(git:27c966b)
Loading...
Searching...
No Matches
dbt_methods.F File Reference

Go to the source code of this file.

Modules

module  dbt_methods
 DBT tensor framework for block-sparse tensor contraction. Representation of n-rank tensors as DBT tall-and-skinny matrices. Support for arbitrary redistribution between different representations. Support for arbitrary tensor contractions.

Functions/Subroutines

subroutine, public dbt_methods::dbt_library_init ()
 Initialize the DBT library (after dbm_library_init).
subroutine, public dbt_methods::dbt_library_finalize ()
 Finalize the DBT library (before dbm_library_finalize), which releases its buffers.
subroutine, public dbt_methods::dbt_clear_mempools ()
 Release the buffers that persist across calls (as dbcsr_clear_mempools), e.g., once a phase with large tensor operations ends. They are allocated again when needed.
subroutine, public dbt_methods::dbt_library_print_stats (mpi_comm, output_unit)
 Print per-contraction statistics (rank-local table, totals over all ranks).
subroutine, public dbt_methods::dbt_copy (tensor_in, tensor_out, order, summation, bounds, move_data, unit_nr)
 Copy tensor data. Redistributes tensor data according to distributions of target and source tensor. Permutes tensor index according to order argument (if present). Source and target tensor formats are arbitrary as long as the following requirements are met:
subroutine, public dbt_methods::dbt_copy_matrix_to_tensor (matrix_in, tensor_out, summation)
 copy matrix to tensor.
subroutine, public dbt_methods::dbt_copy_tensor_to_matrix (tensor_in, matrix_out, summation)
 copy tensor to matrix
subroutine, public dbt_methods::dbt_contract (alpha, tensor_1, tensor_2, beta, tensor_3, contract_1, notcontract_1, contract_2, notcontract_2, map_1, map_2, bounds_1, bounds_2, bounds_3, optimize_dist, pgrid_opt_1, pgrid_opt_2, pgrid_opt_3, filter_eps, flop, move_data, retain_sparsity, unit_nr, log_verbose)
 Contract tensors by multiplying matrix representations. tensor_3(map_1, map_2) := alpha * tensor_1(notcontract_1, contract_1).
subroutine, public dbt_methods::dbt_batched_contract_init (tensor, batch_range_1, batch_range_2, batch_range_3, batch_range_4)
 Initialize batched contraction for this tensor.
subroutine, public dbt_methods::dbt_batched_contract_finalize (tensor, unit_nr)
 finalize batched contraction. This performs all communication that has been postponed in the contraction calls.

Variables

type(contract_stats_type), dimension(:), allocatable, save dbt_methods::contract_stats