Class to generate the directed acyclic graph (DAG) for evaluation. Determined which Evaluators should be called and the order in which to call them such that all dependencies are met with consistency. More...
#include <Phalanx_DAG_Manager.hpp>
Public Member Functions | |
DagManager (const std::string &evaluator_type_name="???") | |
void | requireField (const PHX::FieldTag &v) |
Require a variable to be evaluated. | |
void | registerEvaluator (const Teuchos::RCP< PHX::Evaluator< Traits > > &p) |
Registers an evaluator with the manager. | |
void | setDefaultGraphvizFilenameForErrors (const std::string &file_name) |
Sets the default filename for graphiz file generation for DAG construction errors. | |
void | setWriteGraphvizFileOnError (bool write_file) |
If set to true, a graphviz file will be written during for DAG construction errors. | |
void | sortAndOrderEvaluators () |
void | postRegistrationSetup (typename Traits::SetupData d, PHX::FieldManager< Traits > &vm, const bool &buildDeviceDAG) |
void | evaluateFields (typename Traits::EvalData d) |
Evaluate the required fields using data parallel evaluation on topological sort of tasks. Calls parallel_for for each node in DAG. | |
void | evaluateFieldsDeviceDag (const int &work_size, const int &team_size, const int &vector_size, typename Traits::EvalData d) |
Evaluate the required fields using data parallel evaluation on topological sort of tasks. Uses Device DAG support, calling a single parallel_for for the entire DAG. This could be faster than the call to evaluateFields, but all nodes in the DAG are restricted to the same work_size. This is intended for CUDA builds where kernel launch overhead can be significant. More... | |
void | preEvaluate (typename Traits::PreEvalData d) |
This routine is called before each residual/Jacobian fill. More... | |
void | postEvaluate (typename Traits::PostEvalData d) |
This routine is called after each residual/Jacobian fill. More... | |
void | setEvaluationTypeName (const std::string &evaluation_type_name) |
const std::vector < Teuchos::RCP< PHX::FieldTag > > & | getFieldTags () |
bool | sortingCalled () const |
void | writeGraphvizFile (const std::string filename, bool writeEvaluatedFields, bool writeDependentFields, bool debugRegisteredEvaluators) const |
void | writeGraphvizFileNew (const std::string filename, bool writeEvaluatedFields, bool writeDependentFields) const |
void | print (std::ostream &os) const |
Printing. | |
const std::vector< int > & | getEvaluatorInternalOrdering () const |
Returns the Topological sort ordering. Used for unit testing. | |
const std::vector < PHX::DagNode< Traits > > & | getDagNodes () const |
Returns the internally registered nodes. Used for unit testing. | |
void | analyzeGraph (double &speedup, double ¶llelizability) const |
Returns the speedup and parallelizability of the graph. More... | |
std::vector< Teuchos::RCP < PHX::Evaluator< Traits > > > & | getEvaluatorsBindingField (const PHX::FieldTag &ft) |
Returns all evaluators that either evaluate or require the given field. This is used to bind memory for unmanaged views. More... | |
void | printEvaluatorStartStopMessage (const Teuchos::RCP< std::ostream > &ostr) |
Print to user specified ostream when each evaluator starts and stops. Useful for debugging. Enabled only in debug builds. More... | |
Protected Member Functions | |
void | dfsVisit (PHX::DagNode< Traits > &node, int &time) |
Depth-first search algorithm. | |
void | writeGraphvizDfsVisit (PHX::DagNode< Traits > &node, std::vector< PHX::DagNode< Traits >> &nodes_copy, std::ostream &os, const bool writeEvaluatedFields, const bool writeDependentFields) const |
Depth-first search algorithm specialized for writing graphviz output. | |
void | printEvaluator (const PHX::Evaluator< Traits > &e, std::ostream &os) const |
Helper function. | |
void | createEvaluatorBindingFieldMap () |
Protected Attributes | |
std::vector< Teuchos::RCP < PHX::FieldTag > > | required_fields_ |
Fields required by the user. | |
std::vector< PHX::DagNode < Traits > > | nodes_ |
Vector of all registered evaluators. More... | |
std::unordered_map < std::string, int > | field_to_node_index_ |
Hash map of field key to evaluator index. | |
std::unordered_map < std::string, std::unordered_set< int > > | contributed_field_to_node_index_ |
Hash map of contributed field key to evaluator index. | |
std::vector< Teuchos::RCP < PHX::FieldTag > > | fields_ |
All fields that are needed for the evaluation. | |
std::vector< Teuchos::RCP < Teuchos::Time > > | evalTimers |
Evaluation Order Objects | |
Stores results from a topological sort on the evaluator DAG: the order to call evaluators to evaluate fields correctly. | |
std::vector< int > | topoSortEvalIndex |
std::string | graphviz_filename_for_errors_ |
Use this name for graphviz file output for DAG construction errors. | |
bool | write_graphviz_file_on_error_ |
If set to true, will write graphviz file for DAG construction errors. | |
std::string | evaluation_type_name_ |
bool | sorting_called_ |
Flag to tell the setup has been called. | |
bool | allow_multiple_evaluators_for_same_field_ |
Backwards compatibility option: set to true to disable a check that throws if multiple registered evaluators can evaluate the same field. Original DFS algortihm allowed this. Refactor checks and throws. | |
std::unordered_map < std::string, std::vector < Teuchos::RCP< PHX::Evaluator < Traits > > > > | field_to_evaluators_binding_ |
A map that returns all evaluators that bind the memory of a particular field. Key is unique field identifier. | |
bool | build_device_dag_ |
If set to true, allocated DeviceEvaluators for Device DAG for evaluation. | |
Kokkos::View < PHX::DeviceEvaluatorPtr < Traits > *, PHX::Device > | device_evaluators_ |
Contians pointers to DeviceEvaluators for Device DAG support. | |
Teuchos::RCP< std::ostream > | start_stop_debug_ostream_ |
If non-null, in debug builds, the DAG manager will print when an evaluator starts and stops. | |
Class to generate the directed acyclic graph (DAG) for evaluation. Determined which Evaluators should be called and the order in which to call them such that all dependencies are met with consistency.
void PHX::DagManager< Traits >::analyzeGraph | ( | double & | speedup, |
double & | parallelizability | ||
) | const |
Returns the speedup and parallelizability of the graph.
Estimates are based on execution times. This will return garbage if the evaluateFields() call has not been made to log execution times.
void PHX::DagManager< Traits >::evaluateFieldsDeviceDag | ( | const int & | work_size, |
const int & | team_size, | ||
const int & | vector_size, | ||
typename Traits::EvalData | d | ||
) |
Evaluate the required fields using data parallel evaluation on topological sort of tasks. Uses Device DAG support, calling a single parallel_for for the entire DAG. This could be faster than the call to evaluateFields, but all nodes in the DAG are restricted to the same work_size. This is intended for CUDA builds where kernel launch overhead can be significant.
The parallel_for kernel launch below will not compile on CUDA unless relocatable device code (RDC) is enabled for the nvcc compiler. We also want to build and run phalanx without Device DAG support on CUDA (i.e. RDC off), so this ifdef will hide the RDC required code.
std::vector< Teuchos::RCP< PHX::Evaluator< Traits > > > & PHX::DagManager< Traits >::getEvaluatorsBindingField | ( | const PHX::FieldTag & | ft | ) |
Returns all evaluators that either evaluate or require the given field. This is used to bind memory for unmanaged views.
CAUTION: The returned vector is non-const to rebind memory for fields in evaluators. Be careful not to corrupt the actual vector.
References PHX::FieldTag::identifier().
const std::vector< Teuchos::RCP< PHX::FieldTag > > & PHX::DagManager< Traits >::getFieldTags | ( | ) |
Returns the FieldTags for all fields involved in the evaluation. Will return an empty vector unless the user has built the DAG using one of the following calls: postRegistrationSetup(), postRegistrationSetupForType() or buildDagForType().
WARNING: This is a dangerous power user feature. It returns non-const field tags so that the fields can be sized after the DAG has been created.
void PHX::DagManager< Traits >::postEvaluate | ( | typename Traits::PostEvalData | d | ) |
This routine is called after each residual/Jacobian fill.
This routine is called ONCE on the evaluator after the fill loop over elements is completed. This allows us to evaluate any post fill data. An example is to print out some statistics such as the maximum grid peclet number in a cell.
void PHX::DagManager< Traits >::postRegistrationSetup | ( | typename Traits::SetupData | d, |
PHX::FieldManager< Traits > & | vm, | ||
const bool & | buildDeviceDAG | ||
) |
Calls post registration setup on all evaluators.
void PHX::DagManager< Traits >::preEvaluate | ( | typename Traits::PreEvalData | d | ) |
This routine is called before each residual/Jacobian fill.
This routine is called ONCE on the evaluator before the fill loop over elements is started. This allows us to reset global objects between each fill. An example is to reset an evaluator that monitors the maximum grid peclet number in a cell. This call would zero out the maximum for a new fill.
void PHX::DagManager< Traits >::printEvaluatorStartStopMessage | ( | const Teuchos::RCP< std::ostream > & | ostr | ) |
Print to user specified ostream when each evaluator starts and stops. Useful for debugging. Enabled only in debug builds.
[in] | ostr | RCP to output stream. If set to null, this disables printing. |
void PHX::DagManager< Traits >::sortAndOrderEvaluators | ( | ) |
Builds the evaluation DAG. This should only be called after all required fields and evaluators are registered. Must be called prior to making calls to postRegistrationSetup(), evaluateFields(), preEvaluate(), and postEvaluate(). This can be called multiple times to build a new DAG if requirements have changed or more evaluators have been added.
|
protected |
Vector of all registered evaluators.
This list may include more nodes than what is needed for the DAG evaluation of required fields.