Structure of MPI-related variables.
This structure contains all the variables needed for parallel computations using the MPI library.
Definition at line 24 of file mpivars_struct.h.
| Data Fields |
|---|
|
int |
rank |
Process rank
|
|
int |
nproc |
Total number of processes
|
|
int * |
iproc |
Number of processes along each dimension
|
|
int * |
ip |
Process rank along each dimension
|
|
int * |
is |
Global start index of local domain along each dimension
|
|
int * |
ie |
Global end index of local domain along each dimension
|
|
int * |
bcperiodic |
Flag for periodic BCs along any dimension
|
|
MPI_Comm |
world |
Communicator for all processes
|
|
MPI_Comm * |
comm |
Sub-communicators
|
|
int |
N_IORanks |
Number of IO ranks
|
|
int |
IOParticipant |
Whether this rank will handle file I/O
|
|
int |
CommGroup |
I/O group this rank is a part of
|
|
int |
IORank |
Rank of the process this rank will get I/O from
|
|
int |
GroupStartRank |
Starting rank of the IO group
|
|
int |
GroupEndRank |
Last rank of the IO group
|
|
MPI_Comm |
IOWorld |
Communicator of processes participating in file I/O
|
|
double * |
sendbuf |
Buffer to send data
|
|
double * |
recvbuf |
Buffer to receive data
|
|
int |
maxbuf |
Maximum buffer size
|
|
int |
ncalls |
|
|
double |
wctime |
|
|
double |
wctime_total |
|
|
int * |
cpu_dim |
|
|
double * |
gpu_sendbuf |
|
|
double * |
gpu_recvbuf |
|