name: Splitter

synopsis:


g++ [flags ...] file ... -l /isip/tools/lib/$ISIP_BINARY/lib_mmedia.a

#include <Splitter.h>

static boolean diagnose(Integral::DEBUG debug_level);
SPLIT_MODE getSplitMode();
boolean setSplitMode(SPLIT_MODE arg);
Filename& getParamFile();
boolean setParamFile(Filename& arg);
Filename& getIdentifierList();
boolean setIdentifierList(Filename& arg);
Filename& getMachineList();
boolean setMachineList(Filename& arg);
Filename& getAccumulatorList();
boolean setAccumulatorList(Filename& arg);
MachineDatabase& getMachineDatabase();
boolean getMachineDatabase(MachineDatabase& arg);
boolean createUtterLists(Filename& identifier_list, Filename& machine_list, Filename& accumulator_list, Vector<Filename>& identifiers_lookup, Vector<Machine>& machine_lookup, Vector<Filename>& accumulator_lookup, Vector<SingleLinkedList<Filename> >& items);
boolean divideAndConquer(Filename& identifier_list, Filename& machine_list, Vector<SingleLinkedList<Filename> >& items);

quick start:


Splitter splt_00;

splt_00.setSplitMode(Splitter::SPEED);
splt_00.run();

description:

This class is used to split a file list and launch jobs on machines during parallel training of models in the Baum-Welch re-estimation phase.

dependencies:

String.h

public constants:

define the class name:

static const String CLASS_NAME = L"Splitter";

define the split mode choices:

enum SPLIT_MODE { NONE = 0, SPEED, MEMORY, DEF_SPLIT_MODE = NONE };

define the NameMap objects:
```
static const NameMap SPLIT_MAP;
```

define i/o related constants:

static const String DEF_PARAM;

static const String PARAM_SPLIT_MODE;

static const String PARAM_PARAMETER_FILE;

static const String PARAM_MACHINE_LIST;

static const String PARAM_IDENTIFIER_LIST;

static const String PARAM_ACCUMULATOR_LIST;

static const String PARAM_MACHINE_DATABASE;

define default values and arguments:

static const String DEF_RECOGNIZER;

static const String DEF_RSH_PROTOCOL;

static const String DEF_SSH_PROTOCOL;

define option flags:

static const String OPTION_PARAM_FLAG;

static const String OPTION_LIST_FLAG;

static const String OPTION_ACCUM_FLAG;

error codes:

error codes indicating general errors:

protected data:

declare the split mode:

SPLIT_MODE split_mode_d;

declare the parameter file name:

Filename param_file_d;

declare the list of identifiers:

Filename ident_list_d;

declare the list of machines:

Filename machine_list_d;

declare the list of accumulators:

Filename accum_list_d;

declare the machine database:

MachineDatabase mdb_d;

declare the debugging parameters:
```
static Integral::DEBUG debug_level_d;
```
declare the memory manager:
```
static MemoryManager mgr_d;
```

required public methods:

static methods:
```
static const String& name();
```

debug methods:

boolean debug(const unichar* msg) const;

static boolean setDebug(Integral::DEBUG arg);

static boolean diagnose(Integral::DEBUG debug_level);

destructor/constructor(s):

~Splitter();

Splitter();

Splitter(const Splitter& arg);

assign methods:
```
boolean assign(const Splitter& arg);
```
operator= method:
```
not present.
```

i/o methods:

long sofSize() const;

boolean read(Sof& sof, long tag);

boolean read(Sof& sof, long tag, const String& name);

boolean write(Sof& sof, long tag) const;

boolean write(Sof& sof, long tag, const String& name) const;

boolean readData(Sof& sof, const String& pname = DEF_PARAM, long size = SofParser::FULL_OBJECT, boolean param = true, boolean nested = false);

boolean writeData(Sof& sof, const String& pname =  DEF_PARAM) const;

equality methods:
```
boolean eq(const Splitter& arg) const;
```

memory management methods:

static void* operator new(size_t arg);

static void* operator new[](size_t arg);

static void operator delete(void* arg);

static void operator delete[](void* arg);

static boolean setGrowSize(long arg);

boolean clear(Integral::CMODE cmode = Integral::DEF_CMODE);

class-specific public methods:

declare split mode methods:

SPLIT_MODE getSplitMode();

boolean setSplitMode(SPLIT_MODE arg);

declare parameter file methods:

Filename& getParamFile();

boolean setParamFile(Filename& arg);

declare identifier methods:

Filename& getIdentifierList();

boolean setIdentifierList(Filename& arg);

declare machine list methods:

Filename& getMachineList();

boolean setMachineList(Filename& arg);

declare accumulator list methods:

Filename& getAccumulatorList();

boolean setAccumulatorList(Filename& arg);

declare machine database methods:

MachineDatabase& getMachineDatabase();

boolean getMachineDatabase(MachineDatabase& arg);

declare splitting methods:

boolean createUtterLists(Filename& identifier_list, Filename& machine_list, Filename& accumulator_list, Vector<Filename>& identifiers_lookup, Vector<Machine>& machine_lookup, Vector<Filename>& accumulator_lookup, Vector<SingleLinkedList<Filename> >& items);

boolean divideAndConquer(Filename& identifier_list, Filename& machine_list, Vector<SingleLinkedList<Filename> >& items);

boolean splitJobList(Filename& input_file, Filename& cpus, String& command_to_execute, String& walltime, String& batch_command, Sdb& sdb_list, String& cluster_name);

boolean pollLogFile(Vector& log_files);

boolean waitAndAccumulate(Vector& log_files, String& proc_id_str);

boolean sendEmail(Long& start_time, Long& finish_time, String& begin_time, String& end_time, String& proc_id_str);

String getDiffTime(Long& start_time, Long& finish_time);

declare cluster specific methods:

private methods:

declare private methods:

examples:

This example shows how to use the Splitter object.

notes:

none.