Next:
1 Overview
Up:
ADAPTOR HPF Programmers Guide
Previous:
ADAPTOR HPF Programmers Guide
Index
Contents
1 Overview
2 Execution Model of HPF Programs
2.1 Serial Execution of HPF Programs
2.2 The Multiprocessing Execution Model for HPF
2.3 The Multithreading Execution Model for HPF
2.4 The Hierarchical Execution Model
3 Home of Computations and Work Distribution
3.1 Importance of Work Distribution
3.2 Default Work Distribution in ADAPTOR
3.3 The ON Directive
3.4 Active Processors
3.5 Restrictions for the ON Directive
3.6 Execution of Subroutines
3.7 Execution of I/O Statements
3.8 Serial Procedures
4 Local Computations
4.1 Local Array Assignments
4.2 Local FORALL Statements
4.3 Local Independent Loops
4.4 Independent Loops and NEW Directive
4.5 Independent Loops and RESIDENT Directive
4.6 Importance of Alignment for Local Computations
4.7 PURE Procedures
4.8 Coupling of the ON and RESIDENT Directives
4.9 Parallelism with the ON Directive
4.10 Local Procedures
4.11 Private Variables
5 Communication and Synchronization
5.1 Using Shared Arrays
5.2 Temporary Arrays
5.3 Problems with Extracting Communication
6 Global Communications
6.1 Broadcast
6.2 Spreading
6.3 Reduction Functions
6.4 Reduction Operations in Independent Loops
7 Structured Communication
7.1 Assignments with Regular Sections
7.2 FORALL Statements with Structured Communication
7.3 Shifting
7.4 Transpose
7.5 Matrix Multiplication
8 Unstructured Communication
8.1 Gathering of Data
8.2 Scattering of Data
8.3 The TRACE Directive
9 Shadow Edges and Halos
9.1 Shadow Edges
9.2 Automatic Creation of Shadow Edges
9.3 Halos
10 Explicit and Implicit Remapping
10.1 About the Importance of Remapping
10.2 Explicit Remapping via Array Assignments
10.3 Explicit Remapping via Remapping Directives
10.4 Implicit Remapping at Subroutine Boundaries
10.5 Data Transfers with the ON Directive
11 Pitfalls
11.1 Identifying Overhead of HPF Programs
11.2 Loop Ordering
11.3 Avoiding Temporary Arrays in Array Operations
11.4 Serial Loops
11.5 When Independent Loops Become Serial
11.6 Efficient Communication
11.7 Avoiding Remappings and Reallocations
12 Compiling Subprograms for Processor Subsets
Bibliography
Index
About this document ...
Thomas Brandes 2004-03-18
