MPI for Python - MPI4Py
Contents
1. else data None data comm bcast data root 0 e 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 xx2 for i in range size else data None data comm scatter data root 0 assert data rank 1 xx2 e Gathering Python objects from mpi4py import MPI comm MPIT COMM_ WORLD size comm Get_size rank comm Get_rank data rank 1 xx2 data comm gather data root 0 if rank 0 for i in range size assert data i itl 2 else assert data is None e 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 4 3 Dynamic Process Management Compute Pi e 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 e Worker or child or server side usr bin env python from mpi4py import MPI import numpy comm MPI Comm Get_parent2. size comm Get_size rank comm Get_rank N numpy array 0 dtype i comm Bcast N MPI INT root 0 hi 1 0 y Ni 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 x h dtype d comm Reduce PI MPI DOUBLE None Op MPI SUM root 0 comm Disconnect 4 4 Wrapping with SWIG e C source file helloworld c void sayhello MPI_Comm comm int size rank MPI_Comm_size comm amp size MPI_Comm_rank comm amp rank printf Hello World I am process d of d n rank size e SWIG interface file file helloworld i module helloworld 1 include lt mpi h gt include helloworld c Sinclude mpi4py mpi4py i Smpi4py_typemap Comm MPI_Comm void sayhello MPI_Comm comm e Try it in the Python prompt gt gt gt from mpi4py import MPI gt gt gt import helloworld gt gt gt helloworld sayhello MPI COMM_ WORLD Hello World I am process 0 of 1 4 5 Wrapping with F2Py e 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 e Try it in the Python prompt gt gt gt from mpi4py import MPI gt gt gt import helloworld gt gt gt fcomm MP
3. 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 2 2 Communicators In MPI for Python Comm is the base class of communicators The Intracomm and Intercomm classes are sublcasses of the Comm class The Is_inter method and Is_intra provided for convenience it is 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 meth ods Get_size and Get_rank The associated process group can be retrieved from a communicator by calling the Get_group method which returns an instance of the Group class Set operations with Group objects like like Union Intersect and Difference are fully supported as well as the creation of new communicators from these groups using Create New communicato
4. 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 3 2 Using pip or easy_install If you already have a working MPI either if you installed it from sources or by using a pre built package from your favourite GNU Linux distribution and the mpice compiler wrapper is on your search path you can use pip sudo pip install mpi4py or alternatively setuptools easy_install deprecated sudo easy_install mpi4py Note If the mpicc compiler wrapper is not on your search path or if it has a different name you can use env to pass the environment variable MP ICC providing the full path to the MPI compiler wrapper executable sudo env MPICC path to mpicc pip install mpi4py sudo env MPICC path to mpicc easy_install mpi4py 3 3 Using distutils MPI for Python uses a standard distutils based buildsystem However some distutils commands like build have additional options e mpicc let you specify a special location or name for the mpicc compiler wrapper e mpi let you pass a section with MPI configuration within a special configuration file e configure runs exhaustive tests for checking about missing MPI types constants calls Thi
5. approach Any server application must first call the Open_port function to open a port and the Publish_name func tion to publish a provided service and next call the Accept method within an Intracomm instance 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 Connect method within an Intracomm in stance Both Accept and Connect methods return an Intercomm instance When connection between client server processes is no longer needed all of them must cooperatively call the Disconnect method of the Comm class Additionally server applications should release resources by calling the Unpublish_name and Close_port functions 2 6 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 commu nication 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
6. end users would be the one matching the system version or alternatively the root directory i e SDKROOT Appropriate values for ARCHF LAGS have the form arch lt value gt where lt value gt should be chosen from the following table Intel PowerPC 32 bit 1386 ppc 64 bit x86_64 ppc64 For example assuming your Mac is running Snow Leopard on a 64 bit Intel processor and you want to override the hard wired cross development SDK in Python configuration you can build and install MPI for Python using any of the alternatives below Note that environment variables may need to be passed set both at the build and install steps because sudo may not pass environment variables to subprocesses for security reasons e Alternative 1 env MACOSX_DEPLOYMENT_TARGET 10 6 SDKROOT ARCHFLAGS arch x86_64 python setup py build options sudo env MACOSX_DEPLOYMENT_TARGET 10 6 SDKROOT ARCHFLAGS arch x86_64 python setup py install options e Alternative 2 export MACOSX_DEPLOYMENT_TARGET 10 6 export SDKROOT export ARCHFLAGS arch x86_64 python setup py build options sudo s enter interactive shell as root export MACOSX_DEPLOYMENT_TARGET 10 6 export SDKROOT export ARCHFLAGS arch x86_64 python setup py install options exit 5 3 Building MPI from sources In the list bel
7. 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 Create method at all processes within a communicator and specifying a memory buffer When a window instance is no longer needed the Free method should be called The three one sided MPI operations for remote write read and reduction are available through calling the methods Put Get and 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 Start and Complete methods at the origin process and target process cooperates by calling the Post and Wait methods There is also a collective variant provided by the Fence method Passive target synchronization is more lenient only the origin process calls the Lock and Unlock methods Locks are used to protect remote accesses to th
8. variables and class attributes This allows the compiler to generate very efficient C code from Cython code This makes Cython the ideal language for wrapping for external C libraries and for fast C modules that speed up the execution of Python code e SWIG is a software development tool that connects programs written in C and C with a variety of high level programming languages like Perl Tcl Tk Ruby and Python Issuing header files to SWIG is the simplest approach to interfacing C C libraries from a Python module 2 Design and Interface 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 2 1 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 slower written in pure Python and cPick1le faster written in C modules provide user extensible facilities to serialize generic 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 indepen
9. 18 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 4 1 Point to Point Communication e Python objects pickle under the hood from mpi4py import MPI comm MPI COMM_WORLD rank comm Get_rank if rank 0 data ar Tr Torr 35243 comm send data dest 1 tag 11 elif rank 1 data comm recv source 0 tag 11 e NumPy arrays the fast way from mpi4py import MPI import numpy comm MPI COMM_WORLD rank comm Get_rank pass explicit MPI datatypes if rank data numpy arange 1000 dtype i comm Send data MPI INT dest 1 tag 77 elif rank data numpy empty 1000 dtype i comm Recv data MPI INT source 0 tag 77 automatic MPI datatype discovery if rank data numpy arange 100 dtype numpy float64 comm Send data dest 1 tag 13 elif rank data numpy empty 100 dtype numpy float 64 comm Recv data source 0 tag 13 4 2 Collective Communication e Broadcasting a Python dictionary from mpi4py import MPI comm MPI COMM_WORLD rank comm Get_rank if rank 0 data keyl 7 2 72 2 33 l key2 Cabe xyz
10. 6 3 39 Building MPI from SourceS s z is od on ee ee bo ta de o 17 Bibliography 19 Abstract This document describes the MPI for Python package MPI for Python provides bindings of the Message Passing Interface MPI standard for the Python programming language allowing any Python program to exploit multiple processors This package is constructed on top of the MPI 1 2 specifications and provides an object oriented inter face which closely follows MPI 2 C bindings It supports point to point sends receives and collec tive broadcasts scatters gathers communications of any picklable Python object as well as optimized communications of Python object exposing the single segment buffer interface NumPy arrays builtin bytes string array objects 1 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 messa
11. I COMM_WORLD py2f gt gt gt helloworld sayhello fcomm Hello World I am process 0 of 1 5 Appendix 5 1 MPl enabled Python interpreter Some MPI 1 implementations notably MPICH 1 do require the actual command line arguments to be passed at the time MP1_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 2 X of what is required is shown below include lt Python h gt include lt mpi h gt int main int argc char xargv int status flag MPI_Init 8argc amp argv status Py_Main argc argv MPI_Finalized amp 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 mpilpy X X X python setup py build_exe mpi lt name gt 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 SHOME bin pythonX X mpi Assuming that prefix bin or SHOME bin is listed on your PATH you should be able to ent
12. MPI for Python Release 1 3 Lisandro Dalcin January 20 2012 Contents 1 Introduction 2 E Whatis MPI 0 A A a a 2 1 2 Whatis Python ci ara Soke a a BG A aoe as 2 1 3 Related Projects ooo a Rb eA RO A ee A 3 2 Design and Interface Overview 4 2 1 Communicating Python Objects and Array Data ooo e 4 2722 ACOMMUMGCALOTS lt q ar dr tl ds de o 4 2 3 Point to Point Communications 2 2 e ee ee 5 24 Collective Communications s miis ee ce eee SOK oe Bey Both PAG Ge eA Be we 6 2 Dynamic Process Management coro cra a ee ee ee Pe al eae 6 2 6 One Sided Communications sche he SB se as a RO ee ea e aa 7 2 1 Parallel Input Output cross 86 be RA Oe hae eee wee ea Pe ee el 7 2 8 Environmental Management pto a e484 b85 4 e bee eee es 8 3 Installation 9 3 17 JREGUINEMENIS 4 403 003 porras oe ee ee oe eR ee ede ee GOR 9 3 2 Using pip oreasy_install ee ee ee 9 3 3 Using distatils lt o os cos ie ee o A ow we ee Re ds G 10 4 Tutorial 11 4 1 Point to Point Communication 2 2 0 ee ee 12 4 2 Collective COMMUNICAI N lt se 2 6 4 2042 8 4206848 eS e bd e ee aS 13 4 3 Dynamic Process Management ios e 6 e ee A a ew Oe 14 44 Wrappine with Wills s ss rt oie Bed Bo ee hh IEA me dete eA 14 45 Wrapping with F2PY sosa De eR eee Bare ar 15 5 Appendix 15 5 1 MPleenabled Python interpreter ee 15 5 2 MacOS X and Universal SDK Python builds o o 1
13. Proceedings 11th European PVM MPI Users Group Meeting Budapest Hungary September 2004 mpi lammpi Greg Burns Raja Daoud and James Vaigl LAM An Open Cluster Environment for MPI In Proceedings of Supercomputing Symposium pages 379 386 1994 Hinsen97 Konrad Hinsen The Molecular Modelling Toolkit a case study of a large scientific application in Python In Proceedings of the 6th International Python Conference pages 29 35 San Jose Ca October 1997 Beazley97 David M Beazley and Peter S Lomdahl Feeding a large scale physics application to Python In Proceedings of the 6th International Python Conference pages 21 29 San Jose Ca October 1997
14. d 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 Open MPI mpi openmpi or LAM mpi lammpi 1 2 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 1 MATLAB is a registered trademark of The MathWorks Inc 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 Hins
15. dent of machine architecture issues MPI for Python can communicate any built in or used defined Python object taking advantage of the features provided by the mod 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 im poses important overheads in memory as well as processor usage especially in the scenario of objects with large memory footprints being communicated Pickling generic Python objects ranging from primitive or container built in types to user defined classes necessarily requires computer resources Processing is also needed for dis patching 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 in 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
16. e locked remote window and to protect local load store accesses to a locked local window 2 7 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 t
17. ed in native MPI calls will raise an instance of the module exception class Exception which is a subclass of the standard Python exception RuntimeError Caution Importing with from mpi4py MPI import will cause a name clashing with standard Python Exception base class In order facilitate communicator sharing with other Python modules interfacing MPI based parallel li braries default MPI error handlers ERRORS_RETURN ERRORS_ARE_FATAL can be assigned to and retrieved from communicators windows and files with methods Class Set_errhandler and Class Get_errhandler 3 Installation 3 1 Requirements You need to have the following software properly installed in order to build MPI for Python e A working MPI distribution preferably a MPI 2 one 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 A Python 2 3 to 2 7 or 3 0 to 3 1 distribution with Python library preferably built with shared dynamic libraries Note Mac OS X users employing a Python distribution built with universal binaries may need to tem porarily set the environment variables MACOSX_DEPLOYMENT_TARGET SDKROOT and ARCHFLAGS to appropriate values in the shell before trying to build install MPI for Python Check the instructions at Mac OS X and Universal SDK Python builds in the appendix Note
18. en97 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 1 3 Related Projects As this work started and evolved some ideas were borrowed from well known MPI and Python related open source projects from the Internet e OOMPI It has not relation with Python but is an excellent object oriented approach to MPI It is a C class library specification layered on top of the C bindings that encapsulates MPI into a functional class hierarchy It provides a flexible and intuitive interface by adding some abstractions like Ports and Messages which enrich and simplify the syntax e Pypar Its interface is rather minimal There is no support for communicators or process topologies It does not require the Python interpreter to be modified or recompiled but does not permit interactive parallel runs General picklable Python objects of any type can be communicated There is good support for numeric arrays practically full MPI bandwidth can be achieved e pyMPI It rebuilds the Python interpreter providing a built in module for message passing It does permit interactive parallel runs which are useful for learning and debugging It provides an interface suitable for basic parallel programing There is not full support
19. er your MPI enabled Python interactively for example python2 6 mpi Python 2 6 r26 66714 Jun 8 2009 16 07 26 GCC 4 4 0 20090506 Red Hat 4 4 0 4 on linux2 Type help copyright credits or license for more information gt gt gt import sys gt gt gt sys executable usr bin python2 6 mpi gt gt gt 5 2 Mac OS X and Universal SDK Python builds Mac OS X users employing a Python distribution built with support for Universal applications could have trouble building MPI for Python specially if they want to link against MPI libraries built without such support An other source of trouble could be a Python build using a specific deployment target and cross development SDK configuration Workarounds for such issues are to temporarily set the environment variables MACOSX_DEPLOYMENT_TARGET SDKROOT and or ARCHF LAGS to appropriate values in the shell before trying to build install MPI for Python An appropriate value for MACOSX_DEPLOYMENT_TARGET should be any greater or equal than the one used to build Python and less or equal than your system version The safest choice for end users would be to use the system version e g if you are on Leopard you should try MACOSX_DEPLOYMENT_TARGET 10 5 An appropriate value for SDKROOT is the full path name of any of the SDK s you have at Developer SDKs directory e g SDKROOT Developer SDKs MacOSX10 5 sdk The safest choice for
20. for defining new communicators or process topologies General picklable Python objects can be messaged between processors There is not support for numeric arrays e Scientific Python It provides a collection of Python modules that are useful for scientific computing There is an interface to MPI and BSP Bulk Synchronous Parallel programming The interface is simple but incomplete and does not resemble the MPI specification There is support for numeric arrays Additionally we would like to mention some available tools for scientific computing and software development with Python e NumPy is a package that provides array manipulation and computational capabilities similar to those found in IDL MATLAB or Octave Using NumPy it is possible to write many efficient numerical data processing applications directly in Python without using any C C or Fortran code e SciPy is an open source library of scientific tools for Python gathering a variety of high level science and en gineering modules together as a single package It includes modules for graphics and plotting optimization integration special functions signal and image processing genetic algorithms ODE solvers and others e Cython is a language that makes writing C extensions for the Python language as easy as Python itself The Cython language is very close to the Python language but Cython additionally supports calling C functions and declaring C types on
21. ge passing parallel programming used to be a nightmare in the past because of the many incompat ible 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 is the dominant interpreted programming language In the open source side Octave and Scilab are well known freely distributed software packages provid ing 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 1 1 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 an
22. ittest scripts mpiexec n 5 python test runalltest py or if you have nose unit testing framework installed mpiexec n 5 nosetests w test 4 Tutorial Warning Under construction Contributions very welcome 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 e Communication of generic Python objects You have to use all lowercase methods of the Comm class like send recv bcast Note that isend is available but irecv is not Collective calls like scatter gather allgather alltoall expect return 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 Global reduction operations reduce and allreduce are naively implemented the reduction is actually done at the designated root process or all processes e Communication of buffer provider objects You have to use method names starting with an upper case letter of the Comm class like Send Recv Bcast 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 the count Automatic MPI datatype discovery for NumPy arrays and PEP 31
23. o 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 forPython all MPI input output operations are performed through instances of the File class File handles are obtained by calling the 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 Close method Files even can be deleted by calling method 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 Set_view and Get_view methods respectively Actual input output operations are achieved by many methods combining read and write calls with different be havior 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 noncollecti
24. ow 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 e MPICH 2 tar zxf mpich2 X X X tar gz cd mpich2 X X X configur enable shared prefix usr local mpich2 make make install UN UY 1 Y Ww e Open MPI tar zxf openmpi X X X tar gz cd openmpi X X X configure prefix usr local openmpi make all make install UN UY 11 mM Ww e LAM MPI tar zxf lam X X X tar gz cd lam X X X configur nable shared prefix usr local lam make make install UN UY 1 Y Ww e MPICH 1 tar zxf mpich X X X tar gz cd mpich X X X configur enable sharedlib prefix usr local mpichl make make install MMM Mm Ww Perhaps you will need to set the LD_LIBRARY_PATH environment variable using export setenv or what ap plies 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 e MPICH 2 MPI_DIR usr local mpich2 export LD_LIBRARY_PATH MPI_DIR lib S LD_LIBRARY_PATH e Open MPI MPI_DIR usr local openmpi export LD_LIBRARY_PATH MPI_DIR lib LD_LIBRARY_PATH e LAM MPI MPI_DIR usr local lam export LD_LIBRARY PATH MPI_DIR lib LD_LIBRARY_PATH e MPICH 1 MPI_DIR usr local mpichl export LD_LIBRARY_PATH MPI_DIR lib
25. perations are the following e Barrier synchronization across all group members e 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 e Global reduction operations such as sum maximum minimum etc MPI for Python provides support for almost all collective calls Unfortunately the Alltoallw and Reduce_scatter methods are curently unimplemented In MPI for Python the Bcast Scatter Gather Allgather and Alltoall methods of Comm instances provide support for collective communications of memory buffers The variants bcast scatter gather allgather and alltoall can communicate generic Python objects The vector variants which can communicate different amounts of data to each process Scatterv Gatherv Allgatherv and Alltoallv are also supported they can only communicate objects exposing memory buffers Global reduction operations on memory buffers are accessible through the Reduce Allreduce Scan and Exscan methods The variants reduce allreduce scan and exscan can communi cate generic 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 2 5 Dynamic Process Management In
26. plication program In MPI for Python the Send Recv and Sendrecv methods of communicator objects provide support for blocking point to point communications within Int racomm and Intercomm instances These methods can communicate memory buffers The variants send recv and sendrecv can communicate generic 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 dedi cated 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 re quested operation and test for completion functions which allow to discover whether the requested operation has completed In MPI for Python the Isend and Irecv methods of the Comm class initiate a send and receive oper ation respectively These methods return a Request instance uniquely identifying the started operation Its completion can be managed using the Test Wait and Cancel methods of the Request class 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 acce
27. r instances can be obtained with the Clone method of Comm objects the Dup and Split methods of Intracomm and Intercomm objects and methods Create_intercomm and Merge of Intracomm and Intercomm objects respectively Virtual topologies Cartcomm Graphcomm and Distgraphcomm classes being them specializations of Intracomm class are fully supported New instances can be obtained from intracommunicator instances with factory methods Create_cart and Create_graph of Intracomm class 2 3 Point to Point Communications Point to point communication is a fundamental capability of message passing systems This mechanism enables the transmittal 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 ap
28. s option should be passed in order to build MPI for Python against old MPI 1 implementations possibly providing a subset of MPI 2 Downloading The MPI for Python package is available for download at the project website generously hosted by Google Code You can use curl or wget to get a release tarball curl O http mpi4py googlecode com files mpi4py X X X tar gz wget http mpi4py googlecode com files mpi4py X X X tar gz Building After unpacking the release tarball tar zxf mpi4py X X X tar gz cd mpi4py X X X the distribution is ready for building e If you use a MPI implementation providing a mpicc compiler wrapper e g MPICH 1 2 Open MPI LAM it will be used for compilation and linking This is the preferred and easiest way of building MPI for Python If mpicc is located somewhere in your search path simply run the build command python setup py build If mpice is not in your search path or the compiler wrapper has a different name you can run the build command specifying its location python setup py build mpicc where you have mpicc e Alternatively you can provide all the relevant information about your MPI distribution by editing the file called mpi cfg You can use the default section mpi or add a new custom section for example my_mpi see the examples provided in the mpi c g file mpi include_dirs usr local mpi include libraries mpi library_dirs usr local mpi lib run
29. shared SLD_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 mpice compiler wrapper References mpi std1 MPI Forum MPI A Message Passing Interface Standard International Journal of Supercomputer Applications volume 8 number 3 4 pages 159 416 1994 mpi std2 MPI Forum MPI A Message Passing Interface Standard High Performance Computing Applications volume 12 number 1 2 pages 1 299 1998 mpi using William Gropp Ewing Lusk and Anthony Skjellum Using MPI portable parallel programming with the message passing interface MIT Press 1994 mpi ref Mark Snir Steve Otto Steven Huss Lederman David Walker and Jack Dongarra MPI The Complete Reference volume 1 The MPI Core MIT Press 2nd edition 1998 mpi mpich W Gropp E Lusk N Doss and A Skjellum A high performance portable implementation of the MPI message passing interface standard Parallel Computing 22 6 789 828 September 1996 mpi openmpi Edgar Gabriel Graham E Fagg George Bosilca Thara Angskun Jack J Dongarra Jeffrey M Squyres Vishal Sahay Prabhanjan Kambadur Brian Barrett Andrew Lumsdaine Ralph H Castain David J Daniel Richard L Graham and Timothy S Woodall Open MPI Goals Concept and Design of a Next Generation MPI Implementation In
30. ssed 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 Fur thermore this kind of optimization can also alleviate the extra call overheads associated to interpreted dynamic languages like Python In MPI for Python the Send_init andRecv_init methods of the Comm class create a persistent request 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 Start method and its completion can be managed as previously described 2 4 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 o
31. 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 processes groups can be created by calling the Spawn method within an intracommunicator i e an Intracomm instance This call returns a new intercommunicator i e an Intercomm instance at the parent process group The child process group can retrieve the matching inter communicator by calling the Get_parent class method defined in the Comm class 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
32. time_library_dirs usr local mpi lib other_mpi include_dirs opt mpi include libraries mpi library_dirs opt mpi lib runtime_library_dirs op mpi lib and then run the build command perhaps specifying you custom configuration section python setup py build mpi other_mpi Installing After building the distribution is ready for install If you have root privileges either by log in as the root user of by using sudo and you want to install MPI for Python in your system for all users just do python setup py install The previous steps will install the mpi4py package at standard location prefix lib pythonX X site packages If you do not have root privileges or you want to install MPI for Python for your private use you have two options depending on the target Python version e For Python 2 6 and up python setup py install user e For Python 2 5 and below assuming your home directory is available through the HOME environment vari able python setup py install home HOME Finally add HOME 1ib python or HOME 1ib64 python to your PYTHONPATH environment variable Testing Issuing at the command line mpiexec n 5 python demo helloworld py or in the case of older MPI 1 implementations mpirun np 5 python demo helloworld py will launch a five process run of the Python interpreter and run the test scripts demo helloworld py You can also run all the un
33. ve versions 2 8 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 Caution 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 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 automat ically called when Python processes exit but only if mpi4py actually initialized Therefore there is no need to call Finalize from Python to ensure MPI finalization Implementation Information e The MPI version number can be retrieved from module function Get_version It returns a two integer tuple version subversion e The Get_processor_name function can be used to access the processor name e The values of predefined attributes attached to the world communicator can be obtained by calling the Get_attr method within the COMM_WORLD instance Timers MPI timer functionalities are available through the Wtime and Wtick functions Error Handling Error handling functionality is almost completely supported Errors originat
Download Pdf Manuals
Related Search