PetscBinarySynchronizedRead#

Reads from a binary file, all MPI processes get the same values

Synopsis#

PetscErrorCode PetscBinarySynchronizedRead(MPI_Comm comm, int fd, void *data, PetscInt num, PetscInt *count, PetscDataType type)

Collective

Input Parameters#

  • comm - the MPI communicator

  • fd - the file descriptor

  • num - the maximum number of items to read

  • type - the type of items to read (PETSC_INT, PETSC_REAL, PETSC_SCALAR, etc.)

Output Parameters#

  • data - the buffer, an array of the type that matches the value in type

  • count - the number of items read, optional

Notes#

Does a PetscBinaryRead() followed by an MPI_Bcast()

If count is not provided and the number of items read is less than the maximum number of items to read, then this routine errors.

PetscBinarySynchronizedRead() uses byte swapping to work on all machines. The files are written using big-endian ordering to the file. On little-endian machines the numbers are converted to the big-endian format when they are written to disk. When PETSc is configured using ./configure with --with-64-bit-indices the integers are written to the file as 64-bit integers, this means they can only be read back in when the option --with-64-bit-indices is used.

See Also#

PetscBinaryWrite(), PetscBinaryOpen(), PetscBinaryClose(), PetscBinaryRead(), PetscBinarySynchronizedWrite(), PetscBinarySynchronizedSeek()

Level#

developer

Location#

src/sys/fileio/sysio.c

Edit on GitLab

Index of all Sys routines
Table of Contents for all manual pages
Index of all manual pages