m_mpi_proxy Module Reference

This module serves as a proxy to the parameters and subroutines available in the MPI implementation's MPI module. Specifically, the role of the proxy is to harness basic MPI commands into more complex procedures as to achieve the required communication goals for the post-process. More...

Functions/Subroutines
subroutine	s_initialize_mpi_proxy_module
	Computation of parameters, allocation procedures, and/or any other tasks needed to properly setup the module.
subroutine	s_mpi_bcast_user_inputs
	Since only processor with rank 0 is in charge of reading and checking the consistency of the user provided inputs, these are not available to the remaining processors. This subroutine is then in charge of broadcasting the required information.
subroutine	s_mpi_decompose_computational_domain
	This subroutine takes care of efficiently distributing the computational domain among the available processors as well as recomputing some of the global parameters so that they reflect the configuration of sub-domain that is overseen by the local processor.
subroutine	s_mpi_sendrecv_grid_vars_buffer_regions (pbc_loc, sweep_coord)
	Communicates the buffer regions associated with the grid variables with processors in charge of the neighboring sub-domains. Note that only cell-width spacings feature buffer regions so that no information relating to the cell-boundary locations is communicated.
subroutine	s_mpi_sendrecv_cons_vars_buffer_regions (q_cons_vf, pbc_loc, sweep_coord, q_particle)
	Communicates buffer regions associated with conservative variables with processors in charge of the neighboring sub-domains.
subroutine	s_mpi_gather_spatial_extents (spatial_extents)
	This subroutine gathers the Silo database metadata for the spatial extents in order to boost the performance of the multidimensional visualization.
subroutine	s_mpi_defragment_1d_grid_variable
	This subroutine collects the sub-domain cell-boundary or cell-center locations data from all of the processors and puts back together the grid of the entire computational domain on the rank 0 processor. This is only done for 1D simulations.
subroutine	s_mpi_gather_data_extents (q_sf, data_extents)
	This subroutine gathers the Silo database metadata for the flow variable's extents as to boost performance of the multidimensional visualization.
subroutine	s_mpi_defragment_1d_flow_variable (q_sf, q_root_sf)
	This subroutine gathers the sub-domain flow variable data from all of the processors and puts it back together for the entire computational domain on the rank 0 processor. This is only done for 1D simulations.
subroutine	s_finalize_mpi_proxy_module
	Deallocation procedures for the module.

Variables
Buffers of the conservative variables received/sent from/to neighboring
processors. Note that these variables are structured as vectors rather than arrays.
real(wp), dimension(:), allocatable	q_cons_buffer_in
real(wp), dimension(:), allocatable	q_cons_buffer_out
Receive counts and displacement vector variables, respectively, used in
enabling MPI to gather varying amounts of data from all processes to the root process
integer, dimension(:), allocatable	recvcounts
integer, dimension(:), allocatable	displs
Generic flags used to identify and report MPI errors
integer, private	err_code
integer, private	ierr

Detailed Description

This module serves as a proxy to the parameters and subroutines available in the MPI implementation's MPI module. Specifically, the role of the proxy is to harness basic MPI commands into more complex procedures as to achieve the required communication goals for the post-process.

Function/Subroutine Documentation

s_finalize_mpi_proxy_module()

subroutine m_mpi_proxy::s_finalize_mpi_proxy_module

Deallocation procedures for the module.

Here is the caller graph for this function:

s_initialize_mpi_proxy_module()

subroutine m_mpi_proxy::s_initialize_mpi_proxy_module

Computation of parameters, allocation procedures, and/or any other tasks needed to properly setup the module.

Here is the caller graph for this function:

s_mpi_bcast_user_inputs()

subroutine m_mpi_proxy::s_mpi_bcast_user_inputs

Since only processor with rank 0 is in charge of reading and checking the consistency of the user provided inputs, these are not available to the remaining processors. This subroutine is then in charge of broadcasting the required information.

Here is the caller graph for this function:

s_mpi_decompose_computational_domain()

subroutine m_mpi_proxy::s_mpi_decompose_computational_domain

This subroutine takes care of efficiently distributing the computational domain among the available processors as well as recomputing some of the global parameters so that they reflect the configuration of sub-domain that is overseen by the local processor.

Here is the caller graph for this function:

s_mpi_defragment_1d_flow_variable()

subroutine m_mpi_proxy::s_mpi_defragment_1d_flow_variable	(	real(wp), dimension(0:m), intent(in)	q_sf,
		real(wp), dimension(0:m), intent(inout)	q_root_sf )

This subroutine gathers the sub-domain flow variable data from all of the processors and puts it back together for the entire computational domain on the rank 0 processor. This is only done for 1D simulations.

Parameters

q_sf	Flow variable defined on a single computational sub-domain
q_root_sf	Flow variable defined on the entire computational domain

Here is the caller graph for this function:

s_mpi_defragment_1d_grid_variable()

subroutine m_mpi_proxy::s_mpi_defragment_1d_grid_variable

This subroutine collects the sub-domain cell-boundary or cell-center locations data from all of the processors and puts back together the grid of the entire computational domain on the rank 0 processor. This is only done for 1D simulations.

Here is the caller graph for this function:

s_mpi_gather_data_extents()

subroutine m_mpi_proxy::s_mpi_gather_data_extents	(	real(wp), dimension(:, :, :), intent(in)	q_sf,
		real(wp), dimension(1:2, 0:num_procs - 1), intent(inout)	data_extents )

This subroutine gathers the Silo database metadata for the flow variable's extents as to boost performance of the multidimensional visualization.

Parameters

q_sf	Flow variable defined on a single computational sub-domain
data_extents	The flow variable extents on each of the processor's sub-domain. First dimension of array corresponds to the former's minimum and maximum values, respectively, while second dimension corresponds to each processor's rank.

Here is the caller graph for this function:

s_mpi_gather_spatial_extents()

subroutine m_mpi_proxy::s_mpi_gather_spatial_extents

(

real(wp), dimension(1:, 0:), intent(inout)

spatial_extents

)

This subroutine gathers the Silo database metadata for the spatial extents in order to boost the performance of the multidimensional visualization.

Parameters

spatial_extents

Spatial extents for each processor's sub-domain. First dimension corresponds to the minimum and maximum values, respectively, while the second dimension corresponds to the processor rank.

Here is the caller graph for this function:

s_mpi_sendrecv_cons_vars_buffer_regions()

subroutine m_mpi_proxy::s_mpi_sendrecv_cons_vars_buffer_regions	(	type(scalar_field), dimension(sys_size), intent(inout)	q_cons_vf,
		character(len=3), intent(in)	pbc_loc,
		character, intent(in)	sweep_coord,
		type(scalar_field), intent(inout), optional	q_particle )

Communicates buffer regions associated with conservative variables with processors in charge of the neighboring sub-domains.

Parameters

q_cons_vf	Conservative variables
pbc_loc	Processor boundary condition (PBC) location
sweep_coord	Coordinate direction normal to the processor boundary
q_particle	Projection of the lagrangian particles in the Eulerian framework

Here is the caller graph for this function:

s_mpi_sendrecv_grid_vars_buffer_regions()

subroutine m_mpi_proxy::s_mpi_sendrecv_grid_vars_buffer_regions	(	character(len=3), intent(in)	pbc_loc,
		character, intent(in)	sweep_coord )

Communicates the buffer regions associated with the grid variables with processors in charge of the neighboring sub-domains. Note that only cell-width spacings feature buffer regions so that no information relating to the cell-boundary locations is communicated.

Parameters

pbc_loc	Processor boundary condition (PBC) location
sweep_coord	Coordinate direction normal to the processor boundary

Here is the caller graph for this function:

Variable Documentation

displs

integer, dimension(:), allocatable m_mpi_proxy::displs

err_code

integer, private m_mpi_proxy::err_code

private

ierr

integer, private m_mpi_proxy::ierr

private

q_cons_buffer_in

real(wp), dimension(:), allocatable m_mpi_proxy::q_cons_buffer_in

q_cons_buffer_out

real(wp), dimension(:), allocatable m_mpi_proxy::q_cons_buffer_out

recvcounts

integer, dimension(:), allocatable m_mpi_proxy::recvcounts