A forwarding MPI implementation that can use any other MPI implementation via an MPI ABI

Overview

MPItrampoline

  • MPI wrapper library: GitHub CI
  • MPI trampoline library: GitHub CI
  • MPI integration tests: GitHub CI

MPI is the de-facto standard for inter-node communication on HPC systems, and has been for the past 25 years. While highly successful, MPI is a standard for source code (it defines an API), and is not a standard defining binary compatibility (it does not define an ABI). This means that applications running on HPC systems need to be compiled anew on every system. This is tedious, since the software that is available on every HPC system is slightly different.

This project attempts to remedy this. It defines an ABI for MPI, and provides an MPI implementation based on this ABI. That is, MPItrampoline does not implement any MPI functions itself, it only forwards them to a "real" implementation via this ABI. The advantage is that one can produce "portable" applications that can use any given MPI implementation. For example, this will make it possible to build external packages for Julia via Yggdrasil that run efficiently on almost any HPC system.

A small and simple MPIwrapper library is used to provide this ABI for any given MPI installation. MPIwrapper needs to be compiled for each MPI installation that is to be used with MPItrampoline, but this is quick and easy.

Successfully Tested

  • Debian 11.0 via Docker (MPICH; arm32v5, arm32v7, arm64v8, mips64le, ppc64le, riscv64; C/C++ only)
  • Debian 11.0 via Docker (MPICH; i386, x86-64)
  • macOS laptop (MPICH, OpenMPI; x86-64)
  • macOS via Github Actions (OpenMPI; x86-64)
  • Ubuntu 20.04 via Docker (MPICH; x86-64)
  • Ubuntu 20.04 via Github Actions (MPICH, OpenMPI; x86-64)
  • Blue Waters, HPC system at the NCSA (Cray MPICH; x86-64)
  • Graham, HPC system at Compute Canada (Intel MPI; x86-64)
  • Marconi A3, HPC system at Cineca (Intel MPI; x86-64)
  • Niagara, HPC system at Compute Canada (OpenMPI; x86-64)
  • Summit, HPC system at ORNL (Spectrum MPI; IBM POWER 9)
  • Symmetry, in-house HPC system at the Perimeter Institute (MPICH, OpenMPI; x86-64)

Workflow

Preparing an HPC system

Install MPIwrapper, wrapping the MPI installation you want to use there. You can install MPIwrapper multiple times if you want to wrap more than one MPI implementation.

This is possibly as simple as

cmake -S . -B build -DMPIEXEC_EXECUTABLE=mpiexec -DCMAKE_BUILD_TYPE=RelWithDebInfo -DCMAKE_INSTALL_PREFIX=$HOME/mpiwrapper
cmake --build build
cmake --install build

but nothing is ever simple on an HPC system. It might be necessary to load certain modules, or to specify more cmake MPI configuration options.

The MPIwrapper libraries remain on the HPC system, they are installed independently of any application.

Building an application

Build your application as usual, using MPItrampline as MPI library.

Running an application

At startup time, MPItrampoline needs to be told which MPIwrapper library to use. This is done via the environment variable MPITRAMPOLINE_LIB. You also need to point MPItrampoline's mpiexec to a respective wrapper created by MPIwrapper, using the environment variable MPITRAMPOLINE_MPIEXEC.

For example:

env MPITRAMPOLINE_MPIEXEC=$HOME/mpiwrapper/bin/mpiwrapper-mpiexec MPITRAMPOLINE_LIB=$HOME/mpiwrapper/lib/libmpiwrapper.so mpiexec -n 4 ./your-application

The mpiexec you run here needs to be the one provided by MPItrampoline.

Current state

MPItrampoline uses the C preprocessor to create wrapper functions for each MPI function. This is how MPI_Send is wrapped:

FUNCTION(int, Send,
         (const void *buf, int count, MT(Datatype) datatype, int dest, int tag,
          MT(Comm) comm),
         (buf, count, (MP(Datatype))datatype, dest, tag, (MP(Comm))comm))

Unfortunately, MPItrampoline does not yet wrap the Fortran API. Your help is welcome.

Certain MPI types, constants, and functions are difficult to wrap. Theoretically, there could be MPI libraries where it is not possible to implement the current MPI ABI. If you encounter this, please let me know -- maybe there is a work-around.

Comments
  • Add support for MPI profiling interface

    Add support for MPI profiling interface

    Each standard MPI function can be called with an MPI_ or PMPI_ prefix (quoting from https://www.open-mpi.org/faq/?category=perftools#PMPI), I think MPItrampoline doesn't currently support the PMPI_ calls.

    opened by ocaisa 6
  • Supporting `MPIX_Query_cuda_support()`

    Supporting `MPIX_Query_cuda_support()`

    While not part of the standard, MPIX_Query_cuda_support() is available in a number of MPI implementations (see https://github.com/pmodels/mpich/pull/4741). It's also being used by a number of applications (and hopefully that will grow, see my issue at https://github.com/lammps/lammps/issues/3140 which links back to the support in GROMACS). This would be a valuable inclusion in MPItrampoline since it would then be able to handle the runtime detection of CUDA support in the MPI implementation.

    opened by ocaisa 6
  • Allow overriding default compilation options

    Allow overriding default compilation options

    Currently default compilation options are set during the build of MPItrampoline, it would probably be useful to be able to fully override these, e.g.,

    exec ${MPITRAMPOLINE_CC:[email protected]_C_COMPILER@} ${CFLAGS:-"@CMAKE_C_FLAGS@"} [email protected]_INSTALL_PREFIX@/@CMAKE_INSTALL_INCLUDEDIR@ @LINK_FLAGS@ [email protected]_INSTALL_PREFIX@/@CMAKE_INSTALL_LIBDIR@ -Wl,-rpath,@CMAKE_INSTALL_PREFIX@/@CMAKE_INSTALL_LIBDIR@ "$@" -lmpi -ldl
    
    opened by ocaisa 5
  • Allow use of a fallback/default value for `MPITRAMPOLINE_MPIEXEC`

    Allow use of a fallback/default value for `MPITRAMPOLINE_MPIEXEC`

    It's possible to configure a default MPI library with -DMPITRAMPOLINE_DEFAULT_LIB=XXX, it would be good to be able to also configure a default mpiexec (-DMPITRAMPOLINE_DEFAULT_MPIEXEC=XXX) so that one can have a fully functional fallback in place at build time.

    opened by ocaisa 3
  • Incomplete installation

    Incomplete installation

    Hello ! I am looking at installing MPItrampoline, and following the steps outlined in the README.md, I cannot access files later referenced.

    For instance, on Summit at ORNL, using the following script:

    INSTALLDIR=$HOME/mpiwrapper
    
    module load cmake/3.18
    module load gcc
    
    cmake -S . -B build -DMPIEXEC_EXECUTABLE=mpiexec \
                                   -DCMAKE_BUILD_TYPE=RelWithDebInfo \
                                   -DCMAKE_INSTALL_PREFIX=$INSTALLDIR
    cmake --build build
    cmake --install build
    

    The following installation is generated:

    $ tree mpiwrapper/
    mpiwrapper/
    |-- bin
    |   |-- mpicc
    |   |-- mpicxx
    |   |-- mpiexec
    |   |-- mpifc
    |   `-- mpifort
    |-- include
    |   |-- mpi.h
    |   |-- mpi.mod
    |   |-- mpi_declarations.h
    |   |-- mpi_declarations_fortran.h
    |   |-- mpi_declarations_fortran90.h
    |   |-- mpi_defaults.h
    |   |-- mpi_f08.mod
    |   |-- mpi_version.h
    |   |-- mpiabi.h
    |   |-- mpiabif.h
    |   |-- mpif.h
    |   `-- mpio.h
    |-- lib
    |   |-- cmake
    |   |   `-- MPItrampoline
    |   |       |-- MPItrampolineConfig.cmake
    |   |       |-- MPItrampolineConfigVersion.cmake
    |   |       |-- MPItrampolineTargets-relwithdebinfo.cmake
    |   |       `-- MPItrampolineTargets.cmake
    |   `-- pkgconfig
    |       `-- MPItrampoline.pc
    `-- lib64
        |-- libmpi.a
        `-- libmpifort.a
    
    7 directories, 24 files
    

    The README.md dictates to use the wrapped libraries, but no shared libraries are available here. I looked through the options of the CMakeLists.txt but the only one defined in the project is the fortran flag.

    Am I missing something ?

    opened by spoutn1k 2
  •  Issues building Global Arrays

    Issues building Global Arrays

    I'm trying to build the Global Arrays library with MPItrampoline and have run into some issues. This package is a dependency for Molpro and several other computational chemistry applications. Any assistance that you can provide to get it working would be greatly appreciated.

    The compilation fails at https://github.com/GlobalArrays/ga/blob/f4016b869dfd1a2b2856f74a73dd9452dbbc8ae4/comex/src-mpi-pr/comex.c#L4968 with the error message:

    libtool: compile:  mpicc -DHAVE_CONFIG_H -I. -I./src-common -I./src-mpi-pr -g -O2 -MT src-mpi-pr/comex.lo -MD -MP -MF src-mpi-pr/.deps/comex.Tpo -c src-mpi-pr/comex.c -o src-mpi-pr/comex.o
    src-mpi-pr/comex.c: In function ‘str_mpi_retval’:
    src-mpi-pr/comex.c:4932:9: error: case label does not reduce to an integer constant
             case MPI_SUCCESS       : msg = "MPI_SUCCESS"; break;
             ^
    

    Steps to reproduce:

    git clone -b develop https://github.com/GlobalArrays/ga
    cd ga
    ./autogen.sh
    ./configure --with-mpi-pr --with-blas=no --with-lapack=no --with-scalapack=no --disable-f77
    make
    

    I've tried both the develop and master branches.

    There are lots of configurations which can be used, we're most interested in the recommended port which uses MPI-1 with progress ranks (--with-mpi-pr) but I tried several other configurations without success.

    I'm not sure whether it's an MPItrampoline issue or whether it's non-standard use of MPI within Global Arrays.

    I've attached the full output from build build-ga.txt

    I've tried on

    • RHEL 7.9 with gcc 4.8.5
    • Ubuntu 22.04 with gcc 11.3.0
    opened by nick-wilson 9
  • Issues building CP2K

    Issues building CP2K

    I thought I would give this a full test with Fortran, and CP2K is a good benchmark for that. The build (v8.2) is failing with:

    /project/60005/easybuild/build/CP2K/8.2/gmtfbf-2021a/cp2k-8.2/exts/dbcsr/src/mpi/dbcsr_mpiwrap.F:1669:21:
    
     1669 |       CALL mpi_bcast(msg, msglen, MPI_LOGICAL, source, gid, ierr)
          |                     1
    ......
     3160 |       CALL mpi_bcast(msg, msglen, ${mpi_type1}$, source, gid, ierr)
          |                     2
    Error: Type mismatch between actual argument at (1) and actual argument at (2) (LOGICAL(4)/COMPLEX(4)).
    
    opened by ocaisa 24
  • Building shared and static libraries at once

    Building shared and static libraries at once

    Currently the default behaviour is to only build static libraries. It might be good build both static and shared libraries since then if libmpi.so is in the default search path it is not selected over the library from MPItrampoline (MPItrampoline would shadow libmpi.so and libmpi.a).

    opened by ocaisa 4
Releases(v5.2.0)
Owner
Erik Schnetter
Erik Schnetter
Leveraging OpenAI's Codex to solve cornerstone problems in Music

Music-Codex Leveraging OpenAI's Codex to solve cornerstone problems in Music Please NOTE: Presented generated samples were created by OpenAI's Codex P

Alex 2 Mar 11, 2022
GCNet: Non-local Networks Meet Squeeze-Excitation Networks and Beyond

GCNet for Object Detection By Yue Cao, Jiarui Xu, Stephen Lin, Fangyun Wei, Han Hu. This repo is a official implementation of "GCNet: Non-local Networ

Jerry Jiarui XU 1.1k Dec 29, 2022
Official implementation of our neural-network-based fast diffuse room impulse response generator (FAST-RIR)

This is the official implementation of our neural-network-based fast diffuse room impulse response generator (FAST-RIR) for generating room impulse responses (RIRs) for a given acoustic environment.

12 Jan 13, 2022
PyTorch implementation of SimSiam: Exploring Simple Siamese Representation Learning

SimSiam: Exploring Simple Siamese Representation Learning This is a PyTorch implementation of the SimSiam paper: @Article{chen2020simsiam, author =

Facebook Research 834 Dec 30, 2022
Collects many various multi-modal transformer architectures, including image transformer, video transformer, image-language transformer, video-language transformer and related datasets

The repository collects many various multi-modal transformer architectures, including image transformer, video transformer, image-language transformer, video-language transformer and related datasets

Jun Chen 139 Dec 21, 2022
Repository for open research on optimizers.

Open Optimizers Repository for open research on optimizers. This is a test in sharing research/exploration as it happens. If you use anything from thi

Ariel Ekgren 6 Jun 24, 2022
PFFDTD is an open-source FDTD simulator for 3D room acoustics

PFFDTD is an open-source FDTD simulator for 3D room acoustics

Brian Hamilton 34 Nov 24, 2022
Implementation of TabTransformer, attention network for tabular data, in Pytorch

Tab Transformer Implementation of Tab Transformer, attention network for tabular data, in Pytorch. This simple architecture came within a hair's bread

Phil Wang 420 Jan 05, 2023
PyTorch Code of "Memory In Memory: A Predictive Neural Network for Learning Higher-Order Non-Stationarity from Spatiotemporal Dynamics"

Memory In Memory Networks It is based on the paper Memory In Memory: A Predictive Neural Network for Learning Higher-Order Non-Stationarity from Spati

Yang Li 12 May 30, 2022
FedMM: Saddle Point Optimization for Federated Adversarial Domain Adaptation

This repository contains the code accompanying the paper " FedMM: Saddle Point Optimization for Federated Adversarial Domain Adaptation" Paper link: R

20 Jun 29, 2022
Code for generating the figures in the paper "Capacity of Group-invariant Linear Readouts from Equivariant Representations: How Many Objects can be Linearly Classified Under All Possible Views?"

Code for running simulations for the paper "Capacity of Group-invariant Linear Readouts from Equivariant Representations: How Many Objects can be Lin

Matthew Farrell 1 Nov 22, 2022
Convert onnx models to pytorch.

onnx2torch onnx2torch is an ONNX to PyTorch converter. Our converter: Is easy to use – Convert the ONNX model with the function call convert; Is easy

ENOT 264 Dec 30, 2022
Anomaly detection related books, papers, videos, and toolboxes

Anomaly Detection Learning Resources Outlier Detection (also known as Anomaly Detection) is an exciting yet challenging field, which aims to identify

Yue Zhao 6.7k Dec 31, 2022
A task Provided by A respective Artenal Ai and Ml based Company to complete it

A task Provided by A respective Alternal Ai and Ml based Company to complete it .

Parth Madan 1 Jan 25, 2022
MNE: Magnetoencephalography (MEG) and Electroencephalography (EEG) in Python

MNE-Python MNE-Python software is an open-source Python package for exploring, visualizing, and analyzing human neurophysiological data such as MEG, E

MNE tools for MEG and EEG data analysis 2.1k Dec 28, 2022
2021 National Underwater Robotics Vision Optics

2021-National-Underwater-Robotics-Vision-Optics 2021年全国水下机器人算法大赛-光学赛道-B榜精度第18名 (Kilian_Di的团队:A榜[email pro

Di Chang 9 Nov 04, 2022
An image classification app boilerplate to serve your deep learning models asap!

Image 🖼 Classification App Boilerplate Have you been puzzled by tons of videos, blogs and other resources on the internet and don't know where and ho

Smaranjit Ghose 27 Oct 06, 2022
Code for Motion Representations for Articulated Animation paper

Motion Representations for Articulated Animation This repository contains the source code for the CVPR'2021 paper Motion Representations for Articulat

Snap Research 851 Jan 09, 2023
Efficient semidefinite bounds for multi-label discrete graphical models.

Low rank solvers #################################### benchmark/ : folder with the random instances used in the paper. ############################

1 Dec 08, 2022
Code for "AutoMTL: A Programming Framework for Automated Multi-Task Learning"

AutoMTL: A Programming Framework for Automated Multi-Task Learning This is the website for our paper "AutoMTL: A Programming Framework for Automated M

Ivy Zhang 40 Dec 04, 2022