{{ page.title }}
+ Author: {{ page.author }}+
{{ content }} @@ -11,7 +40,7 @@
{{ page.title }}
Want to contribute?
-This site is hosted entirely on Github. This site is no longer being actively contributed to by the original author (Wes Kendall), but it was placed on Github in the hopes that others would write high-quality MPI tutorials. Click here for more information about how you can contribute.
+This site is hosted entirely on GitHub. This site is no longer being actively contributed to by the original author (Wes Kendall), but it was placed on GitHub in the hopes that others would write high-quality MPI tutorials. Click here for more information about how you can contribute.
diff --git a/about.md b/about.md index 0454158..c749748 100644 --- a/about.md +++ b/about.md @@ -7,13 +7,13 @@ This site is a collaborative space for providing tutorials about MPI (the Messag ## Contributing -This site is hosted as a static page on [Github]({{ site.github.code }}). It is no longer actively contributed to by the original author, and any potential authors are encouraged to fork the repository [here]({{ site.github.code }}) and start writing a lesson! +This site is hosted as a static page on [GitHub]({{ site.github.code }}). It is no longer actively contributed to by the original author, and any potential authors are encouraged to fork the repository [here]({{ site.github.code }}) and start writing a lesson! -Github uses Jekyll, a markdown-based blogging framework for producing static HTML pages. For an introduction on using Jekyll with Github, checkout [this article](https://help.github.com/articles/using-jekyll-with-pages/). +GitHub uses Jekyll, a markdown-based blogging framework for producing static HTML pages. For an introduction on using Jekyll with GitHub, checkout [this article](https://help.github.com/articles/using-jekyll-with-pages/). All lessons are self-contained in their own directories in the *tutorials* directory of the main repository. New tutorials should go under this directory, and any code for the tutorials should go in the *code* directory of the tutorial and provide a Makefile with executable examples. Similarly, the structure of the posts should match other tutorials. -For those that have never used Github or may feel overwhelmed about contributing a tutorial, contact Wes Kendall first at wesleykendall AT gmail DOT com. If you wish to write a tutorial with images as a Microsoft Word document or PDF, I'm happy to translate the lesson into the proper format for the site. +For those that have never used GitHub or may feel overwhelmed about contributing a tutorial, contact Wes Kendall first at wesleykendall AT gmail DOT com. If you wish to write a tutorial with images as a Microsoft Word document or PDF, I'm happy to translate the lesson into the proper format for the site. > **Note** - The tutorials on this site need to remain as informative as possible and encompass useful topics related to MPI. Before writing a tutorial, collaborate with me through email (wesleykendall AT gmail DOT com) if you want to propose a lesson to the beginning MPI tutorial. Similarly, we can also start an advanced MPI tutorial page for more advanced topics. @@ -23,3 +23,12 @@ For those that have never used Github or may feel overwhelmed about contributing Wes Kendall is the original author of mpitutorial.com. As a graduate student at the University of Tennessee, Knoxville, Wes earned his PhD under Jian Huang. His research revolved around large-scale data analysis and visualization, and he worked with the biggest supercomputers in the world. As a graduate student, he interned at Google, Oak Ridge National Labs, and Argonne National Labs. His research also earned him the Supercomputing 2011 Best Student Paper Award. He is currently co-founder and CTO of [Ambition](http://ambition.com), a data-analytics startup funded by YCombinator, Google Ventures, and several other top investment firms. Disappointed with the amount of freely-available content on parallel programming and MPI, Wes started releasing tutorials on the subject after graduate school. Once his startup consumed most of his time, he opened up mpitutorial.com to the public on github.com so that others could start contributing high-quality content. + +### Dwaraka Nath + +Dwaraka Nath is a masters graduate from Birla Institute of Technology and Science, Pilani, India. He loves blogging and occasionally does some code contributions as well. + +You can find more about him on his [personal website](https://www.dwarak.in) and follow him on GitHub at [@dtsdwarak](https://github.com/dtsdwarak). + +### Wesley Bland +Wesley Bland is a researcher in High Performance Computing and a contributor to both MPICH and Open MPI. He graduated from the University of Tennessee, Knoxville with his PhD under Dr. Jack Dongarra. His research involved fault tolerance at scale using MPI. After leaving the university, he went to Argonne National Laboratory where he worked under Dr. Pavan Balaji as a postdoctoral appointee and continued his fault tolerance research while working on MPICH directly. He has also worked at Intel and Meta. diff --git a/advanced-mpi-tutorial.md b/advanced-mpi-tutorial.md new file mode 100644 index 0000000..225b4b3 --- /dev/null +++ b/advanced-mpi-tutorial.md @@ -0,0 +1,10 @@ +--- +layout: page +title: Advanced MPI Tutorial +--- + +Welcome to the advanced MPI tutorial! In this tutorial, you will learn some of the more advanced concepts of MPI. Below are the available lessons, each of which contain example code. + +This tutorial assumes that the reader understands the topics in the basic tutorial. + +* [Introduction to Groups and Communicators]({{ site.baseurl }}/tutorials/introduction-to-groups-and-communicators/) diff --git a/atom.xml b/atom.xml index 2e355d8..229c82d 100644 --- a/atom.xml +++ b/atom.xml @@ -1,5 +1,5 @@ --- -layout: nil +layout: null --- diff --git a/beginner-mpi-tutorial.md b/beginner-mpi-tutorial.md deleted file mode 100644 index 1d47370..0000000 --- a/beginner-mpi-tutorial.md +++ /dev/null @@ -1,27 +0,0 @@ ---- -layout: page -title: Beginner MPI Tutorial ---- - -Welcome to the MPI tutorial for beginners! In this tutorial, you will learn the basic concepts of MPI. Below are the available lessons, each of which contain example code. - -This beginner tutorial assumes that the reader has a basic knowledge of C, some C++, and Linux. - -## Introduction and MPI installation -* [MPI tutorial introduction]({{ site.baseurl }}/tutorials/mpi-introduction/) -* [Installing MPICH2 on a single machine]({{ site.baseurl }}/tutorials/installing-mpich2/) -* [Launching an Amazon EC2 MPI cluster]({{ site.baseurl }}/tutorials/launching-an-amazon-ec2-mpi-cluster/) -* [Running an MPI hello world application]({{ site.baseurl }}/tutorials/mpi-hello-world/) - -## Blocking point-to-point communication -* [Sending and receiving with MPI_Send and MPI_Recv]({{ site.baseurl }}/tutorials/mpi-send-and-receive/) -* [Dynamic receiving with MPI_Probe and MPI_Status]({{ site.baseurl }}/tutorials/dynamic-receiving-with-mpi-probe-and-mpi-status/) -* [Point-to-point communication application - Random walking]({{ site.baseurl }}/tutorials/point-to-point-communication-application-random-walk/) - -## Basic collective communication -* [Collective communication introduction with MPI_Bcast]({{ site.baseurl }}/tutorials/mpi-broadcast-and-collective-communication/) -* [Common collectives - MPI_Scatter, MPI_Gather, and MPI_Allgather]({{ site.baseurl }}/tutorials/mpi-scatter-gather-and-allgather/) -* [Application example - Performing parallel rank computation with basic collectives]({{ site.baseurl }}/tutorials/performing-parallel-rank-with-mpi/) - -## Advanced collective communication -* [Using MPI_Reduce and MPI_Allreduce for parallel number reduction]({{ site.baseurl }}/tutorials/mpi-reduce-and-allreduce/) \ No newline at end of file diff --git a/index.md b/index.md index b298745..b3a666b 100644 --- a/index.md +++ b/index.md @@ -5,11 +5,11 @@ title: A Comprehensive MPI Tutorial Resource Welcome to mpitutorial.com, a website dedicated to providing useful tutorials about the Message Passing Interface (MPI). -## Beginner Tutorial -Wanting to get started learning MPI? Head over to the [beginner MPI tutorial]({{ site.baseurl }}/beginner-mpi-tutorial/). +## Tutorials +Wanting to get started learning MPI? Head over to the [MPI tutorials]({{ site.baseurl }}/tutorials/). ## Recommended Books Recommended books for learning MPI are located [here]({{ site.baseurl }}/recommended-books/). ## About -This site is no longer being actively contributed to by its original author (Wes Kendall). However, mpitutorial.com has been placed on [Github]({{ site.github.code}}) so that others can contribute high-quality content. Click [here]({{ site.baseurl }}/about/) for more details on how to contribute. +This site is no longer being actively contributed to by its original author (Wes Kendall). However, mpitutorial.com has been placed on [GitHub]({{ site.github.code}}) so that others can contribute high-quality content. Click [here]({{ site.baseurl }}/about/) for more details on how to contribute. diff --git a/recommended-books.md b/recommended-books.md index 7bb6542..6dd1208 100644 --- a/recommended-books.md +++ b/recommended-books.md @@ -3,31 +3,34 @@ layout: page title: Recommended Books --- -Here are the books that helped me out the most when learning MPI. Most of these are written by the primary designers of the Message Passing Interface. Some also include tutorials on how to use OpenMP with MPI. The very first book is a compilation of the beginner tutorials of this site as well and helps support mpitutorial.com. +Here are the books that helped me out the most when learning MPI. Most of these are written by the primary designers of the Message Passing Interface. Some also include tutorials on how to use OpenMP with MPI. -## Beginning MPI (An Introduction in C) -This book is a compilation of all of the beginner tutorials on this site. It goes over everything from installing MPI on an Amazon EC2 cluster to the basics of sending and receiving with MPI to performing collective operations and reductions. If you have enjoyed the tutorials on this site and wish to have a copy of it in book format (while also supporting mpitutorial.com), then [click here for more info](http://www.amazon.com/gp/product/B00HM7O0M8/ref=as_li_qf_sp_asin_il_tl?ie=UTF8&camp=1789&creative=9325&creativeASIN=B00HM7O0M8&linkCode=as2&tag=softengiintet-20). - -[](http://www.amazon.com/gp/product/B00HM7O0M8/ref=as_li_qf_sp_asin_il_tl?ie=UTF8&camp=1789&creative=9325&creativeASIN=B00HM7O0M8&linkCode=as2&tag=softengiintet-20) +> **Disclaimer** - The last book is primarily a compilation of these tutorials by the original mpitutorial.com author (Wes Kendall). ## Parallel Programming with MPI -My personal favorite MPI book. The book gives a good overview of parallel computing before delving into all the various topics of MPI programming. It goes into detail about almost every essential MPI routine, and then it provides examples of parallel programs such as matrix multiplication and sorting. The end of the book discusses libraries and debugging routines. [Click here for more info](http://www.amazon.com/gp/product/1558603395/ref=as_li_qf_sp_asin_il?ie=UTF8&tag=softengiintet-20&linkCode=as2&camp=217145&creative=399377&creativeASIN=1558603395). +My personal favorite MPI book. The book gives a good overview of parallel computing before delving into all the various topics of MPI programming. It goes into detail about almost every essential MPI routine, and then it provides examples of parallel programs such as matrix multiplication and sorting. The end of the book discusses libraries and debugging routines. [Click here for more info](http://www.amazon.com/gp/product/1558603395/ref=as_li_qf_sp_asin_il?ie=UTF8&linkCode=as2&camp=217145&creative=399377&creativeASIN=1558603395). -[](http://www.amazon.com/gp/product/1558603395/ref=as_li_qf_sp_asin_il?ie=UTF8&tag=softengiintet-20&linkCode=as2&camp=217145&creative=399377&creativeASIN=1558603395) +[](http://www.amazon.com/gp/product/1558603395/ref=as_li_qf_sp_asin_il?ie=UTF8&linkCode=as2&camp=217145&creative=399377&creativeASIN=1558603395) -## Using MPI - 2nd Edition -This is a more up-to-date book than the previous, but it mostly focuses on the newer and more advanced MPI routines in the second MPI standard. These include parallel I/O, remote memory access, and dynamic process management. The book also discusses using MPI with threads. This is a must have for advanced MPI development. [Click here for more info](http://www.amazon.com/gp/product/0262571323/ref=as_li_tf_il?ie=UTF8&tag=softengiintet-20&linkCode=as2&camp=217145&creative=399377&creativeASIN=0262571323). +## Using MPI - 3rd Edition and Using Advanced MPI - 1st Edition +This is a more up-to-date book than the previous. The "regular" book covers the fundamentals of MPI and the "advnaced" book covers additional topics. The table of contents can be found on [this](https://www.mcs.anl.gov/research/projects/mpi/usingmpi/) website. This is a must have for advanced MPI development. [Click here for Using MPI](https://www.amazon.com/gp/product/0262527391/ref=as_li_tl?ie=UTF8&camp=1789&creative=9325&creativeASIN=0262527391&linkCode=as2&linkId=52cc270bc3a798b37c787740d00c19d1). The "Using Advanced MPI" book is currently out of print. -[](http://www.amazon.com/gp/product/0262571323/ref=as_li_tf_il?ie=UTF8&tag=softengiintet-20&linkCode=as2&camp=217145&creative=399377&creativeASIN=0262571323) +[](https://www.amazon.com/gp/product/0262527391/ref=as_li_tl?ie=UTF8&camp=1789&creative=9325&creativeASIN=0262527391&linkCode=as2&linkId=52cc270bc3a798b37c787740d00c19d1) ## Parallel Programming in C with MPI and OpenMP -This book is a bit older than the others, but it is still a classic. One strong point of this book is the huge amount of parallel programming examples, along with its focus on MPI and OpenMP. Many parallel programs are discussed in great detail, including matrix multiplication, fast fourier transforms, sorting, and combinatorial searching. [Click here for more info](http://www.amazon.com/gp/product/0071232656/ref=as_li_tf_il?ie=UTF8&tag=softengiintet-20&linkCode=as2&camp=217145&creative=399377&creativeASIN=0071232656). +This book is a bit older than the others, but it is still a classic. One strong point of this book is the huge amount of parallel programming examples, along with its focus on MPI and OpenMP. Many parallel programs are discussed in great detail, including matrix multiplication, fast fourier transforms, sorting, and combinatorial searching. [Click here for more info](http://www.amazon.com/gp/product/0071232656/ref=as_li_tf_il?ie=UTF8&linkCode=as2&camp=217145&creative=399377&creativeASIN=0071232656). -[](http://www.amazon.com/gp/product/0071232656/ref=as_li_tf_il?ie=UTF8&tag=softengiintet-20&linkCode=as2&camp=217145&creative=399377&creativeASIN=0071232656) +[](http://www.amazon.com/gp/product/0071232656/ref=as_li_tf_il?ie=UTF8&linkCode=as2&camp=217145&creative=399377&creativeASIN=0071232656) ## MPI: The Complete Reference -A complete reference guide to MPI one and two. The book does not necessarily teach MPI, but it provides a great reference and complete descriptions of every single function. One advantage of this book is that it includes Fortran routines as well as C routines. [Click here for more info](http://www.amazon.com/gp/product/0262692163/ref=as_li_tf_il?ie=UTF8&tag=softengiintet-20&linkCode=as2&camp=217145&creative=399377&creativeASIN=0262692163). +A complete reference guide to MPI one and two. The book does not necessarily teach MPI, but it provides a great reference and complete descriptions of every single function. One advantage of this book is that it includes Fortran routines as well as C routines. [Click here for more info](http://www.amazon.com/gp/product/0262692163/ref=as_li_tf_il?ie=UTF8&linkCode=as2&camp=217145&creative=399377&creativeASIN=0262692163). + +[](http://www.amazon.com/gp/product/0262692163/ref=as_li_tf_il?ie=UTF8&linkCode=as2&camp=217145&creative=399377&creativeASIN=0262692163) + + + +## Beginning MPI (An Introduction in C) +This book is a compilation of all of the beginner tutorials on this site. It goes over everything from installing MPI on an Amazon EC2 cluster to the basics of sending and receiving with MPI to performing collective operations and reductions. If you have enjoyed the tutorials on this site and wish to have a copy of it in book format, then [click here for more info](http://www.amazon.com/gp/product/B00HM7O0M8/ref=as_li_qf_sp_asin_il_tl?ie=UTF8&camp=1789&creative=9325&creativeASIN=B00HM7O0M8&linkCode=as2). -[](http://www.amazon.com/gp/product/0262692163/ref=as_li_tf_il?ie=UTF8&tag=softengiintet-20&linkCode=as2&camp=217145&creative=399377&creativeASIN=0262692163) +[](http://www.amazon.com/gp/product/B00HM7O0M8/ref=as_li_qf_sp_asin_il_tl?ie=UTF8&camp=1789&creative=9325&creativeASIN=B00HM7O0M8&linkCode=as2) - \ No newline at end of file diff --git a/tutorials.md b/tutorials.md new file mode 100644 index 0000000..a80dd8d --- /dev/null +++ b/tutorials.md @@ -0,0 +1,56 @@ +--- +layout: page +title: Tutorials +redirect_from: '/beginner-mpi-tutorial/' +--- + +Welcome to the MPI tutorials! In these tutorials, you will learn a wide array of concepts about MPI. Below are the available lessons, each of which contain example code. + +The tutorials assume that the reader has a basic knowledge of C, some C++, and Linux. + +## Introduction and MPI installation +* [MPI tutorial introduction]({{ site.baseurl }}/tutorials/mpi-introduction/) +([中文版]({{ site.baseurl }}/tutorials/mpi-introduction/zh_cn)) +([日本語]({{ site.baseurl }}/tutorials/mpi-introduction/ja_jp)) +* [Installing MPICH2 on a single machine]({{ site.baseurl }}/tutorials/installing-mpich2/) +([中文版]({{ site.baseurl }}/tutorials/installing-mpich2/zh_cn)) +([日本語]({{ site.baseurl }}/tutorials/installing-mpich2/ja_jp)) +* [Launching an Amazon EC2 MPI cluster]({{ site.baseurl }}/tutorials/launching-an-amazon-ec2-mpi-cluster/) +([日本語]({{ site.baseurl }}/tutorials/launching-an-amazon-ec2-mpi-cluster/ja_jp)) +* [Running an MPI cluster within a LAN]({{ site.baseurl }}/tutorials/running-an-mpi-cluster-within-a-lan/) +([日本語]({{ site.baseurl }}/tutorials/running-an-mpi-cluster-within-a-lan/ja_jp)) +* [Running an MPI hello world application]({{ site.baseurl }}/tutorials/mpi-hello-world/) +([中文版]({{ site.baseurl }}/tutorials/mpi-hello-world/zh_cn)) +([日本語]({{ site.baseurl }}/tutorials/mpi-hello-world/ja_jp)) + +## Blocking point-to-point communication +* [Sending and receiving with MPI_Send and MPI_Recv]({{ site.baseurl }}/tutorials/mpi-send-and-receive/) +([中文版]({{ site.baseurl }}/tutorials/mpi-send-and-receive/zh_cn)) +([日本語]({{ site.baseurl }}/tutorials/mpi-send-and-receive/ja_jp)) +* [Dynamic receiving with MPI_Probe and MPI_Status]({{ site.baseurl }}/tutorials/dynamic-receiving-with-mpi-probe-and-mpi-status/) +([中文版]({{ site.baseurl }}/tutorials/dynamic-receiving-with-mpi-probe-and-mpi-status/zh_cn)) +([日本語]({{ site.baseurl }}/tutorials/dynamic-receiving-with-mpi-probe-and-mpi-status/ja_jp)) +* [Point-to-point communication application - Random walking]({{ site.baseurl }}/tutorials/point-to-point-communication-application-random-walk/) +([中文版]({{ site.baseurl }}/tutorials/point-to-point-communication-application-random-walk/zh_cn)) +([日本語]({{ site.baseurl }}/tutorials/point-to-point-communication-application-random-walk/ja_jp)) + +## Basic collective communication +* [Collective communication introduction with MPI_Bcast]({{ site.baseurl }}/tutorials/mpi-broadcast-and-collective-communication/) +([中文版]({{ site.baseurl }}/tutorials/mpi-broadcast-and-collective-communication/zh_cn)) +([日本語]({{ site.baseurl }}/tutorials/mpi-broadcast-and-collective-communication/ja_jp)) +* [Common collectives - MPI_Scatter, MPI_Gather, and MPI_Allgather]({{ site.baseurl }}/tutorials/mpi-scatter-gather-and-allgather/) +([中文版]({{ site.baseurl }}/tutorials/mpi-scatter-gather-and-allgather/zh_cn)) +([日本語]({{ site.baseurl }}/tutorials/mpi-scatter-gather-and-allgather/ja_jp)) +* [Application example - Performing parallel rank computation with basic collectives]({{ site.baseurl }}/tutorials/performing-parallel-rank-with-mpi/) +([中文版]({{ site.baseurl }}/tutorials/performing-parallel-rank-with-mpi/zh_cn)) +([日本語]({{ site.baseurl }}/tutorials/performing-parallel-rank-with-mpi/ja_jp)) + +## Advanced collective communication +* [Using MPI_Reduce and MPI_Allreduce for parallel number reduction]({{ site.baseurl }}/tutorials/mpi-reduce-and-allreduce/) +([中文版]({{ site.baseurl }}/tutorials/mpi-reduce-and-allreduce/zh_cn)) +([日本語]({{ site.baseurl }}/tutorials/mpi-reduce-and-allreduce/ja_jp)) + +## Groups and communicators +* [Introduction to groups and communicators]({{ site.baseurl }}/tutorials/introduction-to-groups-and-communicators/) +([中文版]({{ site.baseurl }}/tutorials/introduction-to-groups-and-communicators/zh_cn)) +([日本語]({{ site.baseurl }}/tutorials/introduction-to-groups-and-communicators/ja_jp)) diff --git a/tutorials/dynamic-receiving-with-mpi-probe-and-mpi-status/code/check_status.c b/tutorials/dynamic-receiving-with-mpi-probe-and-mpi-status/code/check_status.c index b051601..ebd157b 100644 --- a/tutorials/dynamic-receiving-with-mpi-probe-and-mpi-status/code/check_status.c +++ b/tutorials/dynamic-receiving-with-mpi-probe-and-mpi-status/code/check_status.c @@ -2,7 +2,7 @@ // Copyright 2011 www.mpitutorial.com // This code is provided freely with the tutorials on mpitutorial.com. Feel // free to modify it for your own use. Any distribution of the code must -// either provide a link to www.mpitutorial.com or keep this header in tact. +// either provide a link to www.mpitutorial.com or keep this header intact. // // Example of checking the MPI_Status object from an MPI_Recv call // @@ -27,7 +27,7 @@ int main(int argc, char** argv) { int numbers[MAX_NUMBERS]; int number_amount; if (world_rank == 0) { - // Pick a random amont of integers to send to process one + // Pick a random amount of integers to send to process one srand(time(NULL)); number_amount = (rand() / (float)RAND_MAX) * MAX_NUMBERS; // Send the amount of integers to process one diff --git a/tutorials/dynamic-receiving-with-mpi-probe-and-mpi-status/code/probe.c b/tutorials/dynamic-receiving-with-mpi-probe-and-mpi-status/code/probe.c index 7adb9ce..e994b77 100644 --- a/tutorials/dynamic-receiving-with-mpi-probe-and-mpi-status/code/probe.c +++ b/tutorials/dynamic-receiving-with-mpi-probe-and-mpi-status/code/probe.c @@ -2,7 +2,7 @@ // Copyright 2011 www.mpitutorial.com // This code is provided freely with the tutorials on mpitutorial.com. Feel // free to modify it for your own use. Any distribution of the code must -// either provide a link to www.mpitutorial.com or keep this header in tact. +// either provide a link to www.mpitutorial.com or keep this header intact. // // Example of using MPI_Probe to dynamically allocated received messages // diff --git a/tutorials/dynamic-receiving-with-mpi-probe-and-mpi-status/index.md b/tutorials/dynamic-receiving-with-mpi-probe-and-mpi-status/index.md index 9c9c87b..82af8aa 100644 --- a/tutorials/dynamic-receiving-with-mpi-probe-and-mpi-status/index.md +++ b/tutorials/dynamic-receiving-with-mpi-probe-and-mpi-status/index.md @@ -4,12 +4,13 @@ title: Dynamic Receiving with MPI Probe (and MPI Status) author: Wes Kendall categories: Beginner MPI tags: MPI_Get_count, MPI_Probe +translations: zh_cn,ja_jp redirect_from: '/dynamic-receiving-with-mpi-probe-and-mpi-status/' --- In the [previous lesson]({{ site.baseurl }}/tutorials/mpi-send-and-receive/), I discussed how to use MPI_Send and MPI_Recv to perform standard point-to-point communication. I only covered how to send messages in which the length of the message was known beforehand. Although it is possible to send the length of the message as a separate send / recv operation, MPI natively supports dynamic messages with just a few additional function calls. I will be going over how to use these functions in this lesson. -> **Note** - All of the code for this site is on [Gitub]({{ site.github.repo }}). This tutorial's code is under [tutorials/dynamic-receiving-with-mpi-probe-and-mpi-status/code]({{ site.github.code }}/tutorials/dynamic-receiving-with-mpi-probe-and-mpi-status/code). +> **Note** - All of the code for this site is on [GitHub]({{ site.github.repo }}). This tutorial's code is under [tutorials/dynamic-receiving-with-mpi-probe-and-mpi-status/code]({{ site.github.code }}/tutorials/dynamic-receiving-with-mpi-probe-and-mpi-status/code). ## The MPI_Status structure As covered in the [previous lesson]({{ site.baseurl }}/tutorials/mpi-send-and-receive/), the `MPI_Recv` operation takes the address of an `MPI_Status` structure as an argument (which can be ignored with `MPI_STATUS_IGNORE`). If we pass an `MPI_Status` structure to the `MPI_Recv` function, it will be populated with additional information about the receive operation after it completes. The three primary pieces of information include: @@ -37,7 +38,7 @@ const int MAX_NUMBERS = 100; int numbers[MAX_NUMBERS]; int number_amount; if (world_rank == 0) { - // Pick a random amont of integers to send to process one + // Pick a random amount of integers to send to process one srand(time(NULL)); number_amount = (rand() / (float)RAND_MAX) * MAX_NUMBERS; @@ -133,9 +134,9 @@ mpirun -n 2 ./probe 1 dynamically received 93 numbers from 0 ``` -Although this example is trivial, `MPI_Probe` forms the basis of many dynamic MPI applications. For example, master/slave programs will often make heavy use of `MPI_Probe` when exchanging variable-sized worker messages. As an exercise, make a wrapper around `MPI_Recv` that uses `MPI_Probe` for any dynamic applications you might write. It makes the code look much nicer :-) +Although this example is trivial, `MPI_Probe` forms the basis of many dynamic MPI applications. For example, manager/worker programs will often make heavy use of `MPI_Probe` when exchanging variable-sized worker messages. As an exercise, make a wrapper around `MPI_Recv` that uses `MPI_Probe` for any dynamic applications you might write. It makes the code look much nicer :-) ## Up next Do you feel comfortable using the standard blocking point-to-point communication routines? If so, then you already have the ability to write endless amounts of parallel applications! Let's look at a more advanced example of using the routines you have learned. Check out [the application example using MPI_Send, MPI_Recv, and MPI_Probe]({{ site.baseurl }}/tutorials/point-to-point-communication-application-random-walk/). -Having trouble? Confused? Feel free to leave a comment below and perhaps I or another reader can be of help. \ No newline at end of file +Having trouble? Confused? Feel free to leave a comment below and perhaps I or another reader can be of help. diff --git a/tutorials/dynamic-receiving-with-mpi-probe-and-mpi-status/ja_jp.md b/tutorials/dynamic-receiving-with-mpi-probe-and-mpi-status/ja_jp.md new file mode 100644 index 0000000..e551213 --- /dev/null +++ b/tutorials/dynamic-receiving-with-mpi-probe-and-mpi-status/ja_jp.md @@ -0,0 +1,141 @@ +--- +layout: post +title: MPI Probeを用いた可変長メッセージの受信 - Dynamic Receiving with MPI Probe (and MPI Status) +author: Wes Kendall +categories: Beginner MPI +tags: MPI_Get_count, MPI_Probe +redirect_from: '/dynamic-receiving-with-mpi-probe-and-mpi-status/' +--- + +[前回のレッスン]({{ site.baseurl }}/tutorials/mpi-send-and-receive/)ではMPI_SendとMPI_Recvを使用した基本的なポイントツーポイント通信を学びました。前回はメッセージ長が固定である場合のみを説明しました。可変長メッセージを送る方法として１つ目のメッセージ長をsend/recvするという方法もとれます。しかし、MPIには追加の関数呼び出しだけで可変長のメッセージをサポートすることが可能です。このレッスンではこの方法を学びます。 + +> **Note** - このサイトのコードはすべて[GitHub]({{ site.github.repo }})にあります。このレッスンのコードは[tutorials/dynamic-receiving-with-mpi-probe-and-mpi-status/code]({{ site.github.code }}/tutorials/dynamic-receiving-with-mpi-probe-and-mpi-status/code)にあります。 + +## MPI_Status構造体 - The MPI_Status structure +[前回のレッスン]({{ site.baseurl }}/tutorials/mpi-send-and-receive/)で説明したように、`MPI_Recv`は受信したメッセージに関する情報を`MPI_Status`として受け取ります(`MPI_STATUS_IGNORE`で無視することもできます)。`MPI_Recv`関数に`MPI_Status`を渡していると受信操作が完了した後に次の3つの追加情報を得ることができます。 + +1. **送信者のランク**: 送信者のランクは`MPI_SOURCE`構造体に格納されます。`MPI_Status stat`と宣言すると、ランクには`stat.MPI_SOURCE`でアクセスできます。 +2. **メッセージタグ**: メッセージのタグは`MPI_TAG`に格納されます。`MPI_SOURCE`と同じようにアクセスできます。 +3. **メッセージ長**: メッセージ長は、ステータス構造内には含まれません。そこで`MPI_Get_count`を使用してメッセージの長さを調べる必要があります。 + +```cpp +MPI_Get_count( + MPI_Status* status, + MPI_Datatype datatype, + int* count) +``` + +`MPI_Get_count`関数に`MPI_Status`渡すとメッセージの`datatype`と`count`が返されます。`count`には受信した要素の合計数が入ります。 + +なぜこの２つの情報が必要になるのかを説明します。`MPI_Recv`受信のために`MPI_ANY_SOURCE`を送信者のランクに、`MPI_ANY_TAG`をメッセージのタグとして受信動作を行えます。この場合には`MPI_Status`が送信者とメッセージタグを知る唯一の手がかりになります（訳注：rankとtagを指定してrecvしていない場合、という意味です。）。また、`MPI_Recv`関数は引数として指定した要素数を全て受信することが保証されていないことに注意します。受信した要素数を得ることができます。しかしながら、指定した受信できる数以上の要素が送信された場合はエラーになることに注意してください。`MPI_Get_count`は実際に受信する量を決定するのに使われます。 + +## MPI_Status構造体をクエリする例 - An example of querying the MPI_Status structure + +`MPI_Status`構造体をクエリするプログラム[check_status.c]({{ site.github.code }}/tutorials/dynamic-receiving-with-mpi-probe-and-mpi-status/code/check_status.c)を見ていきます。プログラムは数字を適当な個数送信します。受信者は送信された数字の数を調べます。コードのメイン部分は次のようになります。 + +```cpp +const int MAX_NUMBERS = 100; +int numbers[MAX_NUMBERS]; +int number_amount; +if (world_rank == 0) { + // プロセス1に送る数字の数を決定する + srand(time(NULL)); + number_amount = (rand() / (float)RAND_MAX) * MAX_NUMBERS; + + // その分の数をプロセス1に送る + MPI_Send(numbers, number_amount, MPI_INT, 1, 0, MPI_COMM_WORLD); + printf("0 sent %d numbers to 1\n", number_amount); +} else if (world_rank == 1) { + MPI_Status status; + // 最大でMAX_NUMBERS個のMPI_INTをプロセス0から受け取る + MPI_Recv(numbers, MAX_NUMBERS, MPI_INT, 0, 0, MPI_COMM_WORLD, + &status); + + // メッセージを受信した後、そのメッセージにいくつの整数が含まれていたかを + // 取得する + MPI_Get_count(&status, MPI_INT, &number_amount); + + // 含まれていた数字の数、ランク、タグを出力する。 + printf("1 received %d numbers from 0. Message source = %d, " + "tag = %d\n", + number_amount, status.MPI_SOURCE, status.MPI_TAG); +} +``` + +プロセス0は最大で`MAX_NUMBERS`個の整数をランダムに決めてプロセス1に送信します。プロセス1は最大で`MAX_NUMBERS`個の整数を読み込む`MPI_Recv`を実行します。プロセス1は`MPI_Recv` の引数として `MAX_NUMBERS` を渡しています。繰り返しになりますが、プロセス1が受け取ることができるのは**最大**この個数であることに注意してください。プロセス1では`MPI_Get_count`を呼び出して、実際に受信した`MPI_INT`の数をを調べます。プロセス1は受信したメッセージのサイズを表示すると同時に、`MPI_SOURCE`と`MPI_TAG`、つまり送信元ランクとタグも表示します。 + +ここで注意があります。`MPI_Get_count`で得られるのはデータ型の数でバイト数でないことです。もしユーザーが`MPI_CHAR`をデータ型として(recvを)使用したとすると、返されるデータ量は4倍になります(intを4バイト、charを1バイトと仮定した場合)。このプログラムを[レポジトリ]({{ site.github.code }})の*tutorials*ディレクトリから実行すると、出力はこのようになるでしょう。 + +``` +>>> cd tutorials +>>> ./run.py check_status +mpirun -n 2 ./check_status +0 sent 92 numbers to 1 +1 received 92 numbers from 0. Message source = 0, tag = 0 +``` + +プロセス0はプロセス1にランダムな数の整数を送信して、プロセス1は受信したメッセージに関する情報を出力できました。 + +## MPI_Probeを使用してメッセージサイズを調べる - Using MPI_Probe to find out the message size +`MPI_Status`への理解が深まってきました。前回の例では受信前にすべてのサイズのメッセージを処理できるように大きなバッファを用意しました。実は、`MPI_Probe`を使用すると受信前にメッセージ長を調べることができます。 + +```cpp +MPI_Probe( + int source, + int tag, + MPI_Comm comm, + MPI_Status* status) +``` + +`MPI_Probe`は`MPI_Recv`と似た呼び出しです。`MPI_Probe`は`MPI_Recv`で実際に受信する以外の処理を行うと考えて良いです。`MPI_Recv`と同様に`MPI_Probe`は指定したタグかつ指定した送信者ランクであるメッセージが来るまでブロッキングします。メッセージを受信するとstatus構造体に情報が格納されます。その後にユーザは(そのデータを受信するのに十分なバッファを用意して)`MPI_Recv`で実際にメッセージを受信すれば良いのです。 + +[レポジトリ]({{ site.github.code }}/tutorials/dynamic-receiving-with-mpi-probe-and-mpi-status/code/code)の[probe.c]({{ site.github.code }}/tutorials/dynamic-receiving-with-mpi-probe-and-mpi-status/code/probe.c)にこの例を示します。 + +```cpp +int number_amount; +if (world_rank == 0) { + const int MAX_NUMBERS = 100; + int numbers[MAX_NUMBERS]; + // プロセス1に送る数字の数を決定する + srand(time(NULL)); + number_amount = (rand() / (float)RAND_MAX) * MAX_NUMBERS; + + // その分の数をプロセス1に送る + MPI_Send(numbers, number_amount, MPI_INT, 1, 0, MPI_COMM_WORLD); + printf("0 sent %d numbers to 1\n", number_amount); +} else if (world_rank == 1) { + MPI_Status status; + // プロセス0からのメッセージを"probe"する + MPI_Probe(0, 0, MPI_COMM_WORLD, &status); + + // probeが完了した時、statusにはメッセージ長などの情報が含まれている + // MPI_Get_countを使ってメッセージ長を得る + MPI_Get_count(&status, MPI_INT, &number_amount); + + // probeで判明したメッセージ長文のメモリを確保する + int* number_buf = (int*)malloc(sizeof(int) * number_amount); + + // そのバッファを使用してrecv処理を行う + MPI_Recv(number_buf, number_amount, MPI_INT, 0, 0, + MPI_COMM_WORLD, MPI_STATUS_IGNORE); + printf("1 dynamically received %d numbers from 0.\n", + number_amount); + free(number_buf); +} +``` + +先程の例と同様、プロセス0はプロセス1に送信する変数の数をランダムに決定します。変更点は、プロセス1が `MPI_Probe`を呼び出してプロセス0が送信しようとしている要素の数を(`MPI_Get_count`を使用して)得ることです。この後、プロセス1が適切なサイズのバッファを確保してから数値を受信します。 + +``` +>>> ./run.py probe +mpirun -n 2 ./probe +0 sent 93 numbers to 1 +1 dynamically received 93 numbers from 0 +``` + +ここで示した例はシンプルなものです。`MPI_Probe`は多くのMPIアプリケーションで用いられる基本的な機能です。例えばマネージャ/ワーカプログラムは、可変サイズのメッセージを交換する際に`MPI_Probe`を多用します。練習として、`MPI_Probe`を使用した`MPI_Recv`のラッパーを作成し、動的なアプリケーションを書いてみてください。コードの見栄えが格段に良くなるでしょう :-) + +## Up next +さて、標準的なブロック型ポイント・ツー・ポイント通信に抵抗はなくなってきましたか？そうなら、あなたはすでにいくらでも並列アプリケーションを書く能力を持っています！では、学んだルーチンを使ったより高度な例を見てみましょう。[the application example using MPI_Send, MPI_Recv, and MPI_Probe]({{ site.baseurl }}/tutorials/point-to-point-communication-application-random-walk/) をチェックしてください。 + +困っていますか?混乱していますか?お気軽に下記にコメントを残してください。私や他の読者がお役に立てるかもしれません。 diff --git a/tutorials/dynamic-receiving-with-mpi-probe-and-mpi-status/zh_cn.md b/tutorials/dynamic-receiving-with-mpi-probe-and-mpi-status/zh_cn.md new file mode 100644 index 0000000..f289d30 --- /dev/null +++ b/tutorials/dynamic-receiving-with-mpi-probe-and-mpi-status/zh_cn.md @@ -0,0 +1,181 @@ +--- +layout: post +title: Dynamic Receiving with MPI Probe (and MPI Status) +author: Wes Kendall +categories: Beginner MPI +tags: MPI_Get_count, MPI_Probe +redirect_from: '/dynamic-receiving-with-mpi-probe-and-mpi-status/' +--- + +在 [上一节]({{ site.baseurl }}/tutorials/mpi-send-and-receive/) 中，讨论了如何使用 `MPI_Send` 和 `MPI_Recv` 执行标准的点对点通信。 +但仅仅只介绍了如何发送事先知道消息长度的消息。 +尽管可以将消息的长度作为单独的发送/接收操作发送，但是 MPI 本身仅通过几个额外的函数调用即可支持动态消息。 +在本节中，将讨论如何使用这些功能。 + +> **注意** - 教程所涉及的所有代码都在 [GitHub 上]({{ site.github.repo }})。本文的代码在 [tutorials/dynamic-receiving-with-mpi-probe-and-mpi-status/code]({{ site.github.code }}/tutorials/dynamic-receiving-with-mpi-probe-and-mpi-status/code) 下。 + +## MPI_Status 结构体 + +如 [上节]({{ site.baseurl }}/tutorials/mpi-send-and-receive/) 所述，`MPI_Recv` 将 `MPI_Status` 结构体的地址作为参数（可以使用 `MPI_STATUS_IGNORE` 忽略）。 +如果我们将 `MPI_Status` 结构体传递给 `MPI_Recv` 函数，则操作完成后将在该结构体中填充有关接收操作的其他信息。 +三个主要的信息包括： + +1. **发送端秩**. 发送端的秩存储在结构体的 `MPI_SOURCE` 元素中。也就是说，如果我们声明一个 `MPI_Status stat` 变量，则可以通过 `stat.MPI_SOURCE` 访问秩。 +2. **消息的标签**. 消息的标签可以通过结构体的 `MPI_TAG` 元素访问（类似于 `MPI_SOURCE`）。 +3. **消息的长度**. 消息的长度在结构体中没有预定义的元素。相反，我们必须使用 `MPI_Get_count` 找出消息的长度。 + +```cpp +MPI_Get_count( + MPI_Status* status, + MPI_Datatype datatype, + int* count) +``` + +在 `MPI_Get_count` 函数中，使用者需要传递 `MPI_Status` 结构体，消息的 `datatype`（数据类型），并返回 `count`。 +变量 `count` 是已接收的 `datatype` 元素的数目。 + +为什么需要这些信息？ +事实证明，`MPI_Recv` 可以将 `MPI_ANY_SOURCE` 用作发送端的秩，将 `MPI_ANY_TAG` 用作消息的标签。 +在这种情况下，`MPI_Status` 结构体是找出消息的实际发送端和标签的唯一方法。 +此外，并不能保证 `MPI_Recv` 能够接收函数调用参数的全部元素。 +相反，它只接收已发送给它的元素数量（如果发送的元素多于所需的接收数量，则返回错误。） +`MPI_Get_count` 函数用于确定实际的接收量。 + +## `MPI_Status` 结构体查询的示例 + +查询 `MPI_Status` 结构体的程序在 [check_status.c]({{ site.github.code }}/tutorials/dynamic-receiving-with-mpi-probe-and-mpi-status/code/check_status.c) 中。 +程序将随机数量的数字发送给接收端，然后接收端找出发送了多少个数字。 +代码的主要部分如下所示。 + +```cpp +const int MAX_NUMBERS = 100; +int numbers[MAX_NUMBERS]; +int number_amount; +if (world_rank == 0) { + // Pick a random amount of integers to send to process one + srand(time(NULL)); + number_amount = (rand() / (float)RAND_MAX) * MAX_NUMBERS; + + // Send the amount of integers to process one + MPI_Send(numbers, number_amount, MPI_INT, 1, 0, MPI_COMM_WORLD); + printf("0 sent %d numbers to 1\n", number_amount); +} else if (world_rank == 1) { + MPI_Status status; + // Receive at most MAX_NUMBERS from process zero + MPI_Recv(numbers, MAX_NUMBERS, MPI_INT, 0, 0, MPI_COMM_WORLD, + &status); + + // After receiving the message, check the status to determine + // how many numbers were actually received + MPI_Get_count(&status, MPI_INT, &number_amount); + + // Print off the amount of numbers, and also print additional + // information in the status object + printf("1 received %d numbers from 0. Message source = %d, " + "tag = %d\n", + number_amount, status.MPI_SOURCE, status.MPI_TAG); +} +``` + +如我们所见，进程 0 将最多 `MAX_NUMBERS` 个整数以随机数量发送到进程 1。 +进程 1 然后调用 `MPI_Recv` 以获取总计 `MAX_NUMBERS` 个整数。 +尽管进程 1 以 `MAX_NUMBERS` 作为 `MPI_Recv` 函数参数，但进程 1 将最多接收到此数量的数字。 +在代码中，进程 1 使用 `MPI_INT` 作为数据类型的参数，调用 `MPI_Get_count`，以找出实际接收了多少个整数。 +除了打印出接收到的消息的大小外，进程 1 还通过访问 status 结构体的 `MPI_SOURCE` 和 `MPI_TAG` 元素来打印消息的来源和标签。 + +为了澄清起见，`MPI_Get_count` 的返回值是相对于传递的数据类型而言的。 +如果用户使用 `MPI_CHAR` 作为数据类型，则返回的数量将是原来的四倍（假设整数是四个字节，而 char 是一个字节）。 +如果你从 [库]({{ site.github.code }}) 的 *tutorials* 目录中运行 check_status 程序，则输出应类似于： + +``` +>>> cd tutorials +>>> ./run.py check_status +mpirun -n 2 ./check_status +0 sent 92 numbers to 1 +1 received 92 numbers from 0. Message source = 0, tag = 0 +``` + +正如预期的那样，进程 0 将随机数目的整数发送给进程 1，进程 1 将打印出接收到的消息的有关信息。 + +## 使用 `MPI_Probe` 找出消息大小 + +现在您了解了 `MPI_Status` 的工作原理，现在我们可以使用它来发挥更高级的优势。 +除了传递接收消息并简易地配备一个很大的缓冲区来为所有可能的大小的消息提供处理（就像我们在上一个示例中所做的那样），您可以使用 `MPI_Probe` 在实际接收消息之前查询消息大小。 +函数原型看起来像这样： + +```cpp +MPI_Probe( + int source, + int tag, + MPI_Comm comm, + MPI_Status* status) +``` + +`MPI_Probe` 看起来与 `MPI_Recv` 非常相似。 +实际上，您可以将 `MPI_Probe` 视为 `MPI_Recv`，除了不接收消息外，它们执行相同的功能。 +与 `MPI_Recv` 类似，`MPI_Probe` 将阻塞具有匹配标签和发送端的消息。 +当消息可用时，它将填充 status 结构体。 +然后，用户可以使用 `MPI_Recv` 接收实际的消息。 + +[教程代码]({{ site.github.code }}/tutorials/dynamic-receiving-with-mpi-probe-and-mpi-status/code) 在 probe.c 中有一个示例。 +以下是源代码的主要部分 + +```cpp +int number_amount; +if (world_rank == 0) { + const int MAX_NUMBERS = 100; + int numbers[MAX_NUMBERS]; + // Pick a random amount of integers to send to process one + srand(time(NULL)); + number_amount = (rand() / (float)RAND_MAX) * MAX_NUMBERS; + + // Send the random amount of integers to process one + MPI_Send(numbers, number_amount, MPI_INT, 1, 0, MPI_COMM_WORLD); + printf("0 sent %d numbers to 1\n", number_amount); +} else if (world_rank == 1) { + MPI_Status status; + // Probe for an incoming message from process zero + MPI_Probe(0, 0, MPI_COMM_WORLD, &status); + + // When probe returns, the status object has the size and other + // attributes of the incoming message. Get the message size + MPI_Get_count(&status, MPI_INT, &number_amount); + + // Allocate a buffer to hold the incoming numbers + int* number_buf = (int*)malloc(sizeof(int) * number_amount); + + // Now receive the message with the allocated buffer + MPI_Recv(number_buf, number_amount, MPI_INT, 0, 0, + MPI_COMM_WORLD, MPI_STATUS_IGNORE); + printf("1 dynamically received %d numbers from 0.\n", + number_amount); + free(number_buf); +} +``` + +与上一个示例类似，进程 0 选择随机数量的数字发送给进程 1。 +不同之处在于，进程 1 现在调用 `MPI_Probe`，以找出进程 0 试图发送多少个元素（利用 `MPI_Get_count`）。 +然后，进程 1 分配适当大小的缓冲区并接收数字。 +执行本示例代码，结果看起来类似于： + +``` +>>> ./run.py probe +mpirun -n 2 ./probe +0 sent 93 numbers to 1 +1 dynamically received 93 numbers from 0 +``` + +尽管这个例子很简单，但是 `MPI_Probe` 构成了许多动态 MPI 应用程序的基础。 +例如，控制端/执行子程序在交换变量大小的消息时通常会大量使用 `MPI_Probe`。 +作为练习，对 `MPI_Recv` 进行包装，将 `MPI_Probe` 用于您可能编写的任何动态应用程序。 +它将使代码看起来更美好：-) + +## 接下来 + +对于使用标准的阻塞点对点通信例程的理解是否清晰？ +如果是的，那么您已经有能力编写无数的并行应用程序！ +让我们来看一个使用所学例程的高级示例。 +查看 [使用 `MPI_Send`，`MPI_Recv` 和 `MPI_Probe` 的应用程序示例]({{ site.baseurl }}/tutorials/point-to-point-communication-application-random-walk/). + +遇到麻烦？ 困惑？ +随时在下面发表评论，也许我或其他读者会有所帮助。 diff --git a/tutorials/installing-mpich2/index.md b/tutorials/installing-mpich2/index.md index a7cd9f5..92e032e 100644 --- a/tutorials/installing-mpich2/index.md +++ b/tutorials/installing-mpich2/index.md @@ -4,53 +4,54 @@ title: Installing MPICH2 on a Single Machine author: Wes Kendall categories: Beginner MPI tags: +translations: zh_cn,ja_jp redirect_from: '/installing-mpich2/' --- -MPI is simply a standard which others follow in their implementation. Because of this, there are a wide variety of MPI implementations out there. One of the most popular implementations, MPICH2, will be used for all of the examples provided through this site. Users are free to use any implementation they wish, but only instructions for installing MPICH2 will be provided. Furthermore, the scripts and code provided for the lessons are only guaranteed to execute and run with the lastest version of MPICH2. +MPI is simply a standard which others follow in their implementation. Because of this, there are a wide variety of MPI implementations out there. One of the most popular implementations, MPICH, will be used for all of the examples provided through this site. Users are free to use any implementation they wish, but only instructions for installing MPICH will be provided. Furthermore, the scripts and code provided for the lessons are only guaranteed to execute and run with the lastest version of MPICH. -MPICH2 is a widely-used implementation of MPI that is developed primarily by Argonne National Laboratory in the United States. The main reason for choosing MPICH2 over other implementations is simply because of my familiarity with the interface and because of my close relationship with Argonne National Laboratory. I also encourage others to check out OpenMPI, which is also a widely-used implementation. +MPICH is a widely-used implementation of MPI that is developed primarily by Argonne National Laboratory in the United States. The main reason for choosing MPICH over other implementations is simply because of my familiarity with the interface and because of my close relationship with Argonne National Laboratory. I also encourage others to check out [OpenMPI](https://www.open-mpi.org/), which is also a widely-used implementation. -## Installing MPICH2 -The latest version of MPICH2 is available [here](http://www.mcs.anl.gov/research/projects/mpich2/). The version that I will be using for all of the examples on the site is 1.4, which was released June 16, 2011. Go ahead and download the source code, uncompress the folder, and change into the MPICH2 directory. +## Installing MPICH +The latest version of MPICH is available [here](https://www.mpich.org/). The version that I will be using for all of the examples on the site is 3.3-2, which was released 13 November 2019. Go ahead and download the source code, uncompress the folder, and change into the MPICH directory. ``` ->>> tar -xzf mpich2-1.4.tar.gz ->>> cd mpich2-1.4 +>>> tar -xzf mpich-3-3.2.tar.gz +>>> cd mpich-3-3.2 ``` -Once doing this, you should be able to configure your installation by performing `./configure`. I added a couple of parameters to my configuration to avoid building the MPI Fortran library. If you need to install MPICH2 to a local directory (for example, if you don't have root access to your machine), type `./configure --prefix=/installation/directory/path` For more information about possible configuration parameters, type `./configure --help` +Once doing this, you should be able to configure your installation by performing `./configure`. If you need to install MPICH to a local directory (for example, if you don't have root access to your machine), type `./configure --prefix=/installation/directory/path`. It is possible to avoid building the MPI Fortran library by using `./configure --disable-fortran` if you do not have Fortran compilers. For more information about possible configuration parameters, type `./configure --help` ``` ->>> ./configure --disable-f77 --disable-fc -Configuring MPICH2 version 1.4 with '--disable-f77' '--disable-fc' -Running on system: Darwin Wes-Kendalls-Macbook-Pro.local 10.7.0 Darwin Kernel Version 10.7.0: Sat Jan 29 15:17:16 PST 2011; root:xnu1504.9.37~1/RELEASE_I386 i386 -checking for gcc... gcc +>>> ./configure +Configuring MPICH version 3.3.2 +Running on system: Linux localhost.localdomain 5.8.18-100.fc31.x86_64 #1 SMP Mon Nov 2 20:32:55 UTC 2020 x86_64 x86_64 x86_64 GNU/Linux +checking build system type... x86_64-unknown-linux-gnu ``` When configuration is done, it should say *"Configuration completed."* Once this is through, it is time to build and install MPICH2 with `make; sudo make install`. ``` >>> make; sudo make install -Beginning make -Using variables CC='gcc' CFLAGS=' -O2' LDFLAGS=' ' F77='' FFLAGS=' ' FC='' FCFLAGS=' ' CXX='c++' CXXFLAGS=' -O2' AR='ar' CPP='gcc-E' CPP +make +make all-recursive + ``` -If your build was successful, you should be able to type `mpich2version` and see something similar to this. +If your build was successful, you should be able to type `mpiexec --version` and see something similar to this. ``` ->>> mpich2version -MPICH2 Version: 1.4 -MPICH2 Release date: Thu Jun 16 16:41:08 CDT 2011 -MPICH2 Device: ch3:nemesis -MPICH2 configure: --disable-f77 --disable-fc -MPICH2 CC: gcc -O2 -MPICH2 CXX: c++ -O2 -MPICH2 F77: -MPICH2 FC: +>>> mpiexec --version +HYDRA build details: + Version: 3.3.2 + Release Date: Tue Nov 12 21:23:16 CST 2019 + CC: gcc + CXX: g++ + F77: gfortran + F90: gfortran ``` -Hopefully your build finished successfully. If not, you may have issues with missing dependencies. For any issue, I highly recommend copying and pasting the error message directly into Google. +Hopefully your build finished successfully. If not, you may have issues with missing dependencies. For any issue, I highly recommend copying and pasting the error message directly into Google. ## Up next -Now that you have built MPICH2 locally, you have a couple options of where you can proceed on this site. If you don't have access to a cluster or want to learn more about building a virtual MPI cluster, check out the lesson about [building and running your own cluster on Amazon EC2]({{ site.baseurl }}/tutorials/launching-an-amazon-ec2-mpi-cluster/). If you already have your own cluster or simply want to run the rest of the lessons from your machine, proceed to the [MPI hello world lesson]({{ site.baseurl }}/tutorials/mpi-hello-world/), which provides an overview of the basics of programming and running your first MPI program. \ No newline at end of file +Now that you have built MPICH locally, you have some options of where you can proceed on this site. If you already have the hardware and resources to setup a local cluster, I suggest you proceed to the tutorial about [running an MPI cluster in LAN]({{ site.baseurl }}/tutorials/running-an-mpi-cluster-within-a-lan). If you don't have access to a cluster or want to learn more about building a virtual MPI cluster, check out the lesson about [building and running your own cluster on Amazon EC2]({{ site.baseurl }}/tutorials/launching-an-amazon-ec2-mpi-cluster/). If you have built a cluster in either way or simply want to run the rest of the lessons from your machine, proceed to the [MPI hello world lesson]({{ site.baseurl }}/tutorials/mpi-hello-world/), which provides an overview of the basics of programming and running your first MPI program. diff --git a/tutorials/installing-mpich2/ja_jp.md b/tutorials/installing-mpich2/ja_jp.md new file mode 100644 index 0000000..e73efc5 --- /dev/null +++ b/tutorials/installing-mpich2/ja_jp.md @@ -0,0 +1,55 @@ +--- +layout: post +title: MPIをシングルマシンにインストールする - Installing MPICH2 on a Single Machine +author: Wes Kendall +categories: Beginner MPI +tags: +redirect_from: '/installing-mpich2/' +--- + +MPIとは標準仕様を指すものであるのでMPIの実装は複数存在します。このレッスンでは主な実装の１つであるMPICHを使用します。あなたが望むなら他の実装を使用しても良いですが、このレッスンではMPICHインストール手順を紹介します。また、チュートリアル全体で提供されるスクリプトとコードは、最新バージョンのMPICHでの実行だけを確認しています。 + +MPICHは米国のアルゴンヌ国立研究所が主に開発したメジャーなMPI実装です。MPICHを選択した理由は、私がこのインターフェイスに精通しておりアルゴンヌ国立研究所と縁が深いためです。広く使用されている実装である[OpenMPI](https://www.open-mpi.org/)もぜひ調べてみてください。 + +## MPICHのインストール - Installing MPICH +[ここ](https://www.mpich.org/)からMPICHの最新バージョンを入手できます。このチュートリアルでは3.3-2(2019年11月13日リリース)を利用します。tar.gzファイルをダウンロードし以下のように伸長します。 + +``` +>>> tar -xzf mpich-3-3.2.tar.gz +>>> cd mpich-3-3.2 +``` + +`./configure`でmakeの準備をします。この際にマシンの特権がなくユーザディレクトリにインストールしたい場合は`./configure --prefix=/installation/directory/path`としてインストールディレクトリを指定できます。Fortranへの対応が不要である場合は`./configure --disable-fortran`とします。利用可能なオプションを全て表示するには`./configure --help`とします。 + +``` +>>> ./configure +Configuring MPICH version 3.3.2 +Running on system: Linux localhost.localdomain 5.8.18-100.fc31.x86_64 #1 SMP Mon Nov 2 20:32:55 UTC 2020 x86_64 x86_64 x86_64 GNU/Linux +checking build system type... x86_64-unknown-linux-gnu +``` + +configureが終了して*"Configuration completed."*と表示さたら`make; sudo make install`を使用して MPICH2 をビルド・インストールします。 +``` +>>> make; sudo make install +make +make all-recursive + +``` + +成功したなら`mpiexec --version`でインストールした情報を出力できます。 + +``` +>>> mpiexec --version +HYDRA build details: + Version: 3.3.2 + Release Date: Tue Nov 12 21:23:16 CST 2019 + CC: gcc + CXX: g++ + F77: gfortran + F90: gfortran +``` + +皆さんの環境でもビルドが無事に完了することを祈りますが、依存関係不足などで問題が起こる可能性があるでしょう。このような場合はエラーメッセージをGoogleで検索してみてください。 + +## 次は？ +単一の環境にMPICHを構築できたのでこのサイトで次に進むための選択肢がいくつかあります。ローカルクラスタをセットアップするためのハードウェアとリソースがすでにある場合は、[running an MPI cluster in LAN]({{ site.baseurl }}/tutorials/running-an-mpi-cluster-within-a-lan)に進むことを推奨します。そのようなクラスタにアクセスできない場合や仮想 MPIクラスターの構築について詳しく知りたい場合は、[building and running your own cluster on Amazon EC2]({{ site.baseurl }}/tutorials/launching-an-amazon-ec2-mpi-cluster/)に進んでください。いずれかの方法でクラスターを構築し終わっていたり、残りのレッスンをスタンドアロンで実行する場合は[MPI hello world lesson]({{ site.baseurl }}/tutorials/mpi-hello-world/)に進んでください。Hello Worldではプログラミングの基礎と最初のMPIプログラムの実行の概要が説明されています。 \ No newline at end of file diff --git a/tutorials/installing-mpich2/zh_cn.md b/tutorials/installing-mpich2/zh_cn.md new file mode 100644 index 0000000..cb6abac --- /dev/null +++ b/tutorials/installing-mpich2/zh_cn.md @@ -0,0 +1,73 @@ +--- +layout: post +title: Installing MPICH2 on a Single Machine +author: Wes Kendall +categories: Beginner MPI +tags: +redirect_from: '/installing-mpich2/' +--- + +MPI 只是众多实现中所遵循的一个标准。 +因此，这里有各种各样的 MPI 实现。 +本站点提供的所有示例都将使用最受欢迎的实现之一 MPICH。 +用户可以自由使用他们希望的任何实现，但是仅提供安装 MPICH 的说明。 +此外，仅保证为课程提供的脚本和代码可以在最新版本的 MPICH 上执行和运行。 + +MPICH 是 MPI 的一种广泛使用的实现，主要由美国的 Argonne 国家实验室开发。 +选择 MPICH 而不是其他实现的主要原因是由于我对界面的熟悉以及与 Argonne 国家实验室的密切关系。 +我还鼓励其他人使用 OpenMPI，这也是一种广泛使用的实现。 + +## Installing MPICH + +MPICH 的最新版本可在 [此处](https://www.mpich.org/) 获取。 +网站上所有示例的版本是 3.3-2，该版本于 2019 年 9 月 13 日发布。 +下载源代码，解压缩文件夹，然后切换到 MPICH 目录。 + +``` +>>> tar -xzf mpich3-3.2.tar.gz +>>> cd mpich3-3.2 +``` + +完成之后，您应该能够通过执行 `./configure` 来配置安装。 +如果需要将 MPICH 安装到本地目录（例如，如果您没有对计算机的 root 访问权限），请使用 `./configure --prefix=/installation/directory/path`，如果您没有 Fortran 编译器，则可以通过使用 `./configure --disable-fortran` 来避免构建 MPI Fortran 库。输入 `./configure --help` 以获取有关可能的配置参数的更多信息。 + +``` +>>> ./configure --disable-fortran +Configuring MPICH version 3.3.2 +Running on system: Linux localhost.localdomain 5.8.18-100.fc31.x86_64 #1 SMP Mon Nov 2 20:32:55 UTC 2020 x86_64 x86_64 x86_64 GNU/Linux +checking build system type... x86_64-unknown-linux-gnu +``` + +配置完成后，应显示 *"Configuration completed."* +一旦完成，就该使用`make; sudo make install`命令来构建和安装 MPICH2 了。 + +``` +>>> make; sudo make install +make +make all-recursive + +``` + +如果构建成功，则应该可以输入`mpiexec --version`并看到以下类似的内容。 + +``` +>>> mpiexec --version +HYDRA build details: + Version: 3.3.2 + Release Date: Tue Nov 12 21:23:16 CST 2019 + CC: gcc + CXX: g++ + F77: gfortran + F90: gfortran +``` + +希望您的构建成功完成。 +如果没有，您可能会遇到缺少依赖项的问题。 +对于任何问题，我强烈建议您将错误消息直接复制并粘贴到 Google 中。 + +## 下一步 + +现在，您已经在本地构建了 MPICH，您可以在该站点上进行一些选择。 +如果您已经具有用于设置本地集群的硬件和资源，建议您继续阅读有关 [在 LAN 中运行 MPI 集群]({{ site.baseurl }}/tutorials/running-an-mpi-cluster-within-a-lan) 的教程。 +如果您无权访问集群或想了解有关构建虚拟化 MPI 集群的更多信息，请阅读有关 [在 Amazon EC2 上构建和运行自己的集群]({{ site.baseurl }}/tutorials/launching-an-amazon-ec2-mpi-cluster/)。 +如果您以任何一种方式构建了集群，或者只是想从计算机上运行其余课程，请继续阅读 [MPI hello world 课程]({{ site.baseurl }}/tutorials/mpi-hello-world/)，其中概述了编程和运行第一个 MPI 程序的基础知识。 diff --git a/tutorials/introduction-to-groups-and-communicators/code/comm_groups.c b/tutorials/introduction-to-groups-and-communicators/code/comm_groups.c new file mode 100644 index 0000000..e29edf9 --- /dev/null +++ b/tutorials/introduction-to-groups-and-communicators/code/comm_groups.c @@ -0,0 +1,55 @@ +// Author: Wesley Bland +// Copyright 2015 www.mpitutorial.com +// This code is provided freely with the tutorials on mpitutorial.com. Feel +// free to modify it for your own use. Any distribution of the code must +// either provide a link to www.mpitutorial.com or keep this header intact. +// +// Example using MPI_Comm_split to divide a communicator into subcommunicators +// +#include