rank = _basics.rank
local_rank = _basics.local_rank
cross_rank = _basics.cross_rank
mpi_threads_supported = _basics.mpi_threads_supported
mpi_enabled = _basics.mpi_enabled
mpi_built = _basics.mpi_built
gloo_enabled = _basics.gloo_enabled
gloo_built = _basics.gloo_built
nccl_built = _basics.nccl_built
ddl_built = _basics.ddl_built
ccl_built = _basics.ccl_built
cuda_built = _basics.cuda_built
rocm_built = _basics.rocm_built
dll_path = os.path.join(os.path.dirname(__file__),
'mpi_lib' + get_ext_suffix())
MPI_MXNET_LIB_CTYPES = ctypes.CDLL(dll_path, ctypes.RTLD_GLOBAL)
def allreduce(tensor,
average=True,
name=None,
priority=0,
prescale_factor=1.0,
postscale_factor=1.0):
"""
A function that performs averaging or summation of the input tensor over
all the Horovod processes. The input tensor is not modified.
The reduction operation is keyed by the name. If name is not provided, an
incremented auto-generated name is used. The tensor type and shape must be
def _load_library(name):
"""Loads a .so file containing the specified operators.
Args:
name: The name of the .so file to load.
Raises:
NotFoundError if were not able to load .so file.
"""
filename = resource_loader.get_path_to_datafile(name)
library = load_library.load_op_library(filename)
return library
MPI_LIB = _load_library('mpi_lib' + get_ext_suffix())
_basics = _HorovodBasics(__file__, 'mpi_lib')
# import basic methods
init = _basics.init
shutdown = _basics.shutdown
is_initialized = _basics.is_initialized
start_timeline = _basics.start_timeline
stop_timeline = _basics.stop_timeline
size = _basics.size
local_size = _basics.local_size
rank = _basics.rank
local_rank = _basics.local_rank
mpi_threads_supported = _basics.mpi_threads_supported
mpi_enabled = _basics.mpi_enabled
"""Loads a .so file containing the specified operators.
Args:
name: The name of the .so file to load.
Raises:
NotFoundError if were not able to load .so file.
"""
filename = resource_loader.get_path_to_datafile(name)
library = load_library.load_op_library(filename)
return library
# Check possible symbol not found error from tensorflow version mismatch
try:
print("get_ext_suffix() = {}".format(str(get_ext_suffix())))
MPI_LIB = _load_library('mpi_lib' + get_ext_suffix())
except Exception as e:
check_installed_version('tensorflow', tf.__version__, e)
raise e
else:
check_installed_version('tensorflow', tf.__version__)
_basics = _HorovodBasics(__file__, 'mpi_lib')
# import basic methods
init = _basics.init
shutdown = _basics.shutdown
is_initialized = _basics.is_initialized
start_timeline = _basics.start_timeline
stop_timeline = _basics.stop_timeline
