|
| 1 | +.\" -*- nroff -*- |
| 2 | +.\" Copyright (c) 2015 Research Organization for Information Science |
| 3 | +.\" and Technology (RIST). All rights reserved. |
| 4 | +.\" $COPYRIGHT$ |
| 5 | +.TH MPI_Win_create_dynamic 3 "#OMPI_DATE#" "#PACKAGE_VERSION#" "#PACKAGE_NAME#" |
| 6 | +.SH NAME |
| 7 | +\fBMPI_Win_create_dynamic\fP \- One-sided MPI call that returns a window object for RMA operations. |
| 8 | + |
| 9 | +.SH SYNTAX |
| 10 | +.ft R |
| 11 | +.SH C Syntax |
| 12 | +.nf |
| 13 | +#include <mpi.h> |
| 14 | +MPI_Win_create_dynamic(MPI_Info \fIinfo\fP, MPI_Comm \fIcomm\fP, MPI_Win *\fIwin\fP) |
| 15 | + |
| 16 | +.fi |
| 17 | +.SH Fortran Syntax |
| 18 | +.nf |
| 19 | +INCLUDE 'mpif.h' |
| 20 | +MPI_WIN_CREATE_DYNAMIC(\fIINFO, COMM, WIN, IERROR\fP) |
| 21 | + INTEGER \fIINFO, COMM, WIN, IERROR\fP |
| 22 | + |
| 23 | +.fi |
| 24 | +.SH INPUT PARAMETERS |
| 25 | +.ft R |
| 26 | +.TP 1i |
| 27 | +info |
| 28 | +Info argument (handle). |
| 29 | +.TP 1i |
| 30 | +comm |
| 31 | +Communicator (handle). |
| 32 | + |
| 33 | +.SH OUTPUT PARAMETERS |
| 34 | +.ft R |
| 35 | +.TP 1i |
| 36 | +win |
| 37 | +Window object returned by the call (handle). |
| 38 | +.TP 1i |
| 39 | +IERROR |
| 40 | +Fortran only: Error status (integer). |
| 41 | + |
| 42 | +.SH DESCRIPTION |
| 43 | +.ft R |
| 44 | +MPI_Win_create_dynamic is a one-sided MPI communication collective call executed by all processes in the group of \fIcomm\fP. It returns a window object without memory attached that can be used by these processes to perform RMA operations. |
| 45 | +.sp |
| 46 | +The following info keys are supported: |
| 47 | +.ft R |
| 48 | +.TP 1i |
| 49 | +no_locks |
| 50 | +If set to \fItrue\fP, then the implementation may assume that the local |
| 51 | +window is never locked (by a call to MPI_Win_lock or |
| 52 | +MPI_Win_lock_all). Setting this value if only active synchronization |
| 53 | +may allow the implementation to enable certain optimizations. |
| 54 | +.sp |
| 55 | +.TP 1i |
| 56 | +accumulate_ordering |
| 57 | +By default, accumulate operations from one initiator to one target on |
| 58 | +the same window are strictly ordered. If the info key |
| 59 | +accumulate_ordering is set to \fInone\fP, no ordering of accumulate |
| 60 | +operations guaranteed. They key can also be a comma-separated list of |
| 61 | +required orderings consisting of \fIrar\fP, \fIwar\fP, \fIraw\fP, and \fIwaw\fP for |
| 62 | +read-after-read, write-after-read, read-after-write, and |
| 63 | +write-after-write, respectively. Looser ordering constraints are |
| 64 | +likely to result in improved performance. |
| 65 | +.sp |
| 66 | +.TP 1i |
| 67 | +accumulate_ops |
| 68 | +If set to \fIsame_op\fP, the implementation will assume that all concurrent |
| 69 | +accumulate calls to the same target address will use the same |
| 70 | +operation. If set to \fIsame_op_no_op\fP, then the implementation will |
| 71 | +assume that all concurrent accumulate calls to the same target address |
| 72 | +will use the same operation or MPI_NO_OP. The default is \fIsame_op_no_op\fP. |
| 73 | + |
| 74 | +.SH ERRORS |
| 75 | +Almost all MPI routines return an error value; C routines as the value of the function and Fortran routines in the last argument. |
| 76 | +.sp |
| 77 | +Before the error value is returned, the current MPI error handler is |
| 78 | +called. By default, this error handler aborts the MPI job, except for I/O function errors. The error handler may be changed with MPI_Comm_set_errhandler; the predefined error handler MPI_ERRORS_RETURN may be used to cause error values to be returned. Note that MPI does not guarantee that an MPI program can continue past an error. |
| 79 | + |
| 80 | + |
| 81 | + |
0 commit comments