MPI for Python

Author:: Lisandro Dalcin
Contact:: dalcinl@gmail.com
Date:: Apr 15, 2024

Introduction

Over the last years, high performance computing has become an affordable resource to many more researchers in the scientific community than ever before. The conjunction of quality open source software and commodity hardware strongly influenced the now widespread popularity of Beowulf class clusters and cluster of workstations.

Among many parallel computational models, message-passing has proven to be an effective one. This paradigm is specially suited for (but not limited to) distributed memory architectures and is used in today’s most demanding scientific and engineering application related to modeling, simulation, design, and signal processing. However, portable message-passing parallel programming used to be a nightmare in the past because of the many incompatible options developers were faced to. Fortunately, this situation definitely changed after the MPI Forum released its standard specification.

High performance computing is traditionally associated with software development using compiled languages. However, in typical applications programs, only a small part of the code is time-critical enough to require the efficiency of compiled languages. The rest of the code is generally related to memory management, error handling, input/output, and user interaction, and those are usually the most error prone and time-consuming lines of code to write and debug in the whole development process. Interpreted high-level languages can be really advantageous for this kind of tasks.

For implementing general-purpose numerical computations, MATLAB [1] is the dominant interpreted programming language. In the open source side, Octave and Scilab are well known, freely distributed software packages providing compatibility with the MATLAB language. In this work, we present MPI for Python, a new package enabling applications to exploit multiple processors using standard MPI “look and feel” in Python scripts.

What is MPI?

MPI, [mpi-using] [mpi-ref] the Message Passing Interface, is a standardized and portable message-passing system designed to function on a wide variety of parallel computers. The standard defines the syntax and semantics of library routines and allows users to write portable programs in the main scientific programming languages (Fortran, C, or C++).

Since its release, the MPI specification [mpi-std1] [mpi-std2] has become the leading standard for message-passing libraries for parallel computers. Implementations are available from vendors of high-performance computers and from well known open source projects like MPICH [mpi-mpich] and Open MPI [mpi-openmpi].

What is Python?

Python is a modern, easy to learn, powerful programming language. It has efficient high-level data structures and a simple but effective approach to object-oriented programming with dynamic typing and dynamic binding. It supports modules and packages, which encourages program modularity and code reuse. Python’s elegant syntax, together with its interpreted nature, make it an ideal language for scripting and rapid application development in many areas on most platforms.

The Python interpreter and the extensive standard library are available in source or binary form without charge for all major platforms, and can be freely distributed. It is easily extended with new functions and data types implemented in C or C++. Python is also suitable as an extension language for customizable applications.

Python is an ideal candidate for writing the higher-level parts of large-scale scientific applications [Hinsen97] and driving simulations in parallel architectures [Beazley97] like clusters of PC’s or SMP’s. Python codes are quickly developed, easily maintained, and can achieve a high degree of integration with other libraries written in compiled languages.

Overview

MPI for Python provides an object oriented approach to message passing which grounds on the standard MPI-2 C++ bindings. The interface was designed with focus in translating MPI syntax and semantics of standard MPI-2 bindings for C++ to Python. Any user of the standard C/C++ MPI bindings should be able to use this module without need of learning a new interface.

Communicating Python Objects and Array Data

The Python standard library supports different mechanisms for data persistence. Many of them rely on disk storage, but pickling and marshaling can also work with memory buffers.

The pickle modules provide user-extensible facilities to serialize general Python objects using ASCII or binary formats. The marshal module provides facilities to serialize built-in Python objects using a binary format specific to Python, but independent of machine architecture issues.

MPI for Python can communicate any built-in or user-defined Python object taking advantage of the features provided by the pickle module. These facilities will be routinely used to build binary representations of objects to communicate (at sending processes), and restoring them back (at receiving processes).

Although simple and general, the serialization approach (i.e., pickling and unpickling) previously discussed imposes important overheads in memory as well as processor usage, especially in the scenario of objects with large memory footprints being communicated. Pickling general Python objects, ranging from primitive or container built-in types to user-defined classes, necessarily requires computer resources. Processing is also needed for dispatching the appropriate serialization method (that depends on the type of the object) and doing the actual packing. Additional memory is always needed, and if its total amount is not known a priori, many reallocations can occur. Indeed, in the case of large numeric arrays, this is certainly unacceptable and precludes communication of objects occupying half or more of the available memory resources.

MPI for Python supports direct communication of any object exporting the single-segment buffer interface. This interface is a standard Python mechanism provided by some types (e.g., strings and numeric arrays), allowing access in the C side to a contiguous memory buffer (i.e., address and length) containing the relevant data. This feature, in conjunction with the capability of constructing user-defined MPI datatypes describing complicated memory layouts, enables the implementation of many algorithms involving multidimensional numeric arrays (e.g., image processing, fast Fourier transforms, finite difference schemes on structured Cartesian grids) directly in Python, with negligible overhead, and almost as fast as compiled Fortran, C, or C++ codes.

Communicators

In MPI for Python, Comm is the base class of communicators. The Intracomm and Intercomm classes are subclasses of the Comm class. The Comm.Is_inter method (and Comm.Is_intra, provided for convenience but not part of the MPI specification) is defined for communicator objects and can be used to determine the particular communicator class.

The two predefined intracommunicator instances are available: COMM_SELF and COMM_WORLD. From them, new communicators can be created as needed.

The number of processes in a communicator and the calling process rank can be respectively obtained with methods Comm.Get_size and Comm.Get_rank. The associated process group can be retrieved from a communicator by calling the Comm.Get_group method, which returns an instance of the Group class. Set operations with Group objects like like Group.Union, Group.Intersection and Group.Difference are fully supported, as well as the creation of new communicators from these groups using Comm.Create and Intracomm.Create_group.

New communicator instances can be obtained with the Comm.Clone, Comm.Dup and Comm.Split methods, as well methods Intracomm.Create_intercomm and Intercomm.Merge.

Virtual topologies (Cartcomm, Graphcomm and Distgraphcomm classes, which are specializations of the Intracomm class) are fully supported. New instances can be obtained from intracommunicator instances with factory methods Intracomm.Create_cart and Intracomm.Create_graph.

Point-to-Point Communications

Point to point communication is a fundamental capability of message passing systems. This mechanism enables the transmission of data between a pair of processes, one side sending, the other receiving.

MPI provides a set of send and receive functions allowing the communication of typed data with an associated tag. The type information enables the conversion of data representation from one architecture to another in the case of heterogeneous computing environments; additionally, it allows the representation of non-contiguous data layouts and user-defined datatypes, thus avoiding the overhead of (otherwise unavoidable) packing/unpacking operations. The tag information allows selectivity of messages at the receiving end.

Blocking Communications

MPI provides basic send and receive functions that are blocking. These functions block the caller until the data buffers involved in the communication can be safely reused by the application program.

In MPI for Python, the Comm.Send, Comm.Recv and Comm.Sendrecv methods of communicator objects provide support for blocking point-to-point communications within Intracomm and Intercomm instances. These methods can communicate memory buffers. The variants Comm.send, Comm.recv and Comm.sendrecv can communicate general Python objects.

Nonblocking Communications

On many systems, performance can be significantly increased by overlapping communication and computation. This is particularly true on systems where communication can be executed autonomously by an intelligent, dedicated communication controller.

MPI provides nonblocking send and receive functions. They allow the possible overlap of communication and computation. Non-blocking communication always come in two parts: posting functions, which begin the requested operation; and test-for-completion functions, which allow to discover whether the requested operation has completed.

In MPI for Python, the Comm.Isend and Comm.Irecv methods initiate send and receive operations, respectively. These methods return a Request instance, uniquely identifying the started operation. Its completion can be managed using the Request.Test, Request.Wait and Request.Cancel methods. The management of Request objects and associated memory buffers involved in communication requires a careful, rather low-level coordination. Users must ensure that objects exposing their memory buffers are not accessed at the Python level while they are involved in nonblocking message-passing operations.

Persistent Communications

Often a communication with the same argument list is repeatedly executed within an inner loop. In such cases, communication can be further optimized by using persistent communication, a particular case of nonblocking communication allowing the reduction of the overhead between processes and communication controllers. Furthermore , this kind of optimization can also alleviate the extra call overheads associated to interpreted, dynamic languages like Python.

In MPI for Python, the Comm.Send_init and Comm.Recv_init methods create persistent requests for a send and receive operation, respectively. These methods return an instance of the Prequest class, a subclass of the Request class. The actual communication can be effectively started using the Prequest.Start method, and its completion can be managed as previously described.

Collective Communications

Collective communications allow the transmittal of data between multiple processes of a group simultaneously. The syntax and semantics of collective functions is consistent with point-to-point communication. Collective functions communicate typed data, but messages are not paired with an associated tag; selectivity of messages is implied in the calling order. Additionally, collective functions come in blocking versions only.

The more commonly used collective communication operations are the following.

Barrier synchronization across all group members.
Global communication functions
- Broadcast data from one member to all members of a group.
- Gather data from all members to one member of a group.
- Scatter data from one member to all members of a group.
Global reduction operations such as sum, maximum, minimum, etc.

In MPI for Python, the Comm.Bcast, Comm.Scatter, Comm.Gather, Comm.Allgather, Comm.Alltoall methods provide support for collective communications of memory buffers. The lower-case variants Comm.bcast, Comm.scatter, Comm.gather, Comm.allgather and Comm.alltoall can communicate general Python objects. The vector variants (which can communicate different amounts of data to each process) Comm.Scatterv, Comm.Gatherv, Comm.Allgatherv, Comm.Alltoallv and Comm.Alltoallw are also supported, they can only communicate objects exposing memory buffers.

Global reduction operations on memory buffers are accessible through the Comm.Reduce, Comm.Reduce_scatter, Comm.Allreduce, Intracomm.Scan and Intracomm.Exscan methods. The lower-case variants Comm.reduce, Comm.allreduce, Intracomm.scan and Intracomm.exscan can communicate general Python objects; however, the actual required reduction computations are performed sequentially at some process. All the predefined (i.e., SUM, PROD, MAX, etc.) reduction operations can be applied.

Support for GPU-aware MPI

Several MPI implementations, including Open MPI and MVAPICH, support passing GPU pointers to MPI calls to avoid explicit data movement between host and device. On the Python side, support for handling GPU arrays have been implemented in many libraries related GPU computation such as CuPy, Numba, PyTorch, and PyArrow. To maximize interoperability across library boundaries, two kinds of zero-copy data exchange protocols have been defined and agreed upon: DLPack and CUDA Array Interface (CAI).

MPI for Python provides an experimental support for GPU-aware MPI. This feature requires:

mpi4py is built against a GPU-aware MPI library.
The Python GPU arrays are compliant with either of the protocols.

See the Tutorial section for further information. We note that

Whether or not a MPI call can work for GPU arrays depends on the underlying MPI implementation, not on mpi4py.
This support is currently experimental and subject to change in the future.

Dynamic Process Management

In the context of the MPI-1 specification, a parallel application is static; that is, no processes can be added to or deleted from a running application after it has been started. Fortunately, this limitation was addressed in MPI-2. The new specification added a process management model providing a basic interface between an application and external resources and process managers.

This MPI-2 extension can be really useful, especially for sequential applications built on top of parallel modules, or parallel applications with a client/server model. The MPI-2 process model provides a mechanism to create new processes and establish communication between them and the existing MPI application. It also provides mechanisms to establish communication between two existing MPI applications, even when one did not start the other.

In MPI for Python, new independent process groups can be created by calling the Intracomm.Spawn method within an intracommunicator. This call returns a new intercommunicator (i.e., an Intercomm instance) at the parent process group. The child process group can retrieve the matching intercommunicator by calling the Comm.Get_parent class method. At each side, the new intercommunicator can be used to perform point to point and collective communications between the parent and child groups of processes.

Alternatively, disjoint groups of processes can establish communication using a client/server approach. Any server application must first call the Open_port function to open a port and the Publish_name function to publish a provided service, and next call the Intracomm.Accept method. Any client applications can first find a published service by calling the Lookup_name function, which returns the port where a server can be contacted; and next call the Intracomm.Connect method. Both Intracomm.Accept and Intracomm.Connect methods return an Intercomm instance. When connection between client/server processes is no longer needed, all of them must cooperatively call the Comm.Disconnect method. Additionally, server applications should release resources by calling the Unpublish_name and Close_port functions.

One-Sided Communications

One-sided communications (also called Remote Memory Access, RMA) supplements the traditional two-sided, send/receive based MPI communication model with a one-sided, put/get based interface. One-sided communication that can take advantage of the capabilities of highly specialized network hardware. Additionally, this extension lowers latency and software overhead in applications written using a shared-memory-like paradigm.

The MPI specification revolves around the use of objects called windows; they intuitively specify regions of a process’s memory that have been made available for remote read and write operations. The published memory blocks can be accessed through three functions for put (remote send), get (remote write), and accumulate (remote update or reduction) data items. A much larger number of functions support different synchronization styles; the semantics of these synchronization operations are fairly complex.

In MPI for Python, one-sided operations are available by using instances of the Win class. New window objects are created by calling the Win.Create method at all processes within a communicator and specifying a memory buffer . When a window instance is no longer needed, the Win.Free method should be called.

The three one-sided MPI operations for remote write, read and reduction are available through calling the methods Win.Put, Win.Get, and Win.Accumulate respectively within a Win instance. These methods need an integer rank identifying the target process and an integer offset relative the base address of the remote memory block being accessed.

The one-sided operations read, write, and reduction are implicitly nonblocking, and must be synchronized by using two primary modes. Active target synchronization requires the origin process to call the Win.Start and Win.Complete methods at the origin process, and target process cooperates by calling the Win.Post and Win.Wait methods. There is also a collective variant provided by the Win.Fence method. Passive target synchronization is more lenient, only the origin process calls the Win.Lock and Win.Unlock methods. Locks are used to protect remote accesses to the locked remote window and to protect local load/store accesses to a locked local window.

Parallel Input/Output

The POSIX standard provides a model of a widely portable file system. However, the optimization needed for parallel input/output cannot be achieved with this generic interface. In order to ensure efficiency and scalability, the underlying parallel input/output system must provide a high-level interface supporting partitioning of file data among processes and a collective interface supporting complete transfers of global data structures between process memories and files. Additionally, further efficiencies can be gained via support for asynchronous input/output, strided accesses to data, and control over physical file layout on storage devices. This scenario motivated the inclusion in the MPI-2 standard of a custom interface in order to support more elaborated parallel input/output operations.

The MPI specification for parallel input/output revolves around the use objects called files. As defined by MPI, files are not just contiguous byte streams. Instead, they are regarded as ordered collections of typed data items. MPI supports sequential or random access to any integral set of these items. Furthermore, files are opened collectively by a group of processes.

The common patterns for accessing a shared file (broadcast, scatter, gather, reduction) is expressed by using user-defined datatypes. Compared to the communication patterns of point-to-point and collective communications, this approach has the advantage of added flexibility and expressiveness. Data access operations (read and write) are defined for different kinds of positioning (using explicit offsets, individual file pointers, and shared file pointers), coordination (non-collective and collective), and synchronism (blocking, nonblocking, and split collective with begin/end phases).

In MPI for Python, all MPI input/output operations are performed through instances of the File class. File handles are obtained by calling the File.Open method at all processes within a communicator and providing a file name and the intended access mode. After use, they must be closed by calling the File.Close method. Files even can be deleted by calling method File.Delete.

After creation, files are typically associated with a per-process view. The view defines the current set of data visible and accessible from an open file as an ordered set of elementary datatypes. This data layout can be set and queried with the File.Set_view and File.Get_view methods respectively.

Actual input/output operations are achieved by many methods combining read and write calls with different behavior regarding positioning, coordination, and synchronism. Summing up, MPI for Python provides the thirty (30) methods defined in MPI-2 for reading from or writing to files using explicit offsets or file pointers (individual or shared), in blocking or nonblocking and collective or noncollective versions.

Environmental Management

Initialization and Exit

Module functions Init or Init_thread and Finalize provide MPI initialization and finalization respectively. Module functions Is_initialized and Is_finalized provide the respective tests for initialization and finalization.

Note

MPI_Init() or MPI_Init_thread() is actually called when you import the MPI module from the mpi4py package, but only if MPI is not already initialized. In such case, calling Init or Init_thread from Python is expected to generate an MPI error, and in turn an exception will be raised.

Note

MPI_Finalize() is registered (by using Python C/API function Py_AtExit()) for being automatically called when Python processes exit, but only if mpi4py actually initialized MPI. Therefore, there is no need to call Finalize from Python to ensure MPI finalization.

Implementation Information

The MPI version number can be retrieved from module function Get_version. It returns a two-integer tuple (version, subversion).
The Get_processor_name function can be used to access the processor name.
The values of predefined attributes attached to the world communicator can be obtained by calling the Comm.Get_attr method within the COMM_WORLD instance.

Timers

MPI timer functionalities are available through the Wtime and Wtick functions.

Error Handling

In order to facilitate handle sharing with other Python modules interfacing MPI-based parallel libraries, the predefined MPI error handlers ERRORS_RETURN and ERRORS_ARE_FATAL can be assigned to and retrieved from communicators using methods Comm.Set_errhandler and Comm.Get_errhandler, and similarly for windows and files. New custom error handlers can be created with Comm.Create_errhandler.

When the predefined error handler ERRORS_RETURN is set, errors returned from MPI calls within Python code will raise an instance of the exception class Exception, which is a subclass of the standard Python exception RuntimeError.

Note

After import, mpi4py overrides the default MPI rules governing inheritance of error handlers. The ERRORS_RETURN error handler is set in the predefined COMM_SELF and COMM_WORLD communicators, as well as any new Comm, Win, or File instance created through mpi4py. If you ever pass such handles to C/C++/Fortran library code, it is recommended to set the ERRORS_ARE_FATAL error handler on them to ensure MPI errors do not pass silently.

Warning

Importing with from mpi4py.MPI import * will cause a name clashing with the standard Python Exception base class.

Tutorial

Warning

Under construction. Contributions very welcome!

Tip

Rolf Rabenseifner at HLRS developed a comprehensive MPI-3.1/4.0 course with slides and a large set of exercises including solutions. This material is available online for self-study. The slides and exercises show the C, Fortran, and Python (mpi4py) interfaces. For performance reasons, most Python exercises use NumPy arrays and communication routines involving buffer-like objects.

Tip

Victor Eijkhout at TACC authored the book Parallel Programming for Science and Engineering. This book is available online in PDF and HTML formats. The book covers parallel programming with MPI and OpenMP in C/C++ and Fortran, and MPI in Python using mpi4py.

MPI for Python supports convenient, pickle-based communication of generic Python object as well as fast, near C-speed, direct array data communication of buffer-provider objects (e.g., NumPy arrays).

Communication of generic Python objects

You have to use methods with all-lowercase names, like Comm.send, Comm.recv, Comm.bcast, Comm.scatter, Comm.gather . An object to be sent is passed as a parameter to the communication call, and the received object is simply the return value.

The Comm.isend and Comm.irecv methods return Request instances; completion of these methods can be managed using the Request.test and Request.wait methods.

The Comm.recv and Comm.irecv methods may be passed a buffer object that can be repeatedly used to receive messages avoiding internal memory allocation. This buffer must be sufficiently large to accommodate the transmitted messages; hence, any buffer passed to Comm.recv or Comm.irecv must be at least as long as the pickled data transmitted to the receiver.

Collective calls like Comm.scatter, Comm.gather, Comm.allgather, Comm.alltoall expect a single value or a sequence of Comm.size elements at the root or all process. They return a single value, a list of Comm.size elements, or None.

Note

MPI for Python uses the highest protocol version available in the Python runtime (see the HIGHEST_PROTOCOL constant in the pickle module). The default protocol can be changed at import time by setting the MPI4PY_PICKLE_PROTOCOL environment variable, or at runtime by assigning a different value to the PROTOCOL attribute of the pickle object within the MPI module.
Communication of buffer-like objects

You have to use method names starting with an upper-case letter, like Comm.Send, Comm.Recv, Comm.Bcast, Comm.Scatter, Comm.Gather.

In general, buffer arguments to these calls must be explicitly specified by using a 2/3-list/tuple like [data, MPI.DOUBLE], or [data, count, MPI.DOUBLE] (the former one uses the byte-size of data and the extent of the MPI datatype to define count).

For vector collectives communication operations like Comm.Scatterv and Comm.Gatherv, buffer arguments are specified as [data, count, displ, datatype], where count and displ are sequences of integral values.

Automatic MPI datatype discovery for NumPy/GPU arrays and PEP-3118 buffers is supported, but limited to basic C types (all C/C99-native signed/unsigned integral types and single/double precision real/complex floating types) and availability of matching datatypes in the underlying MPI implementation. In this case, the buffer-provider object can be passed directly as a buffer argument, the count and MPI datatype will be inferred.

If mpi4py is built against a GPU-aware MPI implementation, GPU arrays can be passed to upper-case methods as long as they have either the __dlpack__ and __dlpack_device__ methods or the __cuda_array_interface__ attribute that are compliant with the respective standard specifications. Moreover, only C-contiguous or Fortran-contiguous GPU arrays are supported. It is important to note that GPU buffers must be fully ready before any MPI routines operate on them to avoid race conditions. This can be ensured by using the synchronization API of your array library. mpi4py does not have access to any GPU-specific functionality and thus cannot perform this operation automatically for users.

Running Python scripts with MPI

Most MPI programs can be run with the command mpiexec. In practice, running Python programs looks like:

$ mpiexec -n 4 python script.py

to run the program with 4 processors.

Point-to-Point Communication

Python objects (pickle under the hood):

from mpi4py import MPI

comm = MPI.COMM_WORLD
rank = comm.Get_rank()

if rank == 0:
    data = {'a': 7, 'b': 3.14}
    comm.send(data, dest=1, tag=11)
elif rank == 1:
    data = comm.recv(source=0, tag=11)

Python objects with non-blocking communication:

from mpi4py import MPI

comm = MPI.COMM_WORLD
rank = comm.Get_rank()

if rank == 0:
    data = {'a': 7, 'b': 3.14}
    req = comm.isend(data, dest=1, tag=11)
    req.wait()
elif rank == 1:
    req = comm.irecv(source=0, tag=11)
    data = req.wait()

NumPy arrays (the fast way!):

from mpi4py import MPI
import numpy

comm = MPI.COMM_WORLD
rank = comm.Get_rank()

# passing MPI datatypes explicitly
if rank == 0:
    data = numpy.arange(1000, dtype='i')
    comm.Send([data, MPI.INT], dest=1, tag=77)
elif rank == 1:
    data = numpy.empty(1000, dtype='i')
    comm.Recv([data, MPI.INT], source=0, tag=77)

# automatic MPI datatype discovery
if rank == 0:
    data = numpy.arange(100, dtype=numpy.float64)
    comm.Send(data, dest=1, tag=13)
elif rank == 1:
    data = numpy.empty(100, dtype=numpy.float64)
    comm.Recv(data, source=0, tag=13)

Collective Communication

Broadcasting a Python dictionary:

from mpi4py import MPI

comm = MPI.COMM_WORLD
rank = comm.Get_rank()

if rank == 0:
    data = {'key1' : [7, 2.72, 2+3j],
            'key2' : ( 'abc', 'xyz')}
else:
    data = None
data = comm.bcast(data, root=0)

Scattering Python objects:

from mpi4py import MPI

comm = MPI.COMM_WORLD
size = comm.Get_size()
rank = comm.Get_rank()

if rank == 0:
    data = [(i+1)**2 for i in range(size)]
else:
    data = None
data = comm.scatter(data, root=0)
assert data == (rank+1)**2

Gathering Python objects:

from mpi4py import MPI

comm = MPI.COMM_WORLD
size = comm.Get_size()
rank = comm.Get_rank()

data = (rank+1)**2
data = comm.gather(data, root=0)
if rank == 0:
    for i in range(size):
        assert data[i] == (i+1)**2
else:
    assert data is None

Broadcasting a NumPy array:

from mpi4py import MPI
import numpy as np

comm = MPI.COMM_WORLD
rank = comm.Get_rank()

if rank == 0:
    data = np.arange(100, dtype='i')
else:
    data = np.empty(100, dtype='i')
comm.Bcast(data, root=0)
for i in range(100):
    assert data[i] == i

Scattering NumPy arrays:

from mpi4py import MPI
import numpy as np

comm = MPI.COMM_WORLD
size = comm.Get_size()
rank = comm.Get_rank()

sendbuf = None
if rank == 0:
    sendbuf = np.empty([size, 100], dtype='i')
    sendbuf.T[:,:] = range(size)
recvbuf = np.empty(100, dtype='i')
comm.Scatter(sendbuf, recvbuf, root=0)
assert np.allclose(recvbuf, rank)

Gathering NumPy arrays:

from mpi4py import MPI
import numpy as np

comm = MPI.COMM_WORLD
size = comm.Get_size()
rank = comm.Get_rank()

sendbuf = np.zeros(100, dtype='i') + rank
recvbuf = None
if rank == 0:
    recvbuf = np.empty([size, 100], dtype='i')
comm.Gather(sendbuf, recvbuf, root=0)
if rank == 0:
    for i in range(size):
        assert np.allclose(recvbuf[i,:], i)

Parallel matrix-vector product:

from mpi4py import MPI
import numpy

def matvec(comm, A, x):
    m = A.shape[0] # local rows
    p = comm.Get_size()
    xg = numpy.zeros(m*p, dtype='d')
    comm.Allgather([x,  MPI.DOUBLE],
                   [xg, MPI.DOUBLE])
    y = numpy.dot(A, xg)
    return y

Input/Output (MPI-IO)

Collective I/O with NumPy arrays:

from mpi4py import MPI
import numpy as np

amode = MPI.MODE_WRONLY|MPI.MODE_CREATE
comm = MPI.COMM_WORLD
fh = MPI.File.Open(comm, "./datafile.contig", amode)

buffer = np.empty(10, dtype=np.int)
buffer[:] = comm.Get_rank()

offset = comm.Get_rank()*buffer.nbytes
fh.Write_at_all(offset, buffer)

fh.Close()

Non-contiguous Collective I/O with NumPy arrays and datatypes:

from mpi4py import MPI
import numpy as np

comm = MPI.COMM_WORLD
rank = comm.Get_rank()
size = comm.Get_size()

amode = MPI.MODE_WRONLY|MPI.MODE_CREATE
fh = MPI.File.Open(comm, "./datafile.noncontig", amode)

item_count = 10

buffer = np.empty(item_count, dtype='i')
buffer[:] = rank

filetype = MPI.INT.Create_vector(item_count, 1, size)
filetype.Commit()

displacement = MPI.INT.Get_size()*rank
fh.Set_view(displacement, filetype=filetype)

fh.Write_all(buffer)
filetype.Free()
fh.Close()

Dynamic Process Management

Compute Pi - Master (or parent, or client) side:

#!/usr/bin/env python
from mpi4py import MPI
import numpy
import sys

comm = MPI.COMM_SELF.Spawn(sys.executable,
                           args=['cpi.py'],
                           maxprocs=5)

N = numpy.array(100, 'i')
comm.Bcast([N, MPI.INT], root=MPI.ROOT)
PI = numpy.array(0.0, 'd')
comm.Reduce(None, [PI, MPI.DOUBLE],
            op=MPI.SUM, root=MPI.ROOT)
print(PI)

comm.Disconnect()

Compute Pi - Worker (or child, or server) side:

#!/usr/bin/env python
from mpi4py import MPI
import numpy

comm = MPI.Comm.Get_parent()
size = comm.Get_size()
rank = comm.Get_rank()

N = numpy.array(0, dtype='i')
comm.Bcast([N, MPI.INT], root=0)
h = 1.0 / N; s = 0.0
for i in range(rank, N, size):
    x = h * (i + 0.5)
    s += 4.0 / (1.0 + x**2)
PI = numpy.array(s * h, dtype='d')
comm.Reduce([PI, MPI.DOUBLE], None,
            op=MPI.SUM, root=0)

comm.Disconnect()

GPU-aware MPI + Python GPU arrays

Reduce-to-all CuPy arrays:

from mpi4py import MPI
import cupy as cp

comm = MPI.COMM_WORLD
size = comm.Get_size()
rank = comm.Get_rank()

sendbuf = cp.arange(10, dtype='i')
recvbuf = cp.empty_like(sendbuf)
cp.cuda.get_current_stream().synchronize()
comm.Allreduce(sendbuf, recvbuf)

assert cp.allclose(recvbuf, sendbuf*size)

One-Sided Communication (RMA)

Read from (write to) the entire RMA window:

import numpy as np
from mpi4py import MPI
from mpi4py.util import dtlib

comm = MPI.COMM_WORLD
rank = comm.Get_rank()

datatype = MPI.FLOAT
np_dtype = dtlib.to_numpy_dtype(datatype)
itemsize = datatype.Get_size()

N = 10
win_size = N * itemsize if rank == 0 else 0
win = MPI.Win.Allocate(win_size, comm=comm)

buf = np.empty(N, dtype=np_dtype)
if rank == 0:
    buf.fill(42)
    win.Lock(rank=0)
    win.Put(buf, target_rank=0)
    win.Unlock(rank=0)
    comm.Barrier()
else:
    comm.Barrier()
    win.Lock(rank=0)
    win.Get(buf, target_rank=0)
    win.Unlock(rank=0)
    assert np.all(buf == 42)

Accessing a part of the RMA window using the target argument, which is defined as (offset, count, datatype):

import numpy as np
from mpi4py import MPI
from mpi4py.util import dtlib

comm = MPI.COMM_WORLD
rank = comm.Get_rank()

datatype = MPI.FLOAT
np_dtype = dtlib.to_numpy_dtype(datatype)
itemsize = datatype.Get_size()

N = comm.Get_size() + 1
win_size = N * itemsize if rank == 0 else 0
win = MPI.Win.Allocate(
    size=win_size,
    disp_unit=itemsize,
    comm=comm,
)
if rank == 0:
    mem = np.frombuffer(win, dtype=np_dtype)
    mem[:] = np.arange(len(mem), dtype=np_dtype)
comm.Barrier()

buf = np.zeros(3, dtype=np_dtype)
target = (rank, 2, datatype)
win.Lock(rank=0)
win.Get(buf, target_rank=0, target=target)
win.Unlock(rank=0)
assert np.all(buf == [rank, rank+1, 0])

Wrapping with SWIG

C source:

/* file: helloworld.c */
void sayhello(MPI_Comm comm)
{
  int size, rank;
  MPI_Comm_size(comm, &size);
  MPI_Comm_rank(comm, &rank);
  printf("Hello, World! "
         "I am process %d of %d.\n",
         rank, size);
}

SWIG interface file:

// file: helloworld.i
%module helloworld
%{
#include <mpi.h>
#include "helloworld.c"
}%

%include mpi4py/mpi4py.i
%mpi4py_typemap(Comm, MPI_Comm);
void sayhello(MPI_Comm comm);

Try it in the Python prompt:

>>> from mpi4py import MPI
>>> import helloworld
>>> helloworld.sayhello(MPI.COMM_WORLD)
Hello, World! I am process 0 of 1.

Wrapping with F2Py

Fortran 90 source:

! file: helloworld.f90
subroutine sayhello(comm)
  use mpi
  implicit none
  integer :: comm, rank, size, ierr
  call MPI_Comm_size(comm, size, ierr)
  call MPI_Comm_rank(comm, rank, ierr)
  print *, 'Hello, World! I am process ',rank,' of ',size,'.'
end subroutine sayhello

Compiling example using f2py

$ f2py -c --f90exec=mpif90 helloworld.f90 -m helloworld

Try it in the Python prompt:

>>> from mpi4py import MPI
>>> import helloworld
>>> fcomm = MPI.COMM_WORLD.py2f()
>>> helloworld.sayhello(fcomm)
Hello, World! I am process 0 of 1.

mpi4py

The MPI for Python package.

The Message Passing Interface (MPI) is a standardized and portable message-passing system designed to function on a wide variety of parallel computers. The MPI standard defines the syntax and semantics of library routines and allows users to write portable programs in the main scientific programming languages (Fortran, C, or C++). Since its release, the MPI specification has become the leading standard for message-passing libraries for parallel computers.

MPI for Python provides MPI bindings for the Python programming language, allowing any Python program to exploit multiple processors. This package build on the MPI specification and provides an object oriented interface which closely follows MPI-2 C++ bindings.

Runtime configuration options

mpi4py.rc: This object has attributes exposing runtime configuration options that become effective at import time of the MPI module.

Attributes Summary

`initialize`	Automatic MPI initialization at import
`threads`	Request initialization with thread support
`thread_level`	Level of thread support to request
`finalize`	Automatic MPI finalization at exit
`fast_reduce`	Use tree-based reductions for objects
`recv_mprobe`	Use matched probes to receive objects
`irecv_bufsz`	Default buffer size in bytes for `irecv()`
`errors`	Error handling policy

Attributes Documentation

mpi4py.rc.initialize

Automatic MPI initialization at import.

Type:: bool
Default:: True

See also

MPI4PY_RC_INITIALIZE

mpi4py.rc.threads

Request initialization with thread support.

Type:: bool
Default:: True

See also

MPI4PY_RC_THREADS

mpi4py.rc.thread_level

Level of thread support to request.

Type:: str
Default:: "multiple"
Choices:: "multiple", "serialized", "funneled", "single"

See also

MPI4PY_RC_THREAD_LEVEL

mpi4py.rc.finalize

Automatic MPI finalization at exit.

Type:: None or bool
Default:: None

See also

MPI4PY_RC_FINALIZE

mpi4py.rc.fast_reduce

Use tree-based reductions for objects.

Type:: bool
Default:: True

See also

MPI4PY_RC_FAST_REDUCE

mpi4py.rc.recv_mprobe

Use matched probes to receive objects.

Type:: bool
Default:: True

See also

MPI4PY_RC_RECV_MPROBE

mpi4py.rc.irecv_bufsz

Default buffer size in bytes for irecv().

Type:: int
Default:: 32768

See also

MPI4PY_RC_IRECV_BUFSZ

New in version 4.0.0.

mpi4py.rc.errors

Error handling policy.

Type:: str
Default:: "exception"
Choices:: "exception", "default", "abort", "fatal"

See also

MPI4PY_RC_ERRORS

Example

MPI for Python features automatic initialization and finalization of the MPI execution environment. By using the mpi4py.rc object, MPI initialization and finalization can be handled programmatically:

import mpi4py
mpi4py.rc.initialize = False  # do not initialize MPI automatically
mpi4py.rc.finalize = False    # do not finalize MPI automatically

from mpi4py import MPI # import the 'MPI' module

MPI.Init()      # manual initialization of the MPI environment
...             # your finest code here ...
MPI.Finalize()  # manual finalization of the MPI environment

Environment variables

The following environment variables override the corresponding attributes of the mpi4py.rc and MPI.pickle objects at import time of the MPI module.

Note

For variables of boolean type, accepted values are 0 and 1 (interpreted as False and True, respectively), and strings specifying a YAML boolean value (case-insensitive).

MPI4PY_RC_INITIALIZE

Type:: bool
Default:: True

Whether to automatically initialize MPI at import time of the mpi4py.MPI module.

See also

mpi4py.rc.initialize

New in version 4.0.0.

MPI4PY_RC_FINALIZE

Type:: None | bool
Default:: None
Choices:: None, True, False

Whether to automatically finalize MPI at exit time of the Python process.

See also

mpi4py.rc.finalize

New in version 4.0.0.

MPI4PY_RC_THREADS

Type:: bool
Default:: True

Whether to initialize MPI with thread support.

See also

mpi4py.rc.threads

New in version 3.1.0.

MPI4PY_RC_THREAD_LEVEL

Default:: "multiple"
Choices:: "single", "funneled", "serialized", "multiple"

The level of required thread support.

See also

mpi4py.rc.thread_level

New in version 3.1.0.

MPI4PY_RC_FAST_REDUCE

Type:: bool
Default:: True

Whether to use tree-based reductions for objects.

See also

mpi4py.rc.fast_reduce

New in version 3.1.0.

MPI4PY_RC_RECV_MPROBE

Type:: bool
Default:: True

Whether to use matched probes to receive objects.

See also

mpi4py.rc.recv_mprobe

MPI4PY_RC_IRECV_BUFSZ

Type:: bool
Default:: True

Default buffer size in bytes for irecv().

See also

mpi4py.rc.irecv_bufsz

New in version 4.0.0.

MPI4PY_RC_ERRORS

Default:: "exception"
Choices:: "exception", "default", "abort", "fatal"

Controls default MPI error handling policy.

See also

mpi4py.rc.errors

New in version 3.1.0.

MPI4PY_PICKLE_PROTOCOL

Type:: int
Default:: pickle.HIGHEST_PROTOCOL

Controls the default pickle protocol to use when communicating Python objects.

See also

PROTOCOL attribute of the MPI.pickle object within the MPI module.

New in version 3.1.0.

MPI4PY_PICKLE_THRESHOLD

Type:: int
Default:: 262144

Controls the default buffer size threshold for switching from in-band to out-of-band buffer handling when using pickle protocol version 5 or higher.

See also

THRESHOLD attribute of the MPI.pickle object within the MPI module.

New in version 3.1.2.

Miscellaneous functions

mpi4py.profile(name, *, path=None)

Support for the MPI profiling interface.

Parameters:

name (str) – Name of the profiler library to load.
path (sequence of str, optional) – Additional paths to search for the profiler.

Return type:

None

mpi4py.get_include()

Return the directory in the package that contains header files.

Extension modules that need to compile against mpi4py should use this function to locate the appropriate include directory. Using Python distutils (or perhaps NumPy distutils):

import mpi4py
Extension('extension_name', ...
          include_dirs=[..., mpi4py.get_include()])

Return type:: str

mpi4py.get_config()

Return a dictionary with information about MPI.

Changed in version 4.0.0: By default, this function returns an empty dictionary. However, downstream packagers and distributors may alter such behavior. To that end, MPI information must be provided under an mpi section within a UTF-8 encoded INI-style configuration file mpi.cfg located at the top-level package directory. The configuration file is read and parsed using the configparser module.

Return type:: dict[str, str]

mpi4py.MPI

Classes

Ancillary

`Datatype`	Datatype object.
`Status`	Status object.
`Request`	Request handler.
`Prequest`	Persistent request handler.
`Grequest`	Generalized request handler.
`Op`	Reduction operation.
`Group`	Group of processes.
`Info`	Info object.

Communication

`Comm`	Communication context.
`Intracomm`	Intracommunicator.
`Topocomm`	Topology intracommunicator.
`Cartcomm`	Cartesian topology intracommunicator.
`Graphcomm`	General graph topology intracommunicator.
`Distgraphcomm`	Distributed graph topology intracommunicator.
`Intercomm`	Intercommunicator.
`Message`	Matched message.

One-sided operations

Win

Remote memory access context.

Input/Output

File

File I/O context.

Error handling

`Errhandler`	Error handler.
`Exception`	Exception class.

Auxiliary

`Pickle`	Pickle/unpickle Python objects.
`memory`	alias of `buffer`

Functions

Version inquiry

`Get_version`()	Obtain the version number of the MPI standard.
`Get_library_version`()	Obtain the version string of the MPI library.

Initialization and finalization

`Init`()	Initialize the MPI execution environment.
`Init_thread`([required])	Initialize the MPI execution environment.
`Finalize`()	Terminate the MPI execution environment.
`Is_initialized`()	Indicate whether `Init` has been called.
`Is_finalized`()	Indicate whether `Finalize` has completed.
`Query_thread`()	Return the level of thread support provided by the MPI library.
`Is_thread_main`()	Indicate whether this thread called `Init` or `Init_thread`.

Memory allocation

`Alloc_mem`(size[, info])	Allocate memory for message passing and remote memory access.
`Free_mem`(mem)	Free memory allocated with `Alloc_mem`.

Address manipulation

`Get_address`(location)	Get the address of a location in memory.
`Aint_add`(base, disp)	Return the sum of base address and displacement.
`Aint_diff`(addr1, addr2)	Return the difference between absolute addresses.

Timer

`Wtick`()	Return the resolution of `Wtime`.
`Wtime`()	Return an elapsed time on the calling processor.

Error handling

`Get_error_class`(errorcode)	Convert an error code into an error class.
`Get_error_string`(errorcode)	Return the error string for a given error class or error code.
`Add_error_class`()	Add an error class to the known error classes.
`Add_error_code`(errorclass)	Add an error code to an error class.
`Add_error_string`(errorcode, string)	Associate an error string with an error class or error code.

Dynamic process management

`Open_port`([info])	Return an address used to connect group of processes.
`Close_port`(port_name)	Close a port.
`Publish_name`(service_name, port_name[, info])	Publish a service name.
`Unpublish_name`(service_name, port_name[, info])	Unpublish a service name.
`Lookup_name`(service_name[, info])	Lookup a port name given a service name.

Miscellanea

`Attach_buffer`(buf)	Attach a user-provided buffer for sending in buffered mode.
`Detach_buffer`()	Remove an existing attached buffer.
`Compute_dims`(nnodes, dims)	Return a balanced distribution of processes per coordinate direction.
`Get_processor_name`()	Obtain the name of the calling processor.
`Register_datarep`(datarep, read_fn, write_fn, ...)	Register user-defined data representations.
`Pcontrol`(level)	Control profiling.

Utilities

get_vendor()

Information about the underlying MPI implementation.

Attributes

`UNDEFINED`	Constant `UNDEFINED` of type `int`
`ANY_SOURCE`	Constant `ANY_SOURCE` of type `int`
`ANY_TAG`	Constant `ANY_TAG` of type `int`
`PROC_NULL`	Constant `PROC_NULL` of type `int`
`ROOT`	Constant `ROOT` of type `int`
`BOTTOM`	Constant `BOTTOM` of type `BottomType`
`IN_PLACE`	Constant `IN_PLACE` of type `InPlaceType`
`KEYVAL_INVALID`	Constant `KEYVAL_INVALID` of type `int`
`TAG_UB`	Constant `TAG_UB` of type `int`
`IO`	Constant `IO` of type `int`
`WTIME_IS_GLOBAL`	Constant `WTIME_IS_GLOBAL` of type `int`
`UNIVERSE_SIZE`	Constant `UNIVERSE_SIZE` of type `int`
`APPNUM`	Constant `APPNUM` of type `int`
`LASTUSEDCODE`	Constant `LASTUSEDCODE` of type `int`
`WIN_BASE`	Constant `WIN_BASE` of type `int`
`WIN_SIZE`	Constant `WIN_SIZE` of type `int`
`WIN_DISP_UNIT`	Constant `WIN_DISP_UNIT` of type `int`
`WIN_CREATE_FLAVOR`	Constant `WIN_CREATE_FLAVOR` of type `int`
`WIN_FLAVOR`	Constant `WIN_FLAVOR` of type `int`
`WIN_MODEL`	Constant `WIN_MODEL` of type `int`
`SUCCESS`	Constant `SUCCESS` of type `int`
`ERR_LASTCODE`	Constant `ERR_LASTCODE` of type `int`
`ERR_COMM`	Constant `ERR_COMM` of type `int`
`ERR_GROUP`	Constant `ERR_GROUP` of type `int`
`ERR_TYPE`	Constant `ERR_TYPE` of type `int`
`ERR_REQUEST`	Constant `ERR_REQUEST` of type `int`
`ERR_OP`	Constant `ERR_OP` of type `int`
`ERR_BUFFER`	Constant `ERR_BUFFER` of type `int`
`ERR_COUNT`	Constant `ERR_COUNT` of type `int`
`ERR_TAG`	Constant `ERR_TAG` of type `int`
`ERR_RANK`	Constant `ERR_RANK` of type `int`
`ERR_ROOT`	Constant `ERR_ROOT` of type `int`
`ERR_TRUNCATE`	Constant `ERR_TRUNCATE` of type `int`
`ERR_IN_STATUS`	Constant `ERR_IN_STATUS` of type `int`
`ERR_PENDING`	Constant `ERR_PENDING` of type `int`
`ERR_TOPOLOGY`	Constant `ERR_TOPOLOGY` of type `int`
`ERR_DIMS`	Constant `ERR_DIMS` of type `int`
`ERR_ARG`	Constant `ERR_ARG` of type `int`
`ERR_OTHER`	Constant `ERR_OTHER` of type `int`
`ERR_UNKNOWN`	Constant `ERR_UNKNOWN` of type `int`
`ERR_INTERN`	Constant `ERR_INTERN` of type `int`
`ERR_INFO`	Constant `ERR_INFO` of type `int`
`ERR_FILE`	Constant `ERR_FILE` of type `int`
`ERR_WIN`	Constant `ERR_WIN` of type `int`
`ERR_KEYVAL`	Constant `ERR_KEYVAL` of type `int`
`ERR_INFO_KEY`	Constant `ERR_INFO_KEY` of type `int`
`ERR_INFO_VALUE`	Constant `ERR_INFO_VALUE` of type `int`
`ERR_INFO_NOKEY`	Constant `ERR_INFO_NOKEY` of type `int`
`ERR_ACCESS`	Constant `ERR_ACCESS` of type `int`
`ERR_AMODE`	Constant `ERR_AMODE` of type `int`
`ERR_BAD_FILE`	Constant `ERR_BAD_FILE` of type `int`
`ERR_FILE_EXISTS`	Constant `ERR_FILE_EXISTS` of type `int`
`ERR_FILE_IN_USE`	Constant `ERR_FILE_IN_USE` of type `int`
`ERR_NO_SPACE`	Constant `ERR_NO_SPACE` of type `int`
`ERR_NO_SUCH_FILE`	Constant `ERR_NO_SUCH_FILE` of type `int`
`ERR_IO`	Constant `ERR_IO` of type `int`
`ERR_READ_ONLY`	Constant `ERR_READ_ONLY` of type `int`
`ERR_CONVERSION`	Constant `ERR_CONVERSION` of type `int`
`ERR_DUP_DATAREP`	Constant `ERR_DUP_DATAREP` of type `int`
`ERR_UNSUPPORTED_DATAREP`	Constant `ERR_UNSUPPORTED_DATAREP` of type `int`
`ERR_UNSUPPORTED_OPERATION`	Constant `ERR_UNSUPPORTED_OPERATION` of type `int`
`ERR_NAME`	Constant `ERR_NAME` of type `int`
`ERR_NO_MEM`	Constant `ERR_NO_MEM` of type `int`
`ERR_NOT_SAME`	Constant `ERR_NOT_SAME` of type `int`
`ERR_PORT`	Constant `ERR_PORT` of type `int`
`ERR_QUOTA`	Constant `ERR_QUOTA` of type `int`
`ERR_SERVICE`	Constant `ERR_SERVICE` of type `int`
`ERR_SPAWN`	Constant `ERR_SPAWN` of type `int`
`ERR_BASE`	Constant `ERR_BASE` of type `int`
`ERR_SIZE`	Constant `ERR_SIZE` of type `int`
`ERR_DISP`	Constant `ERR_DISP` of type `int`
`ERR_ASSERT`	Constant `ERR_ASSERT` of type `int`
`ERR_LOCKTYPE`	Constant `ERR_LOCKTYPE` of type `int`
`ERR_RMA_CONFLICT`	Constant `ERR_RMA_CONFLICT` of type `int`
`ERR_RMA_SYNC`	Constant `ERR_RMA_SYNC` of type `int`
`ERR_RMA_RANGE`	Constant `ERR_RMA_RANGE` of type `int`
`ERR_RMA_ATTACH`	Constant `ERR_RMA_ATTACH` of type `int`
`ERR_RMA_SHARED`	Constant `ERR_RMA_SHARED` of type `int`
`ERR_RMA_FLAVOR`	Constant `ERR_RMA_FLAVOR` of type `int`
`ORDER_C`	Constant `ORDER_C` of type `int`
`ORDER_F`	Constant `ORDER_F` of type `int`
`ORDER_FORTRAN`	Constant `ORDER_FORTRAN` of type `int`
`TYPECLASS_INTEGER`	Constant `TYPECLASS_INTEGER` of type `int`
`TYPECLASS_REAL`	Constant `TYPECLASS_REAL` of type `int`
`TYPECLASS_COMPLEX`	Constant `TYPECLASS_COMPLEX` of type `int`
`DISTRIBUTE_NONE`	Constant `DISTRIBUTE_NONE` of type `int`
`DISTRIBUTE_BLOCK`	Constant `DISTRIBUTE_BLOCK` of type `int`
`DISTRIBUTE_CYCLIC`	Constant `DISTRIBUTE_CYCLIC` of type `int`
`DISTRIBUTE_DFLT_DARG`	Constant `DISTRIBUTE_DFLT_DARG` of type `int`
`COMBINER_NAMED`	Constant `COMBINER_NAMED` of type `int`
`COMBINER_DUP`	Constant `COMBINER_DUP` of type `int`
`COMBINER_CONTIGUOUS`	Constant `COMBINER_CONTIGUOUS` of type `int`
`COMBINER_VECTOR`	Constant `COMBINER_VECTOR` of type `int`
`COMBINER_HVECTOR`	Constant `COMBINER_HVECTOR` of type `int`
`COMBINER_INDEXED`	Constant `COMBINER_INDEXED` of type `int`
`COMBINER_HINDEXED`	Constant `COMBINER_HINDEXED` of type `int`
`COMBINER_INDEXED_BLOCK`	Constant `COMBINER_INDEXED_BLOCK` of type `int`
`COMBINER_HINDEXED_BLOCK`	Constant `COMBINER_HINDEXED_BLOCK` of type `int`
`COMBINER_STRUCT`	Constant `COMBINER_STRUCT` of type `int`
`COMBINER_SUBARRAY`	Constant `COMBINER_SUBARRAY` of type `int`
`COMBINER_DARRAY`	Constant `COMBINER_DARRAY` of type `int`
`COMBINER_RESIZED`	Constant `COMBINER_RESIZED` of type `int`
`COMBINER_F90_REAL`	Constant `COMBINER_F90_REAL` of type `int`
`COMBINER_F90_COMPLEX`	Constant `COMBINER_F90_COMPLEX` of type `int`
`COMBINER_F90_INTEGER`	Constant `COMBINER_F90_INTEGER` of type `int`
`IDENT`	Constant `IDENT` of type `int`
`CONGRUENT`	Constant `CONGRUENT` of type `int`
`SIMILAR`	Constant `SIMILAR` of type `int`
`UNEQUAL`	Constant `UNEQUAL` of type `int`
`CART`	Constant `CART` of type `int`
`GRAPH`	Constant `GRAPH` of type `int`
`DIST_GRAPH`	Constant `DIST_GRAPH` of type `int`
`UNWEIGHTED`	Constant `UNWEIGHTED` of type `int`
`WEIGHTS_EMPTY`	Constant `WEIGHTS_EMPTY` of type `int`
`COMM_TYPE_SHARED`	Constant `COMM_TYPE_SHARED` of type `int`
`BSEND_OVERHEAD`	Constant `BSEND_OVERHEAD` of type `int`
`WIN_FLAVOR_CREATE`	Constant `WIN_FLAVOR_CREATE` of type `int`
`WIN_FLAVOR_ALLOCATE`	Constant `WIN_FLAVOR_ALLOCATE` of type `int`
`WIN_FLAVOR_DYNAMIC`	Constant `WIN_FLAVOR_DYNAMIC` of type `int`
`WIN_FLAVOR_SHARED`	Constant `WIN_FLAVOR_SHARED` of type `int`
`WIN_SEPARATE`	Constant `WIN_SEPARATE` of type `int`
`WIN_UNIFIED`	Constant `WIN_UNIFIED` of type `int`
`MODE_NOCHECK`	Constant `MODE_NOCHECK` of type `int`
`MODE_NOSTORE`	Constant `MODE_NOSTORE` of type `int`
`MODE_NOPUT`	Constant `MODE_NOPUT` of type `int`
`MODE_NOPRECEDE`	Constant `MODE_NOPRECEDE` of type `int`
`MODE_NOSUCCEED`	Constant `MODE_NOSUCCEED` of type `int`
`LOCK_EXCLUSIVE`	Constant `LOCK_EXCLUSIVE` of type `int`
`LOCK_SHARED`	Constant `LOCK_SHARED` of type `int`
`MODE_RDONLY`	Constant `MODE_RDONLY` of type `int`
`MODE_WRONLY`	Constant `MODE_WRONLY` of type `int`
`MODE_RDWR`	Constant `MODE_RDWR` of type `int`
`MODE_CREATE`	Constant `MODE_CREATE` of type `int`
`MODE_EXCL`	Constant `MODE_EXCL` of type `int`
`MODE_DELETE_ON_CLOSE`	Constant `MODE_DELETE_ON_CLOSE` of type `int`
`MODE_UNIQUE_OPEN`	Constant `MODE_UNIQUE_OPEN` of type `int`
`MODE_SEQUENTIAL`	Constant `MODE_SEQUENTIAL` of type `int`
`MODE_APPEND`	Constant `MODE_APPEND` of type `int`
`SEEK_SET`	Constant `SEEK_SET` of type `int`
`SEEK_CUR`	Constant `SEEK_CUR` of type `int`
`SEEK_END`	Constant `SEEK_END` of type `int`
`DISPLACEMENT_CURRENT`	Constant `DISPLACEMENT_CURRENT` of type `int`
`DISP_CUR`	Constant `DISP_CUR` of type `int`
`THREAD_SINGLE`	Constant `THREAD_SINGLE` of type `int`
`THREAD_FUNNELED`	Constant `THREAD_FUNNELED` of type `int`
`THREAD_SERIALIZED`	Constant `THREAD_SERIALIZED` of type `int`
`THREAD_MULTIPLE`	Constant `THREAD_MULTIPLE` of type `int`
`VERSION`	Constant `VERSION` of type `int`
`SUBVERSION`	Constant `SUBVERSION` of type `int`
`MAX_PROCESSOR_NAME`	Constant `MAX_PROCESSOR_NAME` of type `int`
`MAX_ERROR_STRING`	Constant `MAX_ERROR_STRING` of type `int`
`MAX_PORT_NAME`	Constant `MAX_PORT_NAME` of type `int`
`MAX_INFO_KEY`	Constant `MAX_INFO_KEY` of type `int`
`MAX_INFO_VAL`	Constant `MAX_INFO_VAL` of type `int`
`MAX_OBJECT_NAME`	Constant `MAX_OBJECT_NAME` of type `int`
`MAX_DATAREP_STRING`	Constant `MAX_DATAREP_STRING` of type `int`
`MAX_LIBRARY_VERSION_STRING`	Constant `MAX_LIBRARY_VERSION_STRING` of type `int`
`DATATYPE_NULL`	Object `DATATYPE_NULL` of type `Datatype`
`PACKED`	Object `PACKED` of type `Datatype`
`BYTE`	Object `BYTE` of type `Datatype`
`AINT`	Object `AINT` of type `Datatype`
`OFFSET`	Object `OFFSET` of type `Datatype`
`COUNT`	Object `COUNT` of type `Datatype`
`CHAR`	Object `CHAR` of type `Datatype`
`WCHAR`	Object `WCHAR` of type `Datatype`
`SIGNED_CHAR`	Object `SIGNED_CHAR` of type `Datatype`
`SHORT`	Object `SHORT` of type `Datatype`
`INT`	Object `INT` of type `Datatype`
`LONG`	Object `LONG` of type `Datatype`
`LONG_LONG`	Object `LONG_LONG` of type `Datatype`
`UNSIGNED_CHAR`	Object `UNSIGNED_CHAR` of type `Datatype`
`UNSIGNED_SHORT`	Object `UNSIGNED_SHORT` of type `Datatype`
`UNSIGNED`	Object `UNSIGNED` of type `Datatype`
`UNSIGNED_LONG`	Object `UNSIGNED_LONG` of type `Datatype`
`UNSIGNED_LONG_LONG`	Object `UNSIGNED_LONG_LONG` of type `Datatype`
`FLOAT`	Object `FLOAT` of type `Datatype`
`DOUBLE`	Object `DOUBLE` of type `Datatype`
`LONG_DOUBLE`	Object `LONG_DOUBLE` of type `Datatype`
`C_BOOL`	Object `C_BOOL` of type `Datatype`
`INT8_T`	Object `INT8_T` of type `Datatype`
`INT16_T`	Object `INT16_T` of type `Datatype`
`INT32_T`	Object `INT32_T` of type `Datatype`
`INT64_T`	Object `INT64_T` of type `Datatype`
`UINT8_T`	Object `UINT8_T` of type `Datatype`
`UINT16_T`	Object `UINT16_T` of type `Datatype`
`UINT32_T`	Object `UINT32_T` of type `Datatype`
`UINT64_T`	Object `UINT64_T` of type `Datatype`
`C_COMPLEX`	Object `C_COMPLEX` of type `Datatype`
`C_FLOAT_COMPLEX`	Object `C_FLOAT_COMPLEX` of type `Datatype`
`C_DOUBLE_COMPLEX`	Object `C_DOUBLE_COMPLEX` of type `Datatype`
`C_LONG_DOUBLE_COMPLEX`	Object `C_LONG_DOUBLE_COMPLEX` of type `Datatype`
`CXX_BOOL`	Object `CXX_BOOL` of type `Datatype`
`CXX_FLOAT_COMPLEX`	Object `CXX_FLOAT_COMPLEX` of type `Datatype`
`CXX_DOUBLE_COMPLEX`	Object `CXX_DOUBLE_COMPLEX` of type `Datatype`
`CXX_LONG_DOUBLE_COMPLEX`	Object `CXX_LONG_DOUBLE_COMPLEX` of type `Datatype`
`SHORT_INT`	Object `SHORT_INT` of type `Datatype`
`INT_INT`	Object `INT_INT` of type `Datatype`
`TWOINT`	Object `TWOINT` of type `Datatype`
`LONG_INT`	Object `LONG_INT` of type `Datatype`
`FLOAT_INT`	Object `FLOAT_INT` of type `Datatype`
`DOUBLE_INT`	Object `DOUBLE_INT` of type `Datatype`
`LONG_DOUBLE_INT`	Object `LONG_DOUBLE_INT` of type `Datatype`
`CHARACTER`	Object `CHARACTER` of type `Datatype`
`LOGICAL`	Object `LOGICAL` of type `Datatype`
`INTEGER`	Object `INTEGER` of type `Datatype`
`REAL`	Object `REAL` of type `Datatype`
`DOUBLE_PRECISION`	Object `DOUBLE_PRECISION` of type `Datatype`
`COMPLEX`	Object `COMPLEX` of type `Datatype`
`DOUBLE_COMPLEX`	Object `DOUBLE_COMPLEX` of type `Datatype`
`LOGICAL1`	Object `LOGICAL1` of type `Datatype`
`LOGICAL2`	Object `LOGICAL2` of type `Datatype`
`LOGICAL4`	Object `LOGICAL4` of type `Datatype`
`LOGICAL8`	Object `LOGICAL8` of type `Datatype`
`INTEGER1`	Object `INTEGER1` of type `Datatype`
`INTEGER2`	Object `INTEGER2` of type `Datatype`
`INTEGER4`	Object `INTEGER4` of type `Datatype`
`INTEGER8`	Object `INTEGER8` of type `Datatype`
`INTEGER16`	Object `INTEGER16` of type `Datatype`
`REAL2`	Object `REAL2` of type `Datatype`
`REAL4`	Object `REAL4` of type `Datatype`
`REAL8`	Object `REAL8` of type `Datatype`
`REAL16`	Object `REAL16` of type `Datatype`
`COMPLEX4`	Object `COMPLEX4` of type `Datatype`
`COMPLEX8`	Object `COMPLEX8` of type `Datatype`
`COMPLEX16`	Object `COMPLEX16` of type `Datatype`
`COMPLEX32`	Object `COMPLEX32` of type `Datatype`
`UNSIGNED_INT`	Object `UNSIGNED_INT` of type `Datatype`
`SIGNED_SHORT`	Object `SIGNED_SHORT` of type `Datatype`
`SIGNED_INT`	Object `SIGNED_INT` of type `Datatype`
`SIGNED_LONG`	Object `SIGNED_LONG` of type `Datatype`
`SIGNED_LONG_LONG`	Object `SIGNED_LONG_LONG` of type `Datatype`
`BOOL`	Object `BOOL` of type `Datatype`
`SINT8_T`	Object `SINT8_T` of type `Datatype`
`SINT16_T`	Object `SINT16_T` of type `Datatype`
`SINT32_T`	Object `SINT32_T` of type `Datatype`
`SINT64_T`	Object `SINT64_T` of type `Datatype`
`F_BOOL`	Object `F_BOOL` of type `Datatype`
`F_INT`	Object `F_INT` of type `Datatype`
`F_FLOAT`	Object `F_FLOAT` of type `Datatype`
`F_DOUBLE`	Object `F_DOUBLE` of type `Datatype`
`F_COMPLEX`	Object `F_COMPLEX` of type `Datatype`
`F_FLOAT_COMPLEX`	Object `F_FLOAT_COMPLEX` of type `Datatype`
`F_DOUBLE_COMPLEX`	Object `F_DOUBLE_COMPLEX` of type `Datatype`
`REQUEST_NULL`	Object `REQUEST_NULL` of type `Request`
`MESSAGE_NULL`	Object `MESSAGE_NULL` of type `Message`
`MESSAGE_NO_PROC`	Object `MESSAGE_NO_PROC` of type `Message`
`OP_NULL`	Object `OP_NULL` of type `Op`
`MAX`	Object `MAX` of type `Op`
`MIN`	Object `MIN` of type `Op`
`SUM`	Object `SUM` of type `Op`
`PROD`	Object `PROD` of type `Op`
`LAND`	Object `LAND` of type `Op`
`BAND`	Object `BAND` of type `Op`
`LOR`	Object `LOR` of type `Op`
`BOR`	Object `BOR` of type `Op`
`LXOR`	Object `LXOR` of type `Op`
`BXOR`	Object `BXOR` of type `Op`
`MAXLOC`	Object `MAXLOC` of type `Op`
`MINLOC`	Object `MINLOC` of type `Op`
`REPLACE`	Object `REPLACE` of type `Op`
`NO_OP`	Object `NO_OP` of type `Op`
`GROUP_NULL`	Object `GROUP_NULL` of type `Group`
`GROUP_EMPTY`	Object `GROUP_EMPTY` of type `Group`
`INFO_NULL`	Object `INFO_NULL` of type `Info`
`INFO_ENV`	Object `INFO_ENV` of type `Info`
`ERRHANDLER_NULL`	Object `ERRHANDLER_NULL` of type `Errhandler`
`ERRORS_RETURN`	Object `ERRORS_RETURN` of type `Errhandler`
`ERRORS_ARE_FATAL`	Object `ERRORS_ARE_FATAL` of type `Errhandler`
`COMM_NULL`	Object `COMM_NULL` of type `Comm`
`COMM_SELF`	Object `COMM_SELF` of type `Intracomm`
`COMM_WORLD`	Object `COMM_WORLD` of type `Intracomm`
`WIN_NULL`	Object `WIN_NULL` of type `Win`
`FILE_NULL`	Object `FILE_NULL` of type `File`
`pickle`	Object `pickle` of type `Pickle`

mpi4py.typing

New in version 4.0.0.

This module provides type aliases used to add type hints to the various functions and methods within the MPI module.

See also

Module typing: Documentation of the typing standard module.

Types Summary

`SupportsBuffer`	Python buffer protocol.
`SupportsDLPack`	DLPack data interchange protocol.
`SupportsCAI`	CUDA Array Interface (CAI) protocol.
`Buffer`	Buffer-like object.
`Bottom`	Start of the address range.
`InPlace`	In-place buffer argument.
`Aint`	Address-sized integral type.
`Count`	Integral type for counts.
`Displ`	Integral type for displacements.
`Offset`	Integral type for offsets.
`TypeSpec`	Datatype specification.
`BufSpec`	Buffer specification.
`BufSpecB`	Buffer specification (block).
`BufSpecV`	Buffer specification (vector).
`BufSpecW`	Buffer specification (generalized).
`TargetSpec`	Target specification.

Types Documentation

mpi4py.typing.SupportsBuffer = <class 'mpi4py.typing.SupportsBuffer'>

Python buffer protocol.

See also

Buffer Protocol

alias of mpi4py.typing.SupportsBuffer

mpi4py.typing.SupportsDLPack = <class 'mpi4py.typing.SupportsDLPack'>

DLPack data interchange protocol.

See also

Python Specification for DLPack

alias of mpi4py.typing.SupportsDLPack

mpi4py.typing.SupportsCAI = <class 'mpi4py.typing.SupportsCAI'>

CUDA Array Interface (CAI) protocol.

See also

CUDA Array Interface (Version 3)

alias of mpi4py.typing.SupportsCAI

mpi4py.typing.Buffer

Buffer-like object.

alias of Union[SupportsBuffer, SupportsDLPack, SupportsCAI]

mpi4py.typing.Bottom

Start of the address range.

alias of Optional[BottomType]

mpi4py.typing.InPlace

In-place buffer argument.

alias of Optional[InPlaceType]

mpi4py.typing.Aint = <class 'numbers.Integral'>

Address-sized integral type.

alias of numbers.Integral

mpi4py.typing.Count = <class 'numbers.Integral'>

Integral type for counts.

alias of numbers.Integral

mpi4py.typing.Displ = <class 'numbers.Integral'>

Integral type for displacements.

alias of numbers.Integral

mpi4py.typing.Offset = <class 'numbers.Integral'>

Integral type for offsets.

alias of numbers.Integral

mpi4py.typing.TypeSpec

Datatype specification.

alias of Union[Datatype, str]

mpi4py.typing.BufSpec

Buffer specification.

Buffer
Tuple[Buffer, Count]
Tuple[Buffer, TypeSpec]
Tuple[Buffer, Count, TypeSpec]
Tuple[Bottom, Count, Datatype]

alias of Union[SupportsBuffer, SupportsDLPack, SupportsCAI, Tuple[Union[SupportsBuffer, SupportsDLPack, SupportsCAI], Integral], Tuple[Union[SupportsBuffer, SupportsDLPack, SupportsCAI], Union[Datatype, str]], Tuple[Union[SupportsBuffer, SupportsDLPack, SupportsCAI], Integral, Union[Datatype, str]], Tuple[Optional[BottomType], Integral, Datatype], List[Any]]

mpi4py.typing.BufSpecB

Buffer specification (block).

Buffer
Tuple[Buffer, Count]
Tuple[Buffer, TypeSpec]
Tuple[Buffer, Count, TypeSpec]

alias of Union[SupportsBuffer, SupportsDLPack, SupportsCAI, Tuple[Union[SupportsBuffer, SupportsDLPack, SupportsCAI], Integral], Tuple[Union[SupportsBuffer, SupportsDLPack, SupportsCAI], Union[Datatype, str]], Tuple[Union[SupportsBuffer, SupportsDLPack, SupportsCAI], Integral, Union[Datatype, str]], List[Any]]

mpi4py.typing.BufSpecV

Buffer specification (vector).

Buffer
Tuple[Buffer, Sequence[Count]]
Tuple[Buffer, Tuple[Sequence[Count], Sequence[Displ]]]
Tuple[Buffer, TypeSpec]
Tuple[Buffer, Sequence[Count], TypeSpec]
Tuple[Buffer, Tuple[Sequence[Count], Sequence[Displ]], TypeSpec]
Tuple[Buffer, Sequence[Count], Sequence[Displ], TypeSpec]
Tuple[Bottom, Tuple[Sequence[Count], Sequence[Displ]], Datatype]
Tuple[Bottom, Sequence[Count], Sequence[Displ], Datatype]

mpi4py.typing.BufSpecW

Buffer specification (generalized).

Tuple[Buffer, Sequence[Datatype]]
Tuple[Buffer, Tuple[Sequence[Count], Sequence[Displ]], Sequence[Datatype]]
Tuple[Buffer, Sequence[Count], Sequence[Displ], Sequence[Datatype]]
Tuple[Bottom, Tuple[Sequence[Count], Sequence[Displ]], Sequence[Datatype]]
Tuple[Bottom, Sequence[Count], Sequence[Displ], Sequence[Datatype]]

alias of Union[Tuple[Union[SupportsBuffer, SupportsDLPack, SupportsCAI], Sequence[Datatype]], Tuple[Union[SupportsBuffer, SupportsDLPack, SupportsCAI], Tuple[Sequence[Integral], Sequence[Integral]], Sequence[Datatype]], Tuple[Union[SupportsBuffer, SupportsDLPack, SupportsCAI], Sequence[Integral], Sequence[Integral], Sequence[Datatype]], Tuple[Optional[BottomType], Tuple[Sequence[Integral], Sequence[Integral]], Sequence[Datatype]], Tuple[Optional[BottomType], Sequence[Integral], Sequence[Integral], Sequence[Datatype]], List[Any]]

mpi4py.typing.TargetSpec

Target specification.

Displ
Tuple[()]
Tuple[Displ]
Tuple[Displ, Count]
Tuple[Displ, Count, Datatype]

alias of Union[Integral, Tuple, Tuple[Integral], Tuple[Integral, Integral], Tuple[Integral, Integral, Union[Datatype, str]], List[Any]]

mpi4py.futures

New in version 3.0.0.

This package provides a high-level interface for asynchronously executing callables on a pool of worker processes using MPI for inter-process communication.

The mpi4py.futures package is based on concurrent.futures from the Python standard library. More precisely, mpi4py.futures provides the MPIPoolExecutor class as a concrete implementation of the abstract class Executor. The submit() interface schedules a callable to be executed asynchronously and returns a Future object representing the execution of the callable. Future instances can be queried for the call result or exception. Sets of Future instances can be passed to the wait() and as_completed() functions.

See also

Module concurrent.futures: Documentation of the concurrent.futures standard module.

MPIPoolExecutor

The MPIPoolExecutor class uses a pool of MPI processes to execute calls asynchronously. By performing computations in separate processes, it allows to side-step the global interpreter lock but also means that only picklable objects can be executed and returned. The __main__ module must be importable by worker processes, thus MPIPoolExecutor instances may not work in the interactive interpreter.

MPIPoolExecutor takes advantage of the dynamic process management features introduced in the MPI-2 standard. In particular, the MPI.Intracomm.Spawn method of MPI.COMM_SELF is used in the master (or parent) process to spawn new worker (or child) processes running a Python interpreter. The master process uses a separate thread (one for each MPIPoolExecutor instance) to communicate back and forth with the workers. The worker processes serve the execution of tasks in the main (and only) thread until they are signaled for completion.

Note

The worker processes must import the main script in order to unpickle any callable defined in the __main__ module and submitted from the master process. Furthermore, the callables may need access to other global variables. At the worker processes, mpi4py.futures executes the main script code (using the runpy module) under the __worker__ namespace to define the __main__ module. The __main__ and __worker__ modules are added to sys.modules (both at the master and worker processes) to ensure proper pickling and unpickling.

Warning

During the initial import phase at the workers, the main script cannot create and use new MPIPoolExecutor instances. Otherwise, each worker would attempt to spawn a new pool of workers, leading to infinite recursion. mpi4py.futures detects such recursive attempts to spawn new workers and aborts the MPI execution environment. As the main script code is run under the __worker__ namespace, the easiest way to avoid spawn recursion is using the idiom if __name__ == '__main__': ... in the main script.

class mpi4py.futures.MPIPoolExecutor(max_workers=None, initializer=None, initargs=(), **kwargs)

An Executor subclass that executes calls asynchronously using a pool of at most max_workers processes. If max_workers is None or not given, its value is determined from the MPI4PY_FUTURES_MAX_WORKERS environment variable if set, or the MPI universe size if set, otherwise a single worker process is spawned. If max_workers is lower than or equal to 0, then a ValueError will be raised.

initializer is an optional callable that is called at the start of each worker process before executing any tasks; initargs is a tuple of arguments passed to the initializer. If initializer raises an exception, all pending tasks and any attempt to submit new tasks to the pool will raise a BrokenExecutor exception.

Other parameters:

python_exe: Path to the Python interpreter executable used to spawn worker processes, otherwise sys.executable is used.
python_args: list or iterable with additional command line flags to pass to the Python executable. Command line flags determined from inspection of sys.flags, sys.warnoptions and sys._xoptions in are passed unconditionally.
mpi_info: dict or iterable yielding (key, value) pairs. These (key, value) pairs are passed (through an MPI.Info object) to the MPI.Intracomm.Spawn call used to spawn worker processes. This mechanism allows telling the MPI runtime system where and how to start the processes. Check the documentation of the backend MPI implementation about the set of keys it interprets and the corresponding format for values.
globals: dict or iterable yielding (name, value) pairs to initialize the main module namespace in worker processes.
main: If set to False, do not import the __main__ module in worker processes. Setting main to False prevents worker processes from accessing definitions in the parent __main__ namespace.
path: list or iterable with paths to append to sys.path in worker processes to extend the module search path.
wdir: Path to set the current working directory in worker processes using os.chdir(). The initial working directory is set by the MPI implementation. Quality MPI implementations should honor a wdir info key passed through mpi_info, although such feature is not mandatory.
env: dict or iterable yielding (name, value) pairs with environment variables to update os.environ in worker processes. The initial environment is set by the MPI implementation. MPI implementations may allow setting the initial environment through mpi_info, however such feature is not required nor recommended by the MPI standard.
use_pkl5: If set to True, use pickle5 with out-of-band buffers for interprocess communication. If use_pkl5 is set to None or not given, its value is determined from the MPI4PY_FUTURES_USE_PKL5 environment variable. Using pickle5 with out-of-band buffers may benefit applications dealing with large buffer-like objects like NumPy arrays. See mpi4py.util.pkl5 for additional information.
backoff: float value specifying the maximum number of seconds a worker thread or process suspends execution with time.sleep() while idle-waiting. If not set, its value is determined from the MPI4PY_FUTURES_BACKOFF environment variable if set, otherwise the default value of 0.001 seconds is used. Lower values will reduce latency and increase execution throughput for very short-lived tasks, albeit at the expense of spinning CPU cores and increased energy consumption.

submit(func, *args, **kwargs)

Schedule the callable, func, to be executed as func(*args, **kwargs) and returns a Future object representing the execution of the callable.

executor = MPIPoolExecutor(max_workers=1)
future = executor.submit(pow, 321, 1234)
print(future.result())

map(func, *iterables, timeout=None, chunksize=1, **kwargs)

Equivalent to map(func, *iterables) except func is executed asynchronously and several calls to func may be made concurrently, out-of-order, in separate processes. The returned iterator raises a TimeoutError if __next__() is called and the result isn’t available after timeout seconds from the original call to map(). timeout can be an int or a float. If timeout is not specified or None, there is no limit to the wait time. If a call raises an exception, then that exception will be raised when its value is retrieved from the iterator. This method chops iterables into a number of chunks which it submits to the pool as separate tasks. The (approximate) size of these chunks can be specified by setting chunksize to a positive integer. For very long iterables, using a large value for chunksize can significantly improve performance compared to the default size of one. By default, the returned iterator yields results in-order, waiting for successive tasks to complete . This behavior can be changed by passing the keyword argument unordered as True, then the result iterator will yield a result as soon as any of the tasks complete.

executor = MPIPoolExecutor(max_workers=3)
for result in executor.map(pow, [2]*32, range(32)):
    print(result)

starmap(func, iterable, timeout=None, chunksize=1, **kwargs)

Equivalent to itertools.starmap(func, iterable). Used instead of map() when argument parameters are already grouped in tuples from a single iterable (the data has been “pre-zipped”). map(func, *iterable) is equivalent to starmap(func, zip(*iterable)).

executor = MPIPoolExecutor(max_workers=3)
iterable = ((2, n) for n in range(32))
for result in executor.starmap(pow, iterable):
    print(result)

shutdown(wait=True, cancel_futures=False)

Signal the executor that it should free any resources that it is using when the currently pending futures are done executing. Calls to submit() and map() made after shutdown() will raise RuntimeError.

If wait is True then this method will not return until all the pending futures are done executing and the resources associated with the executor have been freed. If wait is False then this method will return immediately and the resources associated with the executor will be freed when all pending futures are done executing. Regardless of the value of wait, the entire Python program will not exit until all pending futures are done executing.

If cancel_futures is True, this method will cancel all pending futures that the executor has not started running. Any futures that are completed or running won’t be cancelled, regardless of the value of cancel_futures.

You can avoid having to call this method explicitly if you use the with statement, which will shutdown the executor instance (waiting as if shutdown() were called with wait set to True).

import time
with MPIPoolExecutor(max_workers=1) as executor:
    future = executor.submit(time.sleep, 2)
assert future.done()

bootup(wait=True): Signal the executor that it should allocate eagerly any required resources (in particular, MPI worker processes). If wait is True, then bootup() will not return until the executor resources are ready to process submissions. Resources are automatically allocated in the first call to submit(), thus calling bootup() explicitly is seldom needed.

num_workers: Number or worker processes in the pool.

MPI4PY_FUTURES_MAX_WORKERS: If the max_workers parameter to MPIPoolExecutor is None or not given, the MPI4PY_FUTURES_MAX_WORKERS environment variable provides a fallback value for the maximum number of MPI worker processes to spawn.

New in version 3.1.0.

MPI4PY_FUTURES_USE_PKL5: If the use_pkl5 keyword argument to MPIPoolExecutor is None or not given, the MPI4PY_FUTURES_USE_PKL5 environment variable provides a fallback value for whether the executor should use pickle5 with out-of-band buffers for interprocess communication. Accepted values are 0 and 1 (interpreted as False and True, respectively), and strings specifying a YAML boolean value (case-insensitive). Using pickle5 with out-of-band buffers may benefit applications dealing with large buffer-like objects like NumPy arrays. See mpi4py.util.pkl5 for additional information.

New in version 4.0.0.

MPI4PY_FUTURES_BACKOFF: If the backoff keyword argument to MPIPoolExecutor is not given, the MPI4PY_FUTURES_BACKOFF environment variable can be set to a float value specifying the maximum number of seconds a worker thread or process suspends execution with time.sleep() while idle-waiting. If not set, the default backoff value is 0.001 seconds. Lower values will reduce latency and increase execution throughput for very short-lived tasks, albeit at the expense of spinning CPU cores and increased energy consumption.

New in version 4.0.0.

Note

As the master process uses a separate thread to perform MPI communication with the workers, the backend MPI implementation should provide support for MPI.THREAD_MULTIPLE. However, some popular MPI implementations do not support yet concurrent MPI calls from multiple threads. Additionally, users may decide to initialize MPI with a lower level of thread support. If the level of thread support in the backend MPI is less than MPI.THREAD_MULTIPLE, mpi4py.futures will use a global lock to serialize MPI calls. If the level of thread support is less than MPI.THREAD_SERIALIZED, mpi4py.futures will emit a RuntimeWarning.

Warning

If the level of thread support in the backend MPI is less than MPI.THREAD_SERIALIZED (i.e, it is either MPI.THREAD_SINGLE or MPI.THREAD_FUNNELED), in theory mpi4py.futures cannot be used. Rather than raising an exception, mpi4py.futures emits a warning and takes a “cross-fingers” attitude to continue execution in the hope that serializing MPI calls with a global lock will actually work.

MPICommExecutor

Legacy MPI-1 implementations (as well as some vendor MPI-2 implementations) do not support the dynamic process management features introduced in the MPI-2 standard. Additionally, job schedulers and batch systems in supercomputing facilities may pose additional complications to applications using the MPI_Comm_spawn() routine.

With these issues in mind, mpi4py.futures supports an additional, more traditional, SPMD-like usage pattern requiring MPI-1 calls only. Python applications are started the usual way, e.g., using the mpiexec command. Python code should make a collective call to the MPICommExecutor context manager to partition the set of MPI processes within a MPI communicator in one master processes and many workers processes. The master process gets access to an MPIPoolExecutor instance to submit tasks. Meanwhile, the worker process follow a different execution path and team-up to execute the tasks submitted from the master.

Besides alleviating the lack of dynamic process management features in legacy MPI-1 or partial MPI-2 implementations, the MPICommExecutor context manager may be useful in classic MPI-based Python applications willing to take advantage of the simple, task-based, master/worker approach available in the mpi4py.futures package.

class mpi4py.futures.MPICommExecutor(comm=None, root=0)

Context manager for MPIPoolExecutor. This context manager splits a MPI (intra)communicator comm (defaults to MPI.COMM_WORLD if not provided or None) in two disjoint sets: a single master process (with rank root in comm) and the remaining worker processes. These sets are then connected through an intercommunicator. The target of the with statement is assigned either an MPIPoolExecutor instance (at the master) or None (at the workers).

from mpi4py import MPI
from mpi4py.futures import MPICommExecutor

with MPICommExecutor(MPI.COMM_WORLD, root=0) as executor:
    if executor is not None:
       future = executor.submit(abs, -42)
       assert future.result() == 42
       answer = set(executor.map(abs, [-42, 42]))
       assert answer == {42}

Warning

If MPICommExecutor is passed a communicator of size one (e.g., MPI.COMM_SELF), then the executor instance assigned to the target of the with statement will execute all submitted tasks in a single worker thread, thus ensuring that task execution still progress asynchronously. However, the GIL will prevent the main and worker threads from running concurrently in multicore processors. Moreover, the thread context switching may harm noticeably the performance of CPU-bound tasks. In case of I/O-bound tasks, the GIL is not usually an issue, however, as a single worker thread is used, it progress one task at a time. We advice against using MPICommExecutor with communicators of size one and suggest refactoring your code to use instead a ThreadPoolExecutor.

Command line

Recalling the issues related to the lack of support for dynamic process management features in MPI implementations, mpi4py.futures supports an alternative usage pattern where Python code (either from scripts, modules, or zip files) is run under command line control of the mpi4py.futures package by passing -m mpi4py.futures to the python executable. The mpi4py.futures invocation should be passed a pyfile path to a script (or a zipfile/directory containing a __main__.py file). Additionally, mpi4py.futures accepts -m mod to execute a module named mod, -c cmd to execute a command string cmd, or even - to read commands from standard input (sys.stdin). Summarizing, mpi4py.futures can be invoked in the following ways:

$ mpiexec -n numprocs python -m mpi4py.futures pyfile [arg] ...
$ mpiexec -n numprocs python -m mpi4py.futures -m mod [arg] ...
$ mpiexec -n numprocs python -m mpi4py.futures -c cmd [arg] ...
$ mpiexec -n numprocs python -m mpi4py.futures - [arg] ...

Before starting the main script execution, mpi4py.futures splits MPI.COMM_WORLD in one master (the process with rank 0 in MPI.COMM_WORLD) and numprocs - 1 workers and connects them through an MPI intercommunicator. Afterwards, the master process proceeds with the execution of the user script code, which eventually creates MPIPoolExecutor instances to submit tasks. Meanwhile, the worker processes follow a different execution path to serve the master. Upon successful termination of the main script at the master, the entire MPI execution environment exists gracefully. In case of any unhandled exception in the main script, the master process calls MPI.COMM_WORLD.Abort(1) to prevent deadlocks and force termination of entire MPI execution environment.

Warning

Running scripts under command line control of mpi4py.futures is quite similar to executing a single-process application that spawn additional workers as required. However, there is a very important difference users should be aware of. All MPIPoolExecutor instances created at the master will share the pool of workers. Tasks submitted at the master from many different executors will be scheduled for execution in random order as soon as a worker is idle. Any executor can easily starve all the workers (e.g., by calling MPIPoolExecutor.map() with long iterables). If that ever happens, submissions from other executors will not be serviced until free workers are available.

See also

Command line: Documentation on Python command line interface.

Parallel tasks

The mpi4py.futures package favors an embarrassingly parallel execution model involving a series of sequential tasks independent of each other and executed asynchronously. Albeit unnatural, MPIPoolExecutor can still be used for handling workloads involving parallel tasks, where worker processes communicate and coordinate each other via MPI.

mpi4py.futures.get_comm_workers(): Access an intracommunicator grouping MPI worker processes.

Executing parallel tasks with mpi4py.futures requires following some rules, cf. highlighted lines in example cpi.py :

Use MPIPoolExecutor.num_workers to determine the number of worker processes in the executor and submit exactly one callable per worker process using the MPIPoolExecutor.submit() method.
The submitted callable must use get_comm_workers() to access an intracommunicator grouping MPI worker processes. Afterwards, it is highly recommended calling the Barrier() method on the communicator. The barrier synchronization ensures that every worker process is executing the submitted callable exactly once. Afterwards, the parallel task can safely perform any kind of point-to-point or collective operation using the returned communicator.
The Future instances returned by MPIPoolExecutor.submit() should be collected in a sequence. Use wait() with the sequence of Future instances to ensure logical completion of the parallel task.

Utilities

The mpi4py.futures package provides additional utilities for handling Future instances.

mpi4py.futures.collect(fs)

Gather a collection of futures in a new future.

Parameters:: fs – Collection of futures.
Returns:: New future producing as result a list with results from fs.

mpi4py.futures.compose(future, resulthook=None, excepthook=None)

Compose the completion of a future with result and exception handlers.

Parameters:

future – Input future instance.
resulthook – Function to be called once the input future completes with success. Once the input future finish running with success, its result value is the input argument for resulthook. The result of resulthook is set as the result of the output future. If resulthook is None, the output future is completed directly with the result of the input future.
excepthook – Function to be called once the input future completes with failure. Once the input future finish running with failure, its exception value is the input argument for excepthook. If excepthook returns an Exception instance, it is set as the exception of the output future. Otherwise, the result of excepthook is set as the result of the output future. If excepthook is None, the output future is set as failed with the exception from the input future.

Returns:

Output future instance to be completed once the input future is completed and either resulthook or excepthook finish executing.

Examples

Computing the Julia set

The following julia.py script computes the Julia set and dumps an image to disk in binary PGM format. The code starts by importing MPIPoolExecutor from the mpi4py.futures package. Next, some global constants and functions implement the computation of the Julia set. The computations are protected with the standard if __name__ == '__main__': ... idiom. The image is computed by whole scanlines submitting all these tasks at once using the map method. The result iterator yields scanlines in-order as the tasks complete. Finally, each scanline is dumped to disk.

julia.py

from mpi4py.futures import MPIPoolExecutor

x0, x1, w = -2.0, +2.0, 640*2
y0, y1, h = -1.5, +1.5, 480*2
dx = (x1 - x0) / w
dy = (y1 - y0) / h

c = complex(0, 0.65)

def julia(x, y):
    z = complex(x, y)
    n = 255
    while abs(z) < 3 and n > 1:
        z = z**2 + c
        n -= 1
    return n

def julia_line(k):
    line = bytearray(w)
    y = y1 - k * dy
    for j in range(w):
        x = x0 + j * dx
        line[j] = julia(x, y)
    return line

if __name__ == '__main__':

    with MPIPoolExecutor() as executor:
        image = executor.map(julia_line, range(h))
        with open('julia.pgm', 'wb') as f:
            f.write(b'P5 %d %d %d\n' % (w, h, 255))
            for line in image:
                f.write(line)

The recommended way to execute the script is by using the mpiexec command specifying one MPI process (master) and (optional but recommended) the desired MPI universe size, which determines the number of additional dynamically spawned processes (workers). The MPI universe size is provided either by a batch system or set by the user via command-line arguments to mpiexec or environment variables. Below we provide examples for MPICH and Open MPI implementations [1]. In all of these examples, the mpiexec command launches a single master process running the Python interpreter and executing the main script. When required, mpi4py.futures spawns the pool of 16 worker processes. The master submits tasks to the workers and waits for the results. The workers receive incoming tasks, execute them, and send back the results to the master.

When using MPICH implementation or its derivatives based on the Hydra process manager, users can set the MPI universe size via the -usize argument to mpiexec:

$ mpiexec -n 1 -usize 17 python julia.py

or, alternatively, by setting the MPIEXEC_UNIVERSE_SIZE environment variable:

$ env MPIEXEC_UNIVERSE_SIZE=17 mpiexec -n 1 python julia.py

In the Open MPI implementation, the MPI universe size can be set via the -host argument to mpiexec:

$ mpiexec -n 1 -host localhost:17 python julia.py

Another way to specify the number of workers is to use the mpi4py.futures-specific environment variable MPI4PY_FUTURES_MAX_WORKERS:

$ env MPI4PY_FUTURES_MAX_WORKERS=16 mpiexec -n 1 python julia.py

Note that in this case, the MPI universe size is ignored.

Alternatively, users may decide to execute the script in a more traditional way, that is, all the MPI processes are started at once. The user script is run under command-line control of mpi4py.futures passing the -m flag to the python executable:

$ mpiexec -n 17 python -m mpi4py.futures julia.py

As explained previously, the 17 processes are partitioned in one master and 16 workers. The master process executes the main script while the workers execute the tasks submitted by the master.

Computing Pi (parallel task)

The number $\pi$ can be approximated via numerical integration with the simple midpoint rule, that is:

\[\pi = \int_{0}^{1} \frac{4}{1+x^2} \,dx \approx \frac{1}{n} \sum_{i=1}^{n} \frac{4}{1 + \left[\frac{1}{n} \left(i-\frac{1}{2}\right) \right]^2} .\]

The following cpi.py script computes such approximations using mpi4py.futures with a parallel task involving a collective reduction operation. Highlighted lines correspond to the rules discussed in Parallel tasks.

cpi.py

import math
import sys
from mpi4py.futures import MPIPoolExecutor, wait
from mpi4py.futures import get_comm_workers


def compute_pi(n):
    # Access intracommunicator and synchronize
    comm = get_comm_workers()
    comm.Barrier()

    rank = comm.Get_rank()
    size = comm.Get_size()

    # Local computation
    h = 1.0 / n
    s = 0.0
    for i in range(rank + 1, n + 1, size):
        x = h * (i - 0.5)
        s += 4.0 / (1.0 + x**2)
    pi_partial = s * h

    # Parallel reduce-to-all
    pi = comm.allreduce(pi_partial)

    # All workers return the same value
    return pi


if __name__ == '__main__':
    n = int(sys.argv[1]) if len(sys.argv) > 1 else 256

    with MPIPoolExecutor() as executor:
        # Submit exactly one callable per worker
        P = executor.num_workers
        fs = [executor.submit(compute_pi, n) for _ in range(P)]

        # Wait for all workers to finish
        wait(fs)

        # Get result from the first future object.
        # In this particular example, due to using reduce-to-all,
        # all the other future objects hold the same result value.
        pi = fs[0].result()
        print(
            f"pi: {pi:.16f}, error: {abs(pi - math.pi):.3e}",
            f"({n:d} intervals, {P:d} workers)",
        )

To run in modern MPI-2 mode:

$ env MPI4PY_FUTURES_MAX_WORKERS=4 mpiexec -n 1 python cpi.py 128
pi: 3.1415977398528137, error: 5.086e-06 (128 intervals, 4 workers)

$ env MPI4PY_FUTURES_MAX_WORKERS=8 mpiexec -n 1 python cpi.py 512
pi: 3.1415929714812316, error: 3.179e-07 (512 intervals, 8 workers)

To run in legacy MPI-1 mode:

$ mpiexec -n 5 python -m mpi4py.futures cpi.py 128
pi: 3.1415977398528137, error: 5.086e-06 (128 intervals, 4 workers)

$ mpiexec -n 9 python -m mpi4py.futures cpi.py 512
pi: 3.1415929714812316, error: 3.179e-07 (512 intervals, 8 workers)

Citation

If mpi4py.futures been significant to a project that leads to an academic publication, please acknowledge our work by citing the following article [mpi4py-futures]:

[mpi4py-futures]

M. Rogowski, S. Aseeri, D. Keyes, and L. Dalcin, mpi4py.futures: MPI-Based Asynchronous Task Execution for Python, IEEE Transactions on Parallel and Distributed Systems, 34(2):611-622, 2023. https://doi.org/10.1109/TPDS.2022.3225481

mpi4py.util

New in version 3.1.0.

The mpi4py.util package collects miscellaneous utilities within the intersection of Python and MPI.

mpi4py.util.dtlib

New in version 3.1.0.

The mpi4py.util.dtlib module provides converter routines between NumPy and MPI datatypes.

mpi4py.util.dtlib.from_numpy_dtype(dtype)

Convert NumPy datatype to MPI datatype.

Parameters:: dtype (DTypeLike) – NumPy dtype-like object.
Return type:: Datatype

mpi4py.util.dtlib.to_numpy_dtype(datatype)

Convert MPI datatype to NumPy datatype.

Parameters:: datatype (Datatype) – MPI datatype.
Return type:: dtype[Any]

mpi4py.util.pkl5

New in version 3.1.0.

pickle protocol 5 (see PEP 574) introduced support for out-of-band buffers, allowing for more efficient handling of certain object types with large memory footprints.

MPI for Python uses the traditional in-band handling of buffers. This approach is appropriate for communicating non-buffer Python objects, or buffer-like objects with small memory footprints. For point-to-point communication, in-band buffer handling allows for the communication of a pickled stream with a single MPI message, at the expense of additional CPU and memory overhead in the pickling and unpickling steps.

The mpi4py.util.pkl5 module provides communicator wrapper classes reimplementing pickle-based point-to-point and collective communication methods using pickle protocol 5. Handling out-of-band buffers necessarily involves multiple MPI messages, thus increasing latency and hurting performance in case of small size data. However, in case of large size data, the zero-copy savings of out-of-band buffer handling more than offset the extra latency costs. Additionally, these wrapper methods overcome the infamous 2 GiB message count limit (MPI-1 to MPI-3).

Note

Support for pickle protocol 5 is available in the pickle module within the Python standard library since Python 3.8. Previous Python 3 releases can use the pickle5 backport, which is available on PyPI and can be installed with:

python -m pip install pickle5

class mpi4py.util.pkl5.Request

Request.

Custom request class for nonblocking communications.

Note

Request is not a subclass of mpi4py.MPI.Request

Free()

Free a communication request.

Return type:: None

cancel()

Cancel a communication request.

Return type:: None

get_status(status=None)

Non-destructive test for the completion of a request.

Parameters:: status (Status | None) –
Return type:: bool

test(status=None)

Test for the completion of a request.

Parameters:: status (Status | None) –
Return type:: tuple[bool, Any | None]

wait(status=None)

Wait for a request to complete.

Parameters:: status (Status | None) –
Return type:: Any

classmethod testall(requests, statuses=None)

Test for the completion of all requests.

Classmethod:

classmethod waitall(requests, statuses=None)

Wait for all requests to complete.

Classmethod:

class mpi4py.util.pkl5.Message

Message.

Custom message class for matching probes.

Note

Message is not a subclass of mpi4py.MPI.Message

recv(status=None)

Blocking receive of matched message.

Parameters:: status (Status | None) –
Return type:: Any

irecv()

Nonblocking receive of matched message.

Return type:: Request

classmethod probe(comm, source=ANY_SOURCE, tag=ANY_TAG, status=None)

Blocking test for a matched message.

Classmethod:

classmethod iprobe(comm, source=ANY_SOURCE, tag=ANY_TAG, status=None)

Nonblocking test for a matched message.

Classmethod:

class mpi4py.util.pkl5.Comm

Communicator.

Base communicator wrapper class.

send(obj, dest, tag=0)

Blocking send in standard mode.

Parameters:

obj (Any) –
dest (int) –
tag (int) –

Return type:

None

bsend(obj, dest, tag=0)

Blocking send in buffered mode.

Parameters:

obj (Any) –
dest (int) –
tag (int) –

Return type:

None

ssend(obj, dest, tag=0)

Blocking send in synchronous mode.

Parameters:

obj (Any) –
dest (int) –
tag (int) –

Return type:

None

isend(obj, dest, tag=0)

Nonblocking send in standard mode.

Parameters:

obj (Any) –
dest (int) –
tag (int) –

Return type:

Request

ibsend(obj, dest, tag=0)

Nonblocking send in buffered mode.

Parameters:

obj (Any) –
dest (int) –
tag (int) –

Return type:

Request

issend(obj, dest, tag=0)

Nonblocking send in synchronous mode.

Parameters:

obj (Any) –
dest (int) –
tag (int) –

Return type:

Request

recv(buf=None, source=ANY_SOURCE, tag=ANY_TAG, status=None)

Blocking receive.

Parameters:

buf (Buffer | None) –
source (int) –
tag (int) –
status (Status | None) –

Return type:

Any

irecv(buf=None, source=ANY_SOURCE, tag=ANY_TAG)

Nonblocking receive.

Warning

This method cannot be supported reliably and raises RuntimeError.

Parameters:

buf (Buffer | None) –
source (int) –
tag (int) –

Return type:

Request

sendrecv(sendobj, dest, sendtag=0, recvbuf=None, source=ANY_SOURCE, recvtag=ANY_TAG, status=None)

Send and receive.

Parameters:

sendobj (Any) –
dest (int) –
sendtag (int) –
recvbuf (Buffer | None) –
source (int) –
recvtag (int) –
status (Status | None) –

Return type:

Any

mprobe(source=ANY_SOURCE, tag=ANY_TAG, status=None)

Blocking test for a matched message.

Parameters:

source (int) –
tag (int) –
status (Status | None) –

Return type:

Message

improbe(source=ANY_SOURCE, tag=ANY_TAG, status=None)

Nonblocking test for a matched message.

Parameters:

source (int) –
tag (int) –
status (Status | None) –

Return type:

Message | None

bcast(obj, root=0)

Broadcast.

New in version 3.1.0.

Parameters:

obj (Any) –
root (int) –

Return type:

Any

gather(sendobj, root=0)

Gather.

New in version 4.0.0.

Parameters:

sendobj (Any) –
root (int) –

Return type:

list[Any] | None

scatter(sendobj, root=0)

Scatter.

New in version 4.0.0.

Parameters:

sendobj (Sequence[Any] | None) –
root (int) –

Return type:

Any

allgather(sendobj)

Gather to All.

New in version 4.0.0.

Parameters:: sendobj (Any) –
Return type:: list[Any]

alltoall(sendobj)

All to All Scatter/Gather.

New in version 4.0.0.

Parameters:: sendobj (Sequence[Any]) –
Return type:: list[Any]

class mpi4py.util.pkl5.Intracomm

Intracommunicator.

Intracommunicator wrapper class.

class mpi4py.util.pkl5.Intercomm

Intercommunicator.

Intercommunicator wrapper class.

Examples

test-pkl5-1.py

import numpy as np
from mpi4py import MPI
from mpi4py.util import pkl5

comm = pkl5.Intracomm(MPI.COMM_WORLD)  # comm wrapper
size = comm.Get_size()
rank = comm.Get_rank()
dst = (rank + 1) % size
src = (rank - 1) % size

sobj = np.full(1024**3, rank, dtype='i4')  # > 4 GiB
sreq = comm.isend(sobj, dst, tag=42)
robj = comm.recv (None, src, tag=42)
sreq.Free()

assert np.min(robj) == src
assert np.max(robj) == src

test-pkl5-2.py

import numpy as np
from mpi4py import MPI
from mpi4py.util import pkl5

comm = pkl5.Intracomm(MPI.COMM_WORLD)  # comm wrapper
size = comm.Get_size()
rank = comm.Get_rank()
dst = (rank + 1) % size
src = (rank - 1) % size

sobj = np.full(1024**3, rank, dtype='i4')  # > 4 GiB
sreq = comm.isend(sobj, dst, tag=42)

status = MPI.Status()
rmsg = comm.mprobe(status=status)
assert status.Get_source() == src
assert status.Get_tag() == 42
rreq = rmsg.irecv()
robj = rreq.wait()

sreq.Free()
assert np.max(robj) == src
assert np.min(robj) == src

mpi4py.util.pool

New in version 4.0.0.

See also

This module intends to be a drop-in replacement for the multiprocessing.pool interface from the Python standard library. The Pool class exposed here is implemented as a thin wrapper around MPIPoolExecutor.

Note

The mpi4py.futures package offers a higher level interface for asynchronously pushing tasks to MPI worker process, allowing for a clear separation between submitting tasks and waiting for the results.

class mpi4py.util.pool.Pool

Pool using MPI processes as workers.

__init__(processes=None, initializer=None, initargs=(), **kwargs)

Initialize a new Pool instance.

Parameters:

processes – Number of worker processes.
initializer – An callable used to initialize workers processes.
initargs – A tuple of arguments to pass to the initializer.

Note

Additional keyword arguments are passed down to the MPIPoolExecutor constructor.

Warning

The maxtasksperchild and context arguments of multiprocessing.pool.Pool are not supported. Specifying maxtasksperchild or context with a value other than None will issue a warning of category UserWarning.

apply(func, args=(), kwds={})

Call func with arguments args and keyword arguments kwds.

Equivalent to func(*args, **kwds).

apply_async(func, args=(), kwds={}, callback=None, error_callback=None): Asynchronous version of apply() returning ApplyResult.

map(func, iterable, chunksize=None)

Apply func to each element in iterable.

Equivalent to list(map(func, iterable)).

Block until all results are ready and return them in a list.

The iterable is choped into a number of chunks which are submitted as separate tasks. The (approximate) size of these chunks can be specified by setting chunksize to a positive integer.

Consider using imap() or imap_unordered() with explicit chunksize for better efficiency.

map_async(func, iterable, chunksize=None, callback=None, error_callback=None): Asynchronous version of map() returning MapResult.

imap(func, iterable, chunksize=1)

Like map() but return an iterator.

Equivalent to map(func, iterable).

imap_unordered(func, iterable, chunksize=1): Like imap() but ordering of results is arbitrary.

starmap(func, iterable, chunksize=None)

Apply func to each argument tuple in iterable.

Equivalent to list(itertools.starmap(func, iterable)).

Block until all results are ready and return them in a list.

The iterable is choped into a number of chunks which are submitted as separate tasks. The (approximate) size of these chunks can be specified by setting chunksize to a positive integer.

Consider using istarmap() or istarmap_unordered() with explicit chunksize for better efficiency.

starmap_async(func, iterable, chunksize=None, callback=None, error_callback=None): Asynchronous version of starmap() returning MapResult.

istarmap(func, iterable, chunksize=1)

Like starmap() but return an iterator.

Equivalent to itertools.starmap(func, iterable).

istarmap_unordered(func, iterable, chunksize=1): Like istarmap() but ordering of results is arbitrary.

close(): Prevent any more tasks from being submitted to the pool.

terminate(): Stop the worker processes without completing pending tasks.

join(): Wait for the worker processes to exit.

class mpi4py.util.pool.ThreadPool

Bases: Pool

Pool using threads as workers.

class mpi4py.util.pool.AsyncResult

Asynchronous result.

get(timeout=None)

Return the result when it arrives.

If timeout is not None and the result does not arrive within timeout seconds then raise TimeoutError.

If the remote call raised an exception then that exception will be reraised.

wait(timeout=None): Wait until the result is available or timeout seconds pass.

ready(): Return whether the call has completed.

successful()

Return whether the call completed without raising an exception.

If the result is not ready then raise ValueError.

class mpi4py.util.pool.ApplyResult

Bases: AsyncResult

Result type of apply_async().

class mpi4py.util.pool.MapResult

Bases: AsyncResult

Result type of map_async() and starmap_async().

mpi4py.util.sync

New in version 4.0.0.

The mpi4py.util.sync module provides parallel synchronization utilities.

Sequential execution

class mpi4py.util.sync.Sequential

Sequential execution.

Context manager for sequential execution within a group of MPI processes.

The implementation is based in MPI-1 point-to-point communication. A process with rank i waits in a blocking receive until the previous process rank i-1 finish executing and signals the next rank i with a send.

__init__(comm, tag=0)

Initialize sequential execution.

Parameters:

comm (Intracomm) – Intracommunicator context.
tag (int) – Tag for point-to-point communication.

Return type:

None

__enter__()

Enter sequential execution.

Return type:: Self

__exit__(*exc)

Exit sequential execution.

Parameters:: exc (object) –
Return type:: None

begin()

Begin sequential execution.

Return type:: None

end()

End sequential execution.

Return type:: None

Global counter

class mpi4py.util.sync.Counter

Parallel counter.

Produce consecutive values within a group of MPI processes. The counter interface is close to that of itertools.count.

The implementation is based in MPI-3 one-sided operations. A root process (typically rank 0) holds the counter, and its value is queried and incremented with an atomic RMA fetch-and-add operation.

__init__(comm, start=0, step=1, typecode='i', root=0, info=INFO_NULL)

Initialize counter object.

Parameters:

comm (Intracomm) – Intracommunicator context.
start (int) – Start value.
step (int) – Increment value.
typecode (str) – Type code as defined in the array module.
root (int) – Process rank holding the counter memory.
info (Info) – Info object for RMA context creation.

Return type:

None

__iter__()

Implement iter(self).

Return type:: Self

__next__()

Implement next(self).

Return type:: int

next(incr=None)

Return current value and increment.

Parameters:: incr (int | None) – Increment value.
Returns:: The counter value before incrementing.
Return type:: int

free()

Free counter resources.

Return type:: None

Mutual exclusion

class mpi4py.util.sync.Mutex

Parallel mutex.

Establish a critical section or mutual exclusion among MPI processes. The mutex interface is close to that of threading.Lock. However, its intended uses and specific semantics are somewhat different:

A mutex should be used within a group of MPI processes, not threads.
Once acquired, a mutex is held and owned by a process until released.
Trying to acquire a mutex already held raises RuntimeError.
Trying to release a mutex not yet held raises RuntimeError.

This mutex implementation uses the scalable and fair spinlock algorithm from [mcs-paper] and took inspiration from the MPI-3 RMA implementation of [uam-book].

[mcs-paper]

John M. Mellor-Crummey and Michael L. Scott. Algorithms for scalable synchronization on shared-memory multiprocessors. ACM Transactions on Computer Systems, 9(1):21-65, February 1991. https://doi.org/10.1145/103727.103729

[uam-book]

William Gropp, Torsten Hoefler, Rajeev Thakur, Ewing Lusk. Using Advanced MPI - Modern Features of the Message-Passing Interface. Chapter 4, Section 4.7, Pages 130-131. The MIT Press, November 2014. https://mitpress.mit.edu/9780262527637/using-advanced-mpi/

__init__(comm, info=INFO_NULL)

Initialize mutex object.

Parameters:

comm (Intracomm) – Intracommunicator context.
info (Info) – Info object for RMA context creation.

Return type:

None

__enter__()

Acquire mutex.

Return type:: Self

__exit__(*exc)

Release mutex.

Parameters:: exc (object) –
Return type:: None

acquire(blocking=True)

Acquire mutex, blocking or non-blocking.

Parameters:: blocking (bool) – If True, block until the mutex is held.
Returns:: True if mutex is held, False otherwise.
Return type:: bool

release()

Release mutex.

Return type:: None

locked()

Return whether the mutex is held.

Return type:: bool

free()

Free mutex resources.

Return type:: None

class mpi4py.util.sync.RMutex

Parallel recursive mutex.

Establish a critical section or mutual exclusion among MPI processes. The mutex interface is close to that of threading.RLock, allowing for recursive acquire and release operations. However, the mutex should be used within a group of MPI processes, not threads.

The implementation is based on a Mutex providing mutual exclusion and a counter tracking the recursion level.

__init__(comm, info=INFO_NULL)

Initialize recursive mutex object.

Parameters:

comm (Intracomm) – Intracommunicator context.
info (Info) – Info object for RMA context creation.

Return type:

None

__enter__()

Acquire mutex.

Return type:: Self

__exit__(*exc)

Release mutex.

Parameters:: exc (object) –
Return type:: None

acquire(blocking=True)

Acquire mutex, blocking or non-blocking.

Parameters:: blocking (bool) – If True, block until the mutex is held.
Returns:: True if mutex is held, False otherwise.
Return type:: bool

release()

Release mutex.

Return type:: None

count()

Return recursion count.

Return type:: int

free()

Free mutex resources.

Return type:: None

Examples

test-sync-1.py

from mpi4py import MPI
from mpi4py.util.sync import Counter, Sequential

comm = MPI.COMM_WORLD

counter = Counter(comm)
with Sequential(comm):
   value = next(counter)
counter.free()

assert comm.rank == value

test-sync-2.py

from mpi4py import MPI
from mpi4py.util.sync import Counter, Mutex

comm = MPI.COMM_WORLD

mutex = Mutex(comm)
counter = Counter(comm)
with mutex:
   value = next(counter)
counter.free()
mutex.free()

assert (
   list(range(comm.size)) ==
   sorted(comm.allgather(value))
)

mpi4py.run

New in version 3.0.0.

At import time, mpi4py initializes the MPI execution environment calling MPI_Init_thread() and installs an exit hook to automatically call MPI_Finalize() just before the Python process terminates. Additionally, mpi4py overrides the default ERRORS_ARE_FATAL error handler in favor of ERRORS_RETURN, which allows translating MPI errors in Python exceptions. These departures from standard MPI behavior may be controversial, but are quite convenient within the highly dynamic Python programming environment. Third-party code using mpi4py can just from mpi4py import MPI and perform MPI calls without the tedious initialization/finalization handling. MPI errors, once translated automatically to Python exceptions, can be dealt with the common try…except…finally clauses; unhandled MPI exceptions will print a traceback which helps in locating problems in source code.

Unfortunately, the interplay of automatic MPI finalization and unhandled exceptions may lead to deadlocks. In unattended runs, these deadlocks will drain the battery of your laptop, or burn precious allocation hours in your supercomputing facility.

Exceptions and deadlocks

Consider the following snippet of Python code. Assume this code is stored in a standard Python script file and run with mpiexec in two or more processes.

deadlock.py

from mpi4py import MPI
assert MPI.COMM_WORLD.Get_size() > 1
rank = MPI.COMM_WORLD.Get_rank()
if rank == 0:
    1/0
    MPI.COMM_WORLD.send(None, dest=1, tag=42)
elif rank == 1:
    MPI.COMM_WORLD.recv(source=0, tag=42)

Process 0 raises ZeroDivisionError exception before performing a send call to process 1. As the exception is not handled, the Python interpreter running in process 0 will proceed to exit with non-zero status. However, as mpi4py installed a finalizer hook to call MPI_Finalize() before exit, process 0 will block waiting for other processes to also enter the MPI_Finalize() call. Meanwhile, process 1 will block waiting for a message to arrive from process 0, thus never reaching to MPI_Finalize(). The whole MPI execution environment is irremediably in a deadlock state.

To alleviate this issue, mpi4py offers a simple, alternative command line execution mechanism based on using the -m flag and implemented with the runpy module. To use this features, Python code should be run passing -m mpi4py in the command line invoking the Python interpreter. In case of unhandled exceptions, the finalizer hook will call MPI_Abort() on the MPI_COMM_WORLD communicator, thus effectively aborting the MPI execution environment.

Warning

When a process is forced to abort, resources (e.g. open files) are not cleaned-up and any registered finalizers (either with the atexit module, the Python C/API function Py_AtExit(), or even the C standard library function atexit()) will not be executed. Thus, aborting execution is an extremely impolite way of ensuring process termination. However, MPI provides no other mechanism to recover from a deadlock state.

Command line

The use of -m mpi4py to execute Python code on the command line resembles that of the Python interpreter.

mpiexec -n numprocs python -m mpi4py pyfile [arg] ...
mpiexec -n numprocs python -m mpi4py -m mod [arg] ...
mpiexec -n numprocs python -m mpi4py -c cmd [arg] ...
mpiexec -n numprocs python -m mpi4py - [arg] ...

<pyfile>: Execute the Python code contained in pyfile, which must be a filesystem path referring to either a Python file, a directory containing a __main__.py file, or a zipfile containing a __main__.py file.

-m <mod>: Search sys.path for the named module mod and execute its contents.

-c <cmd>: Execute the Python code in the cmd string command.

-: Read commands from standard input (sys.stdin).

See also

Command line: Documentation on Python command line interface.

mpi4py.bench

New in version 3.0.0.

Reference

mpi4py.MPI

Message Passing Interface.

Citation

If MPI for Python been significant to a project that leads to an academic publication, please acknowledge that fact by citing the project.

M. Rogowski, S. Aseeri, D. Keyes, and L. Dalcin, mpi4py.futures: MPI-Based Asynchronous Task Execution for Python, IEEE Transactions on Parallel and Distributed Systems, 34(2):611-622, 2023. https://doi.org/10.1109/TPDS.2022.3225481
L. Dalcin and Y.-L. L. Fang, mpi4py: Status Update After 12 Years of Development, Computing in Science & Engineering, 23(4):47-54, 2021. https://doi.org/10.1109/MCSE.2021.3083216
L. Dalcin, P. Kler, R. Paz, and A. Cosimo, Parallel Distributed Computing using Python, Advances in Water Resources, 34(9):1124-1139, 2011. https://doi.org/10.1016/j.advwatres.2011.04.013
L. Dalcin, R. Paz, M. Storti, and J. D’Elia, MPI for Python: performance improvements and MPI-2 extensions, Journal of Parallel and Distributed Computing, 68(5):655-662, 2008. https://doi.org/10.1016/j.jpdc.2007.09.005
L. Dalcin, R. Paz, and M. Storti, MPI for Python, Journal of Parallel and Distributed Computing, 65(9):1108-1115, 2005. https://doi.org/10.1016/j.jpdc.2005.03.010

Installation

Build backends

mpi4py supports two different build backends: setuptools (default), scikit-build-core (CMake-based), and meson-python (Meson-based). The build backend can be selected by setting the MPI4PY_BUILD_BACKEND environment variable.

MPI4PY_BUILD_BACKEND

Choices:: "setuptools", "scikit-build-core", "meson-python"
Default:: "setuptools"

Request a build backend for building mpi4py from sources.

Using setuptools

Tip

Set the MPI4PY_BUILD_BACKEND environment variable to "setuptools" to use the setuptools build backend.

When using the default setuptools build backend, mpi4py relies on the legacy Python distutils framework to build C extension modules. The following environment variables affect the build configuration.

MPI4PY_BUILD_MPICC: The mpicc compiler wrapper command is searched for in the executable search path (PATH environment variable) and used to compile the mpi4py.MPI C extension module. Alternatively, use the MPI4PY_BUILD_MPICC environment variable to the full path or command corresponding to the MPI-aware C compiler.

MPI4PY_BUILD_MPILD: The mpicc compiler wrapper command is also used for linking the mpi4py.MPI C extension module. Alternatively, use the MPI4PY_BUILD_MPILD environment variable to specify the full path or command corresponding to the MPI-aware C linker.

MPI4PY_BUILD_MPICFG: If the MPI implementation does not provide a compiler wrapper, or it is not installed in a default system location, all relevant build information like include/library locations and library lists can be provided in an ini-style configuration file under a [mpi] section. mpi4py can then be asked to use the custom build information by setting the MPI4PY_BUILD_MPICFG environment variable to the full path of the configuration file. As an example, see the mpi.cfg file located in the top level mpi4py source directory.

MPI4PY_BUILD_CONFIGURE: Some vendor MPI implementations may not provide complete coverage of the MPI standard, or may provide partial features of newer MPI standard versions while advertising support for an older version. Setting the MPI4PY_BUILD_CONFIGURE environment variable to a non-empty string will trigger the run of exhaustive checks for the availability of all MPI constants, predefined handles, and routines.

The following environment variables are aliases for the ones described above. Having shorter names, they are convenient for occasional use in the command line. Its usage is not recommended in automation scenarios like packaging recipes, deployment scripts, and container image creation.

MPICC: Convenience alias for MPI4PY_BUILD_MPICC.

MPILD: Convenience alias for MPI4PY_BUILD_MPILD.

MPICFG: Convenience alias for MPI4PY_BUILD_MPICFG.

Using scikit-build-core

Tip

Set the MPI4PY_BUILD_BACKEND environment variable to "scikit-build-core" to use the scikit-build-core build backend.

When using the scikit-build-core build backend, mpi4py delegates all of MPI build configuration to CMake’s FindMPI module. Besides the obvious advantage of cross-platform support, this delegation to CMake may be convenient in build environments exposing vendor software stacks via intricate module systems. Note however that mpi4py will not be able to look for MPI routines available beyond the MPI standard version the MPI implementation advertises to support (via the MPI_VERSION and MPI_SUBVERSION macro constants in the mpi.h header file), any missing MPI constant or symbol will prevent a successful build.

Using meson-python

Tip

Set the MPI4PY_BUILD_BACKEND environment variable to "meson-python" to use the meson-python build backend.

When using the meson-python build backend, mpi4py delegates build tasks to the Meson build system.

Warning

mpi4py support for the meson-python build backend is experimental. For the time being, users must set the CC environment variable to the command or path corresponding to the mpicc C compiler wrapper.

Using pip

You can install the latest mpi4py release from its source distribution at PyPI using pip:

$ python -m pip install mpi4py

You can also install the in-development version with:

$ python -m pip install git+https://github.com/mpi4py/mpi4py

or:

$ python -m pip install https://github.com/mpi4py/mpi4py/tarball/master

Note

Installing mpi4py from its source distribution (available at PyPI) or Git source code repository (available at GitHub) requires a C compiler and a working MPI implementation with development headers and libraries.

Warning

pip keeps previously built wheel files on its cache for future reuse. If you want to reinstall the mpi4py package using a different or updated MPI implementation, you have to either first remove the cached wheel file with:

$ python -m pip cache remove mpi4py

or ask pip to disable the cache:

$ python -m pip install --no-cache-dir mpi4py

Using conda

The conda-forge community provides ready-to-use binary packages from an ever growing collection of software libraries built around the multi-platform conda package manager. Three MPI implementations are available on conda-forge: Open MPI (Linux and macOS), MPICH (Linux and macOS), and Microsoft MPI (Windows). You can install mpi4py and your preferred MPI implementation using the conda package manager:

to use MPICH do:

$ conda install -c conda-forge mpi4py mpich

to use Open MPI do:

$ conda install -c conda-forge mpi4py openmpi

to use Microsoft MPI do:

$ conda install -c conda-forge mpi4py msmpi

MPICH and many of its derivatives are ABI-compatible. You can provide the package specification mpich=X.Y.*=external_* (where X and Y are the major and minor version numbers) to request the conda package manager to use system-provided MPICH (or derivative) libraries. Similarly, you can provide the package specification openmpi=X.Y.*=external_* to use system-provided Open MPI libraries.

The openmpi package on conda-forge has built-in CUDA support, but it is disabled by default. To enable it, follow the instruction outlined during conda install. Additionally, UCX support is also available once the ucx package is installed.

Warning

Binary conda-forge packages are built with a focus on compatibility. The MPICH and Open MPI packages are build in a constrained environment with relatively dated OS images. Therefore, they may lack support for high-performance features like cross-memory attach (XPMEM/CMA). In production scenarios, it is recommended to use external (either custom-built or system-provided) MPI installations. See the relevant conda-forge documentation about using external MPI libraries .

Linux

On Fedora Linux systems (as well as RHEL and their derivatives using the EPEL software repository), you can install binary packages with the system package manager:

using dnf and the mpich package:

$ sudo dnf install python3-mpi4py-mpich

using dnf and the openmpi package:

$ sudo dnf install python3-mpi4py-openmpi

Please remember to load the correct MPI module for your chosen MPI implementation:

for the mpich package do:

$ module load mpi/mpich-$(arch)
$ python -c "from mpi4py import MPI"

for the openmpi package do:

$ module load mpi/openmpi-$(arch)
$ python -c "from mpi4py import MPI"

On Ubuntu Linux and Debian Linux systems, binary packages are available for installation using the system package manager:

$ sudo apt install python3-mpi4py

Note that on Ubuntu/Debian systems, the mpi4py package uses Open MPI. To use MPICH, install the libmpich-dev and python3-dev packages (and any other required development tools). Afterwards, install mpi4py from sources using pip.

macOS

macOS users can install mpi4py using the Homebrew package manager:

$ brew install mpi4py

Note that the Homebrew mpi4py package uses Open MPI. Alternatively, install the mpich package and next install mpi4py from sources using pip.

Windows

Windows users can install mpi4py from binary wheels hosted on the Python Package Index (PyPI) using pip:

$ python -m pip install mpi4py

Windows wheels require a separate, system-wide installation of the Microsoft MPI runtime package.

Development

Prerequisites

You need to have the following software properly installed in order to build MPI for Python:

Python 3.6 or above.
The Cython compiler.
A working MPI implementation like MPICH or Open MPI, preferably supporting MPI-4 and built with shared/dynamic libraries.

Note

If you want to build some MPI implementation from sources, check the instructions at Building MPI from sources in the appendix.

Note

Some MPI-1 implementations do require the actual command line arguments to be passed in MPI_Init(). In this case, you will need to use a rebuilt, MPI-enabled, Python interpreter executable. MPI for Python has some support for alleviating you from this task. Check the instructions at MPI-enabled Python interpreter in the appendix.

Optionally, consider installing the following packages:

NumPy for enabling comprehensive testing of MPI communication.
CuPy for enabling comprehensive testing with a GPU-aware MPI.
Sphinx to build documentation.

Building

MPI for Python uses setuptools-based build system that relies on the setup.py file. Some setuptools commands (e.g., build) accept additional options:

--mpi=: Lets you pass a section with MPI configuration within a special configuration file. Alternatively, you can use the MPICFG environment variable.

--mpicc=: Specify the path or name of the mpicc C compiler wrapper. Alternatively, use the MPICC environment variable.

--mpild=: Specify the full path or name for the MPI-aware C linker. Alternatively, use the MPILD environment variable. If not set, the mpicc C compiler wrapper is used for linking.

--configure: Runs exhaustive tests for checking about missing MPI types, constants, and functions. This option should be passed in order to build MPI for Python against old MPI-1, MPI-2, or MPI-3 implementations, possibly providing a subset of MPI-4.

If you use a MPI implementation providing a mpicc C compiler wrapper (e.g., MPICH or Open MPI), it will be used for compilation and linking. This is the preferred and easiest way to build MPI for Python.

If mpicc is found in the executable search path (PATH environment variable), simply run the build command:

$ python setup.py build

If mpicc is not in your search path or the compiler wrapper has a different name, you can run the build command specifying its location, either via the --mpicc command option or using the MPICC environment variable:

$ python setup.py build --mpicc=/path/to/mpicc
$ MPICC=/path/to/mpicc python setup.py build

Alternatively, you can provide all the relevant information about your MPI implementation by editing the mpi.cfg file located in the top level source directory. You can use the default section [mpi] or add a new custom section, for example [other_mpi] (see the examples provided in the mpi.cfg file as a starting point to write your own section):

[mpi]
include_dirs         = /usr/local/mpi/include
libraries            = mpi
library_dirs         = /usr/local/mpi/lib
runtime_library_dirs = /usr/local/mpi/lib

[other_mpi]
include_dirs         = /opt/mpi/include ...
libraries            = mpi ...
library_dirs         = /opt/mpi/lib ...
runtime_library_dirs = /opt/mpi/lib ...

...

and then run the build command specifying you custom configuration section:

$ python setup.py build --mpi=other_mpi
$ MPICFG=other_mpi python setup.py build

After building, the package is ready for installation in development mode:

$ python setup.py develop --user

Alternatively, you can generate a binary wheel file in the dist/ directory with:

$ python setup.py bdist_wheel

Testing

To quickly test the installation:

$ mpiexec -n 5 python -m mpi4py.bench helloworld
Hello, World! I am process 0 of 5 on localhost.
Hello, World! I am process 1 of 5 on localhost.
Hello, World! I am process 2 of 5 on localhost.
Hello, World! I am process 3 of 5 on localhost.
Hello, World! I am process 4 of 5 on localhost.

$ mpiexec -n 5 python -m mpi4py.bench ringtest -l 10 -n 1048576
time for 10 loops = 0.00361614 seconds (5 processes, 1048576 bytes)

If you installed from a git clone or the source distribution, issuing at the command line:

$ mpiexec -n 5 python demo/helloworld.py

will launch a five-process run of the Python interpreter and run the test script demo/helloworld.py from the source distribution.

You can also run all the unittest scripts:

$ mpiexec -n 5 python test/main.py

or, if you have nose unit testing framework installed:

$ mpiexec -n 5 nosetests

or, if you have py.test unit testing framework installed:

$ mpiexec -n 5 py.test

Appendix

MPI-enabled Python interpreter

Warning

These days it is no longer required to use the MPI-enabled Python interpreter in most cases, and, therefore, it is not built by default anymore because it is too difficult to reliably build a Python interpreter across different distributions. If you know that you still really need it, see below on how to use the build_exe and install_exe commands.

Some MPI-1 implementations (notably, MPICH 1) do require the actual command line arguments to be passed at the time MPI_Init() is called. In this case, you will need to use a re-built, MPI-enabled, Python interpreter binary executable. A basic implementation (targeting Python 3.9) of what is required is shown below:

#include <Python.h>
#include <mpi.h>

int main(int argc, char *argv[])
{
   int status, flag;
   MPI_Init(&argc, &argv);
   status = Py_BytesMain(argc, argv);
   MPI_Finalized(&flag);
   if (!flag) MPI_Finalize();
   return status;
}

The source code above is straightforward; compiling it should also be. However, the linking step is more tricky: special flags have to be passed to the linker depending on your platform. In order to alleviate you for such low-level details, MPI for Python provides some pure-distutils based support to build and install an MPI-enabled Python interpreter executable:

$ cd mpi4py-X.X.X
$ python setup.py build_exe [--mpi=<name>|--mpicc=/path/to/mpicc]
$ [sudo] python setup.py install_exe [--install-dir=$HOME/bin]

After the above steps you should have the MPI-enabled interpreter installed as prefix/bin/pythonX.X-mpi (or $HOME/bin/pythonX.X-mpi). Assuming that prefix/bin (or $HOME/bin) is listed on your PATH, you should be able to enter your MPI-enabled Python interactively, for example:

$ python3.9-mpi
Python 3.9.6 (default, Jul 16 2021, 00:00:00)
[GCC 11.1.1 20210531 (Red Hat 11.1.1-3)] on linux
Type "help", "copyright", "credits" or "license" for more information.
>>> import sys
>>> sys.executable
'/usr/local/bin/python3.9-mpi'
>>>

Building MPI from sources

In the list below you have some executive instructions for building some of the open-source MPI implementations out there with support for shared/dynamic libraries on POSIX environments.

MPICH

$ tar -zxf mpich-X.X.X.tar.gz
$ cd mpich-X.X.X
$ ./configure --enable-shared --prefix=/usr/local/mpich
$ make
$ make install

Open MPI

$ tar -zxf openmpi-X.X.X tar.gz
$ cd openmpi-X.X.X
$ ./configure --prefix=/usr/local/openmpi
$ make all
$ make install

MPICH 1

$ tar -zxf mpich-X.X.X.tar.gz
$ cd mpich-X.X.X
$ ./configure --enable-sharedlib --prefix=/usr/local/mpich1
$ make
$ make install

Perhaps you will need to set the LD_LIBRARY_PATH environment variable (using export, setenv or what applies to your system) pointing to the directory containing the MPI libraries . In case of getting runtime linking errors when running MPI programs, the following lines can be added to the user login shell script (.profile, .bashrc, etc.).

MPICH

MPI_DIR=/usr/local/mpich
export LD_LIBRARY_PATH=$MPI_DIR/lib:$LD_LIBRARY_PATH

Open MPI

MPI_DIR=/usr/local/openmpi
export LD_LIBRARY_PATH=$MPI_DIR/lib:$LD_LIBRARY_PATH

MPICH 1
```
MPI_DIR=/usr/local/mpich1
export LD_LIBRARY_PATH=$MPI_DIR/lib/shared:$LD_LIBRARY_PATH:
export MPICH_USE_SHLIB=yes
```
Warning

MPICH 1 support for dynamic libraries is not completely transparent. Users should set the environment variable MPICH_USE_SHLIB to yes in order to avoid link problems when using the mpicc compiler wrapper.

LICENSE

Redistribution and use in source and binary forms, with or without modification, are permitted provided that the following conditions are met:

Redistributions of source code must retain the above copyright notice, this list of conditions and the following disclaimer.
Redistributions in binary form must reproduce the above copyright notice, this list of conditions and the following disclaimer in the documentation and/or other materials provided with the distribution.

THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER AND CONTRIBUTORS “AS IS” AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.

CHANGES