Lucas Nesi 03d52ba8eb add priorities on data requests from mpi 4 éve
..
load_balancer 45572515f0 mpi: Fix data_movement interface 4 éve
mpi 03d52ba8eb add priorities on data requests from mpi 4 éve
nmad 72e39acbfe nmad: decrement number of pending requests just after destroying a request 4 éve
Makefile.am c08118bb6d mpi/src/Makefile.am: add missing SUBDIRS 4 éve
starpu_mpi.c 03d52ba8eb add priorities on data requests from mpi 4 éve
starpu_mpi_cache.c 9054490002 happy 2021 4 éve
starpu_mpi_cache.h 9054490002 happy 2021 4 éve
starpu_mpi_cache_stats.c 9054490002 happy 2021 4 éve
starpu_mpi_cache_stats.h 9054490002 happy 2021 4 éve
starpu_mpi_collective.c 9054490002 happy 2021 4 éve
starpu_mpi_coop_sends.c 03d52ba8eb add priorities on data requests from mpi 4 éve
starpu_mpi_datatype.c 5ce089e2d4 Add starpu_mpi_datatype_node_register and starpu_mpi_interface_datatype_node_register 4 éve
starpu_mpi_datatype.h 5ce089e2d4 Add starpu_mpi_datatype_node_register and starpu_mpi_interface_datatype_node_register 4 éve
starpu_mpi_fortran.c 9054490002 happy 2021 4 éve
starpu_mpi_fxt.h 4a18080596 Store type (point2point or collective) and priority of comms in traces 4 éve
starpu_mpi_helper.c 9054490002 happy 2021 4 éve
starpu_mpi_init.c 9054490002 happy 2021 4 éve
starpu_mpi_init.h 9054490002 happy 2021 4 éve
starpu_mpi_private.c 29e210dd15 mpi/nmad: fix formatting and add comments 4 éve
starpu_mpi_private.h 7a7abdc1c7 mpi: perform reduction only over contributing nodes 4 éve
starpu_mpi_req.c bf516bbc43 mpi: defer choosing the target node to when data acquisition is complete semantically 4 éve
starpu_mpi_select_node.c 9054490002 happy 2021 4 éve
starpu_mpi_select_node.h 9054490002 happy 2021 4 éve
starpu_mpi_stats.c 9054490002 happy 2021 4 éve
starpu_mpi_stats.h 9054490002 happy 2021 4 éve
starpu_mpi_task_insert.c 03d52ba8eb add priorities on data requests from mpi 4 éve
starpu_mpi_task_insert.h 9054490002 happy 2021 4 éve
starpu_mpi_task_insert_fortran.c ae6818ee93 revert fortran task insert ; remove init after redux_data ; example now uses 2 contributing nodes over 1 handle instead of 2 over 2 4 éve