3.1 Logic and binary variables
2.3.2 Constraint logic programming
2.3.3 Ordering
2.3.4 Piecewise-linear models
2.3.5 Modeling mixed integer nonlinear programs
2.4 Solution methods
2.4.1 Branch-and-bound
2.4.2 Extensions to basic branch-and-bound
2.4.3 Column generation
2.4.4 Bender's decomposition
2.4.5 Other approaches
2.4.6 Modeling languages
2.5 Conclusion
3. Case study: instruction set customization
3.1 Introduction
3.2 Overview
3.3 Formulation: parameters and decision variables
3.4 Formulation: constraints
3.5 Formulation: objective
3.6 Modeling limitations
3.7 Evaluation
3.7.1 Methodology
3.7.2 Results
3.8 Related work
3.9 Conclusions
4. Case study: data center resource management
4.1 Introduction
4.2 Overview
4.3 Formulation: parameters and decision variables
4.4 Formulation: constraints
4.5 Formulation: objective
4.6 Modeling limitations
4.7 Evaluation
4.7.1 Methodology
4.7.2 Results
4.8 Related work
4.9 Conclusions
5. Case study: spatial architecture scheduling
5.1 Introduction
5.2 Overview
5.3 Formulation: parameters and decision variables
5.4 Formulation: constraints
5.5 Formulation: objective
5.6 Architecture-specific modeling
5.6.1 Architecture-specific details for TRIPS
5.6.2 Architecture-specific details for DySER
5.6.3 Architecture-specific details for PLUG
5.7 Modeling limitations
5.8 Evaluation
5.8.1 Methodology
5.8.2 Results
5.9 Related work
5.10 Discussion and conclusions
6. Case study: resource allocation in tiled architectures
6.1 Introduction
6.2 Overview
6.3 Formulation: parameters and decision variables
6.4 Formulation: constraints
6.5 Formulation: objective
6.6 Modeling limitations
6.7 Evaluation
6.7.1 Methodology
6.7.2 Results
6.8 Related work
6.9 Conclusions
7. Conclusions
7.1 Properties of a MILP-friendly problem
7.2 Understanding the limitations of MILP
7.2.1 Properties of optimization problems unsuitable to MILP
7.2.2 Example problems poorly suited to MILP
7.3 Implementing your optimization problems in MILP
7.3.1 First steps
7.3.2 Dealing with MILP challenges
7.3.3 Optimizing and tuning models
7.4 Lessons learned
Bibliography
Authors' biographies.