Synopses & Reviews
High Performance Compilers for Parallel Computing
provides a clear understanding of the analysis and optimization methods used in modern commercial research compilers for parallel systems. By the author of the classic 1989 monograph Optimizing Supercompilers for Supercomputers
, this book covers the knowledge and skills necessary to build a competitive, advanced compiler for parallel or high-performance computers. Starting with a review of basic terms and algorithms used - such as graphs, trees, and matrix algebra - Wolfe
shares the lessons of his 20 years experience developing compiler products. He provides a complete catalog of program restructuring methods that have proven useful in the discovery of parallelism or performance optimization and discusses compiling details for each type of parallel system described, from simple code generation, through basic and aggressive optimizations. A wide variety of parallel systems are presented, from bus-based cache-coherent shared memory multiprocessors and vector computers, to message-passing multicomputers and large-scale shared memory systems.
About the Author
As co-founder in 1979 of Kuck and Associates, Inc., Michael Wolfe
helped develop KAP restructuring, parallelizing compiler software. In 1988, Wolfe joined the Oregon Graduate Institute of Science and Technology
faculty, directing research on language and compiler issues for high performance computer systems. His current research includes development and implementation of program restructuring transformations to optimize programs for execution on parallel computers, refinement and application of recent results in analysis techniques to low level compiler optimizations, and analysis of data dependence decision algorithms.
Table of Contents
1. High Performance Systems.
An Example Program: Matrix Multiplication.
Structure of a Compiler.
2. Programming Language Features.
Languages for High Performance.
Sequential and Parallel Loops.
3. Basic Graph Concepts.
Sets, Tuples, Logic.
4. Review of Linear Algebra.
Real Vectors and Matrices.
Integer Matrices and Lattices.
Linear System of Equations.
System of Integer Equations.
Systems of Linear Inequalities.
Systems of Integer Linear Inequalities.
Extreme Values of Affine Functions.
5. Data Dependence.
Data Dependence in Loops.
Data Dependence in Conditionals.
Data Dependence in Parallel Loops.
Program Dependence Graph.
6. Scalar Analysis with Factored Use-Def Chains.
Constructing Factored Use-Def Chains.
FUD Chains for Arrays.
Finding All Reaching Definitions.
Implicit References in FUD Chains.
InductionVariables Using FUD Chains.
Constant Propagation with FUD Chains.
Data Dependence for Scalars.
7. Data Dependence Analysis for Arrays.
Building the Dependence System.
Dependence System Solvers.
Summary of Solvers.
Run-time Dependence Testing.
8. Other Dependence Problems.
Array Region Analysis.
9. Loop Restructuring.
Linear Loop Transformations.
Other Loop Transformations.
10. Optimizing for Locality.
Single Reference to Each Array.
Fission and Fusion for Locality.
11. Concurrency Analysis.
Code for Concurrent Loops.
Concurrency from Sequential Loops.
Concurrency from Parallel Loops.
Exceptions and Debuggers.
12. Vector Analysis.
Vector Code from Sequential Loops.
Vector Code from Forall Loops.
Roundoff Error, Exceptions, and Debuggers.
13. Message-Passing Machines.
Parallel Code for Array Assignment.
Remote Data Access.
Automatic Data Layout.
Multiple Array Assignments.
14. Scalable Shared-Memory Machines.
Global Cache Coherence.
Local Cache Coherence.
Latency Tolerant Machines.