Compaq Fortran
User Manual for
Tru64 UNIX and
Linux Alpha Systems

Contents

Preface
Preface	Preface
Chapter 1
1	Getting Started
1.1	Compaq Fortran Programming Environment
1.2	Commands to Create and Run an Executable Program
1.3	Creating and Running a Program Using a Module and Separate Function
1.3.1	Commands to Create the Executable Program
1.3.2	Running the Sample Program
1.3.3	Debugging the Sample Program
1.4	f90 or fort Command and Related Software Components
1.4.1	f90 or fort Driver Program
1.4.2	cpp, fpp, and Other Preprocessors
1.4.3	Compaq Fortran Compiler
1.4.4	Other Compilers
1.4.5	Linker (ld)
1.5	Program Development Stages and Tools
1.6	Compaq Fortran and Standards It Conforms To
Chapter 2
2	Compiling and Linking Compaq Fortran Programs
2.1	f90 Command: Files and Options
2.1.1	File Suffixes and Source Forms
2.1.2	Format of the f90 and fort Commands
2.1.3	Creating and Using Module Files
2.1.3.1	Creating Module Files
2.1.3.2	Using Module Files
2.1.4	INCLUDE Statement and Using Include Files
2.1.5	Output Files: Executable, Object, and Temporary
2.1.5.1	Naming Output Files
2.1.5.2	Temporary Files
2.1.6	Using Multiple Input Files: Effect on Output Files
2.1.7	Examples of the f90 and fort Commands
2.1.7.1	Compiling and Linking Multiple Files
2.1.7.2	Retaining an Object File and Preventing Linking
2.1.7.3	Compiling Fortran 95/90 and C Source Files and Linking an Object File
2.1.7.4	Renaming the Output File
2.1.7.5	Specifying an Additional Linker Library
2.1.7.6	Requesting Additional Optimizations
2.1.8	Using Listing Files
2.2	Driver Programs and Passing Options to cc and ld
2.2.1	make Facility
2.2.2	Options Passed to the cc Driver or ld Linker
2.3	Compiler Limits, Diagnostic Messages, and Error Conditions
2.3.1	Compiler Limits
2.3.2	Compiler Diagnostic Messages and Error Conditions
2.3.3	Linker Diagnostic Messages and Error Conditions
2.4	Compilation Control: Statements and Directives
2.5	Linking Object Libraries
2.5.1	Specifying Additional Object Libraries
2.5.2	Specifying Types of Object Libraries
2.5.3	Specifying Shared Object Libraries
2.6	Creating Shared Libraries
2.6.1	Creating a Shared Library with a Single f90 Command
2.6.2	Creating a Shared Library with f90 and ld Commands
2.6.3	Choosing How to Create a Shared Library
2.6.4	Shared Library Restrictions
2.6.5	Installing Shared Libraries (TU*X ONLY)
Chapter 3
3	f90 and fort Command-Line Options
3.1	Overview of Command-Line Options
3.2	f90 and fort Command Categories and Options
3.3	-align keyword --- Data Alignment
3.4	-annotations keyword --- Place Optimization Information in Source Listing
3.5	-arch keyword --- Specify Type of Code Instructions Generated
3.6	-assume buffered_io --- Buffered Output
3.7	-assume byterecl --- Units for Unformatted File Record Length
3.8	-assume cc_omp --- Enable Conditional Compilation for OpenMP
3.9	-assume dummy_aliases --- Dummy Variable Aliasing
3.10	-assume gfullpath --- Source File Path for Debugging
3.11	-assume minus0 --- Standard Semantics for Minus Zero
3.12	-assume noaccuracy_sensitive, -fp_reorder --- Reorder Floating-Point Calculations
3.13	-assume noprotect_constants --- Remove Protection from Constants
3.14	-assume nosource_include --- INCLUDE file search
3.15	-assume nounderscore --- Underscore on External Names
3.16	-assume no2underscores --- Two Underscores on External Names
3.17	-assume pthreads_lock --- Thread Lock Selection for Parallel Execution
3.18	-automatic, -static --- Local Variable Allocation
3.19	-c --- Inhibit Linking and Retain Object File
3.20	-call_shared, -non_shared, -shared --- Shared Library Use
3.21	-ccdefault keyword --- Carriage Control for Terminals
3.22	-check arg_temp_created --- Check for Copy of Temporary Arguments
3.23	-check bounds, -C, -check_bounds --- Boundary Run-Time Checking
3.24	-check format --- Format Mismatches at Run Time
3.25	-check nopower --- Allow Special Floating-Point Expressions
3.26	-check omp_bindings --- OpenMP Fortran API Binding Rules Checking
3.27	-check output_conversion --- Truncated Format Mismatches at Run Time
3.28	-check overflow --- Integer Overflow Run-Time Checking
3.29	-check underflow --- Floating-Point Underflow Run-Time Checking
3.30	-convert keyword --- Unformatted Numeric Data Conversion
3.31	-cpp and Related Options --- Run C Preprocessor
3.31.1	-M --- Request cpp Dependency Lists for make
3.31.2	-P --- Retain cpp Intermediate Files
3.31.3	-Wp,-xxx --- Pass Specified Option to cpp
3.32	-Dname, -Dname=def, -Dname="string" --- Define Symbol Names
3.33	-d_lines --- Debugging Statement Indicator, Column 1
3.34	-double_size 128, -double_size 64 --- Double Precision Data Size
3.35	-error_limit num, -noerror_limit --- Limit Error Messages
3.36	-extend_source --- Line Length for Fixed-Format Source
3.37	-f66, -66, -nof77, -onetrip, -1 --- Use FORTRAN 66 Semantics
3.38	-f77, -nof66 --- Use FORTRAN 77 Semantics
3.39	-f77rtl --- Use Fortran 77 Run-Time Behavior
3.40	-fast --- Set Options to Improve Run-Time Performance
3.41	-feedback file, -gen_feedback, -cord --- Create and Use Feedback Files
3.42	-fixed, -free --- Fortran Source Format
3.43	-fpconstant --- Handling of Floating-Point Constants
3.44	-fpen --- Control Arithmetic Exception Handling and Reporting
3.44.1	Hints on Using These Options
3.45	-fpp --- Run Fortran Preprocessor
3.46	-fprm keyword --- Control Floating-Point Rounding Mode
3.47	-fuse_xref --- Cross-Reference Information for Compaq FUSE
3.48	-g0, -g1, -g2 or -g, -g3, -ladebug --- Traceback and Symbol Table Information
3.49	-granularity keyword --- Control Shared Memory Access to Data
3.50	-hpf, -hpf num, and Related Options --- Compile HPF Programs for Parallel Execution
3.50.1	-assume bigarrays --- Run-time Checking for Distributed Small Array Dimensions
3.50.2	-assume nozsize --- Omit Zero-Sized Array Checking
3.50.3	-nearest_neighbor, -nearest_neighbor num, or -nonearest_neighbor --- Nearest Neighbor Optimization
3.50.4	-show hpf --- Show HPF Parallelization Information
3.50.5	-hpf_target --- Message Passing Protocol for Parallel Programs
3.51	-I --- Remove Directory from Include Search Path
3.52	-Idir --- Add Directory for Module and Include File Search
3.53	-i2, -i4, -i8, -integer_size num --- Integer and Logical Data Size
3.54	-inline keyword, -noinline --- Control Procedure Inlining
3.55	-intconstant --- Handling of Integer Constants
3.56	-K --- Keep Temporary Files
3.57	-L --- Remove ld Directory Search Path
3.58	-Ldir --- Add Directory to ld Search Path
3.59	-lstring --- Add Library Name to ld Search
3.60	-machine_code
3.61	-math_library keyword --- Fast or Accurate Math Library Routines
3.62	-mixed_str_len_arg --- Specify Length of Character Arguments
3.63	-module directory --- Specify Directory for Creating Modules Files
3.64	-mp --- Enable Parallel Processing Using Directed Decomposition
3.65	-names keyword --- Case Control of Source and External Names
3.66	-noaltparam --- Alternative PARAMETER Syntax
3.67	-nofor_main --- Allow Non-Fortran Main Program
3.68	-nohpf_main, -nowsf_main --- Compile HPF Global Routine for Nonparallel Main Program
3.69	-noinclude --- Omit Standard Directory Search for INCLUDE Files
3.70	-norun --- Do Not Run the Compiler
3.71	-o output --- Name Output File
3.72	-O0, -O1, -O2, -O3, -O4 or -O, -O5 --- Specify Optimization Level
3.73	-om --- Request Nonshared Object Optimizations
3.74	-omp --- Enable OpenMP Parallel Processing Using Directed Decomposition
3.75	-pad_source --- Pad Short Source Records with Spaces
3.76	-pipeline --- Activate Software Pipelining Optimization
3.77	-p0, -p1 or -p, and -pg --- Profiling Support
3.78	-real_size number, -r8, -r16 --- Floating-Point Data Size
3.79	-recursive --- Request Recursive Execution
3.80	-reentrancy keyword --- Control Use of Threaded Run-Time Library
3.81	-S --- Create Assembler File
3.82	-show keyword, -machine_code --- Control Listing File Content
3.83	-source_listing --- Create a Source Listing File
3.84	-speculate keyword --- Speculative Execution Optimization
3.85	-std, -std90, -std95 --- Perform Fortran Standards Checking
3.86	-synchronous_exceptions --- Report Exceptions More Precisely
3.87	-syntax_only --- Do Not Create Object File
3.88	-threads, -pthread --- Link Using Threaded Run-Time Library
3.89	-transform_loops --- Activate Loop Transformation Optimizations
3.90	-tune keyword --- Specify Alpha Processor Implementation
3.91	-U --- Activates Case Sensitivity
3.92	-Uname --- Undefine Preprocessor Symbol Name
3.93	-u
3.94	-unroll num --- Specify Number for Loop Unroll Optimization
3.95	-V --- Create Listing File
3.96	-v --- Verbose Command Processing Display
3.97	-version, -what --- Show Compaq Fortran Version Information
3.98	-vms --- OpenVMS Fortran Compatibility
3.99	-Wl,-xxx --- Pass Specified Option to ld
3.100	-warn keyword, -u, -nowarn, -w, -w1 --- Warning Messages and Compiler Checking
3.101	-warning_severity keyword --- Elevate Severity of Warning Messages
3.102	-what
3.103	-wsf
Chapter 4
4	Using the Ladebug Debugger
4.1	Overview of Ladebug and dbx Debuggers
4.2	Compaq Fortran Options for Debugging
4.3	Running the Debugger
4.3.1	Creating the Executable Program and Running the Debugger
4.3.1.1	Invoking Ladebug
4.3.1.2	Invoking dbx
4.3.2	Debugger Commands and Breakpoints
4.3.3	Ladebug Limitations
4.4	Sample Program and Debugging Session
4.5	Summary of Debugger Commands
4.6	Displaying Variables
4.6.1	Compaq Fortran Module Variables
4.6.2	Compaq Fortran Common Block Variables
4.6.3	Compaq Fortran Derived-Type Variables
4.6.4	Compaq Fortran Record Variables
4.6.5	Compaq Fortran Pointer Variables
4.6.5.1	Fortran 95/90 Pointers
4.6.5.2	CRAY-Style Pointers
4.6.6	Compaq Fortran Array Variables
4.6.6.1	Array Sections
4.6.6.2	Assignment to Arrays
4.6.7	Complex Variables
4.6.8	Compaq Fortran Data Types
4.7	Expressions in Debugger Commands
4.7.1	Fortran Operators
4.7.2	Procedures
4.8	Debugging Mixed-Language Programs with Ladebug
4.9	Debugging a Program that Generates an Exception
4.10	Locating Unaligned Data
4.10.1	Locating Unaligned Data With Ladebug
4.10.2	Locating Unaligned Data With dbx
4.11	Using Alternate Entry Points
4.12	Debugging Optimized Programs
Chapter 5
5	Performance: Making Programs Run Faster
5.1	Efficient Compilation and the Software Environment
5.1.1	Install the Latest Version of Compaq Fortran and Performance Products
5.1.2	Compile Using Multiple Source Files and Appropriate f90 Options
5.1.3	Process Shell Environment and Related Influences on Performance
5.2	Using the time Command to Measure Performance
5.3	Using Profiling Tools
5.3.1	Program Counter Sampling (prof)
5.3.2	Call Graph Sampling (gprof)
5.3.3	Basic Block Counting (pixie and prof)
5.3.4	Source Line CPU Cycle Use (prof and pixie)
5.3.5	Creating and Using Feedback Files and Optionally cord
5.3.6	Atom Toolkit
5.4	Data Alignment Considerations
5.4.1	Causes of Unaligned Data and Ensuring Natural Alignment
5.4.2	Checking for Inefficient Unaligned Data
5.4.3	Ordering Data Declarations to Avoid Unaligned Data
5.4.3.1	Arranging Data Items in Common Blocks
5.4.3.2	Arranging Data Items in Derived-Type Data
5.4.3.3	Arranging Data Items in Compaq Fortran Record Structures
5.4.4	Options Controlling Alignment
5.5	Using Arrays Efficiently
5.5.1	Accessing Arrays Efficiently
5.5.2	Passing Array Arguments Efficiently
5.6	Improving Overall I/O Performance
5.6.1	Use Unformatted Files Instead of Formatted Files
5.6.2	Write Whole Arrays or Strings
5.6.3	Write Array Data in the Natural Storage Order
5.6.4	Use Memory for Intermediate Results
5.6.5	Enable Implied-DO Loop Collapsing
5.6.6	Use of Variable Format Expressions
5.6.7	Efficient Use of Record Buffers and Disk I/O
5.6.8	Specify RECL
5.6.9	Use the Optimal Record Type
5.6.10	Reading from a Redirected Standard Input File
5.7	Additional Source Code Guidelines for Run-Time Efficiency
5.7.1	Avoid Small Integer and Small Logical Data Items
5.7.2	Avoid Mixed Data Type Arithmetic Expressions
5.7.3	Use Efficient Data Types
5.7.4	Avoid Using Slow Arithmetic Operators
5.7.5	Avoid Using EQUIVALENCE Statements
5.7.6	Use Statement Functions and Internal Subprograms
5.7.7	Code DO Loops for Efficiency
5.8	Optimization Levels: the -On Option
5.8.1	Optimizations Performed at All Optimization Levels
5.8.2	Local (Minimal) Optimizations
5.8.2.1	Common Subexpression Elimination
5.8.2.2	Integer Multiplication and Division Expansion
5.8.2.3	Compile-Time Operations
5.8.2.4	Value Propagation
5.8.2.5	Dead Store Elimination
5.8.2.6	Register Usage
5.8.2.7	Mixed Real/Complex Operations
5.8.3	Global Optimizations
5.8.4	Additional Global Optimizations
5.8.4.1	Loop Unrolling
5.8.4.2	Code Replication to Eliminate Branches
5.8.5	Automatic Inlining
5.8.5.1	Interprocedure Analysis
5.8.5.2	Inlining Procedures
5.8.6	Software Pipelining
5.8.7	Loop Transformation
5.9	Other Options Related to Optimization
5.9.1	Setting Multiple Options with the -fast Option
5.9.2	Controlling the Number of Times a Loop Is Unrolled
5.9.3	Controlling the Inlining of Procedures
5.9.4	Requesting Optimized Code for a Specific Processor Generation
5.9.5	Requesting the Speculative Execution Optimization
5.9.6	Request Nonshared Object Optimizations
5.9.7	Arithmetic Reordering Optimizations
5.9.8	Dummy Aliasing Assumption
Chapter 6
6	Parallel Compiler Directives and Their Programming Environment
6.1	OpenMP Fortran API Compiler Directives
6.1.1	Command-Line Option and Directives Format
6.1.1.1	Directive Prefixes
6.1.1.2	Directive Prefixes for Conditional Compilation
6.1.2	Summary Descriptions of OpenMP Fortran API Compiler Directives
6.1.3	Parallel Processing Thread Model
6.1.4	Privatizing Named Common Blocks: THREADPRIVATE Directive
6.1.5	Controlling Data Scope Attributes
6.1.6	Parallel Region: PARALLEL and END PARALLEL Directives
6.1.7	Worksharing Constructs
6.1.7.1	DO and END DO directives
6.1.7.2	SECTIONS, SECTION, and END SECTIONS Directives
6.1.7.3	SINGLE and END SINGLE Directives
6.1.8	Combined Parallel/Worksharing Constructs
6.1.8.1	PARALLEL DO and END PARALLEL DO Directives
6.1.8.2	PARALLEL SECTIONS and END PARALLEL SECTIONS Directives
6.1.9	Synchronization Constructs
6.1.9.1	ATOMIC Directive
6.1.9.2	BARRIER Directive
6.1.9.3	CRITICAL and END CRITICAL Directives
6.1.9.4	FLUSH Directive
6.1.9.5	MASTER and END MASTER Directives
6.1.9.6	ORDERED and END ORDERED Directives
6.1.10	Specifying Schedule Type and Chunk Size
6.2	Compaq Fortran Parallel Compiler Directives
6.2.1	Command-Line Option and Directives Format
6.2.1.1	Directive Prefixes
6.2.2	Summary Descriptions of Compaq Fortran Parallel Compiler Directives
6.2.3	Parallel Processing Thread Model
6.2.4	Privatizing Named Common Blocks: TASKCOMMON or INSTANCE Directives
6.2.5	Controlling Data Scope Attributes
6.2.6	Parallel Region: PARALLEL and END PARALLEL Directives
6.2.7	Worksharing Constructs
6.2.7.1	PDO and END PDO Directives
6.2.7.2	PSECTIONS, SECTION, and END PSECTIONS Directives
6.2.7.3	SINGLE PROCESS and END SINGLE PROCESS Directives
6.2.8	Combined Parallel/Worksharing Constructs
6.2.8.1	PARALLEL DO and END PARALLEL DO Directives
6.2.8.2	PARALLEL SECTIONS and END PARALLEL SECTIONS Directives
6.2.9	Synchronization Constructs
6.2.9.1	BARRIER Directive
6.2.9.2	CRITICAL SECTION and END CRITICAL SECTION Directives
6.2.10	Specifying a Default Chunk Size
6.2.11	Specifying a Default Schedule Type
6.2.12	Terminating Loop Execution Early: PDONE Directive
6.3	Decomposing Loops for Parallel Processing
6.3.1	Steps in Using Directed Decomposition
6.3.2	Resolving Dependences Manually
6.3.2.1	Resolving Dependences Involving Temporary Variables
6.3.2.2	Resolving Loop-Carried Dependences
6.3.2.3	Loop Alignment
6.3.2.4	Code Replication
6.3.2.5	Loop Distribution
6.3.2.6	Restructuring a Loop into an Inner and Outer Nest
6.3.2.7	Dependences Requiring Locks
6.3.3	Coding Restrictions
6.3.4	Manual Optimization
6.3.4.1	Interchanging Loops
6.3.4.2	Balancing the Workload
6.4	Environment Variables for Adjusting the Run-Time Environment
6.5	Calls to Programs Written in Other Languages
6.6	Compiling, Linking, and Running Parallelized Programs on SMP Systems
6.7	Debugging Parallelized Programs
6.7.1	Debugger Limitations for Parallelized Programs
6.7.2	Debugging Parallel Regions
6.7.3	Debugging Shared Variables