PG Namespace Reference

Parallel globals namespace — MPI rank/size and message tag definitions. More...

Enumerations
enum	{   server_branch_this , server_done_simulating , client_simulate_this , client_restart_run ,   client_shutdown , client_debug_resume , server_branch_meta , client_peer_assign ,   client_peer_restart , client_send_to_peer , client_run_held , server_send_complete ,   server_done_with_report , server_output_data , client_checkpoint }
	MPI message tags for Server/Client communication. More...

Variables
int	rank
	MPI rank of this process. More...
int	size
	Total number of MPI processes. More...
bool	log = true
	Enable/disable rank-specific logging. More...

Detailed Description

Parallel globals namespace — MPI rank/size and message tag definitions.

Enumeration Type Documentation

anonymous enum

anonymous enum

MPI message tags for Server/Client communication.

Enumerator
server_branch_this	Unused placeholder (occupies tag value 0)
server_done_simulating	Client -> Server: empty completion signal, client is free. The server handles it; Client.cpp sends server_done_with_report instead.
client_simulate_this	Server -> Client: run the initial (first) sequence.
client_restart_run	Server -> Client: run a pending task of a resumed checkpoint with server-provided data.
client_shutdown	Server -> Client: no more work, shut down.
client_debug_resume	Unused (no sender or handler)
server_branch_meta	Client A -> Server: branch metadata only (rank-local branch id + num_branches)
client_peer_assign	Server -> Client A: peer assignment list for direct sends.
client_peer_restart	Client A -> Client B: direct peer-to-peer restart data.
client_send_to_peer	Server -> Client A: send held data to another client.
client_run_held	Server -> Client A: run task using own held data.
server_send_complete	Client -> Server: data send finished, client is free again.
server_done_with_report	Client -> Server: sequence completed with SequenceReport payload.
server_output_data	Client -> Server: SequenceOutputData (before report)
client_checkpoint	Server -> Client: send held_data for checkpoint gather.

Definition at line 177 of file ParallelGlobal.h.

Variable Documentation

log

bool PG::log = true

inline

Enable/disable rank-specific logging.

Definition at line 174 of file ParallelGlobal.h.

Referenced by Probability::beta_inc(), Timer::calculateRequiredTime(), Client::complete_sequence(), Server::dispatch_available_tasks(), Server::dispatch_one_resume_task(), Server::dispatch_resume_tasks(), Probability::gamma_log(), Client::hold_branch_data(), Server::merge_outputs(), AdsidacCallback::onBranchingPoint(), Client::parallel_restartforanewsequence_(), Server::Server(), DiagnosisActionUnit::setActionPrereqsEvaluated(), InitiatingEventController::setNeedToCheckForActivation(), Server::Shutdown(), Server::Simulation(), Client::Work(), and Server::write_checkpoint().

rank

int PG::rank

inline

MPI rank of this process.

Definition at line 172 of file ParallelGlobal.h.

Referenced by log_seq_fingerprint_mpi(), main(), and AdsidacCallback::onBranchingPoint().

size

int PG::size

inline

Total number of MPI processes.

Definition at line 173 of file ParallelGlobal.h.

Referenced by ads_membuf_get_trimmed_(), bncore::BatchWorkspace::build_message_schedule(), bncore::BatchWorkspace::build_node_to_clique_map(), InformationLoadModule::calculateAverageInformationLoadPIFValue(), DiagnoseInvestigator::CalculateTheConfidenceLeveloftheDiagnoseInvestigator(), Server::find_free_client_except(), Server::find_next_free_client_round_robin(), lz4_compress(), lz4_decompress(), Server::merge_outputs(), bncore::BatchWorkspace::rebuild_clique_potentials(), Server::Server(), Server::Shutdown(), Server::Simulation(), Server::write_checkpoint(), and write_file_or_die().