-------------------------------------------------------------------------------- UTILITY -------------------------------------------------------------------------------- -debug_parser_input Print a single preprocessed input file before sending to parser and exit -debug_preprocessor_input Print a single pre-annotated input file before sending to preprocessor and exit -dump-options Create file options.cetus with default options -dump-system-options Create system wide file options.cetus with default options -expand-all-header Expand all header file #includes into code -expand-user-header Expand user (non-standard) header file #includes into code -help Print this message -load-options Load options from file options.cetus -macro Sets macros for the specified names with comma-separated list (no space is allowed). e.g., -macro=ARCH=i686,OS=linux -outdir=dirname Set the output directory name (default is cetus_output) -parser=parsername Name of parser to be used for parsing source file -preprocessor=command Set the preprocessor command to use -preserve-KR-function Preserves K&R-style function declaration -skip-procedures=proc1,proc2,... Causes all passes that observe this flag to skip the listed procedures -verbosity=N Degree of status messages (0-4) that you wish to see (default is 0) -version Print the version information -------------------------------------------------------------------------------- ANALYSIS -------------------------------------------------------------------------------- -alias=N Specify level of alias analysis =0 disable alias analysis (assume no alias) =1 advanced interprocedural analysis (default) Uses interprocedural points-to analysis =2 assume no alias when points-to analysis is too conservative -callgraph Print the static call graph to stdout -ddt=N Perform Data Dependence Testing =1 banerjee-wolfe test (default) =2 range test -parallelize-loops Annotate loops with Parallelization decisions =1 parallelizes outermost loops (default) =2 parallelizes every loop =3 parallelizes outermost loops with report =4 parallelizes every loop with report -privatize Perform scalar/array privatization analysis -range=N Specifies the accuracy of symbolic analysis with value ranges =0 disable range computation (minimal symbolic analysis) =1 enable local range computation (default) =2 enable inter-procedural computation (experimental) -reduction=N Perform reduction variable analysis =1 enable only scalar reduction analysis (default) =2 enable array reduction analysis and transformation -------------------------------------------------------------------------------- TRANSFORM -------------------------------------------------------------------------------- -induction Perform induction variable substitution -normalize-loops Normalize for loops so they begin at 0 and have a step of 1 -normalize-return-stmt Normalize return statements for all procedures -profile-loops=N Inserts loop-profiling calls =1 every loop =2 outermost loop =3 every omp parallel =4 outermost omp parallel =5 every omp for =6 outermost omp for -teliminate-branch=N Eliminates unreachable branch targets =0 disable (default) =1 enable =2 leave old statements as comments -tinline=mode=0|1|2|3|4:depth=0|1:pragma=0|1:debug=0|1:foronly=0|1:complement=0|1:functions=foo,bar,... (Experimental) Perform simple subroutine inline expansion tranformation mode =0 inline inside main function (default) =1 inline inside selected functions provided in the "functions" sub-option =2 inline selected functions provided in the "functions" sub-option, when invoked =3 inline according to the "inlinein" pragmas =4 inline according to both "inlinein" and "inline" pragmas depth =0 perform inlining recursively i.e. within callees (and their callees) as well (default) =1 perform 1-level inlining pragma =0 do not honor "noinlinein" and "noinline" pragmas =1 honor "noinlinein" and "noinline" pragmas (default) debug =0 remove inlined (and other) functions if they are no longer executed (default) =1 do not remove the inlined (and other) functions even if they are no longer executed foronly =0 try to inline all function calls depending on other options (default) =1 try to inline function calls inside for loops only complement =0 consider the functions provided in the command line with "functions" sub-option (default) =1 consider all functions except the ones provided in the command line with "functions" sub-option functions =[comma-separated list] consider the provided functions. (Note 1: This sub-option is meaningful for modes 1 and 2 only) (Note 2: It is used with "complement" sub-option to determine which functions should be considered.) -tsingle-call Transform all statements so they contain at most one function call -tsingle-declarator Transform all variable declarations so they contain at most one declarator -tsingle-return Transform all procedures so they have a single return statement -------------------------------------------------------------------------------- CODEGEN -------------------------------------------------------------------------------- -ompGen=N Generate OpenMP pragma =1 comment out existing OpenMP pragmas (default) =2 remove existing OpenMP pragmas =3 remove existing OpenMP and Cetus pragmas =4 keep all pragmas -profitable-omp=N Inserts runtime for selecting profitable omp parallel region (See the API documentation for more details) =0 disable =1 Model-based loop selection (default) =2 Profile-based loop selection