mpi

问题 I am trying to program an MPI_Alltoallv using an MPI Derived datatype using MPI_Type_create_struct. I could not find any examples solving this particular problem. Most examples like this perform communication(Send/Recv) using a single struct member, whereas I am targeting an array of structs. Following is a simpler test code that attempts a MPI_Sendrecv operation on an array of structs created using DDT: #include <stdio.h> #include <stdlib.h> #include <mpi.h> #include <stddef.h> typedef

问题 Following is a simple program where all Slaves process sends a message to the Master process. The program when executed runs correctly sometimes and raises Segmentation Fault the others. int token; if(rank == 0) { for (int irank = 1; irank < world_size; irank++) { MPI_Recv(&token, 1, MPI_INT, irank, 0, MPI_COMM_WORLD, MPI_STATUS_IGNORE); cout << "Master: Token = " << token << endl; } } if(rank != 0) { token = 1; cout << "Slave: Token = " << token << endl; MPI_Send(&token, 1, MPI_INT, 0, 0,

问题 How do I access a global vector from an individual thread in MPI? I'm using a library - specifically, an ODE solver library - called CVODE (part of SUNDIALS). The library works with MPI, so that multiple threads are running in parallel. They are all running the same code. Each thread sends the thread "next to" it a piece of data. But I want one of the threads (rank=0) to print out the state of the data at some points. The library includes functions so that each thread can access their own

问题 I just finished installing OpenMDAOv1.3 on our super computer. The installation was successful and all tests either passed or skipped. However, when I ran the tests I got the following warning: *path/OpenMDAO/openmdao/core/driver.py:228: UserWarning: parallel derivs %s specified but not running under MPI warnings.warn("parallel derivs %s specified but not running under MPI") I'm not sure what to do about this (if anything) and so I am looking for information about the implications of the

问题 Following my previous question : Unable to implement MPI_Intercomm_create The problem of MPI_INTERCOMM_CREATE has been solved. But when I try to implement a basic send receive operations between process 0 of color 0 (globally rank = 0) and process 0 of color 1 (ie globally rank = 2), the code just hangs up after printing received buffer. the code: program hello include 'mpif.h' implicit none integer tag,ierr,rank,numtasks,color,new_comm,inter1,inter2 integer sendbuf,recvbuf,tag,stat(MPI

问题 I am building openMPI 1.8.5 on Ubuntu 12.04 with CUDA 6.5 installed and tested with default samples. I intend to run it on a single node with following configuration: Dell Precision T7400 Dual Xeon X5450 Nvidia GT730/Tesla C1060 The configure command issued was $ ./configure --prefix=/usr --with-cuda=/usr/local/cuda In the generated config.log, it is clear the the configure script was not able to find cuda.h and cuda_runtime_api.h in /usr/loca/cuda/include, which do exist. For cuda.h:

问题 I am using MPI for solving PDE. For this, I breakdown the 2D domain into different cells (size of each of these cells is " xcell,ycell " with xcell = size_x_domain/(number of X subdomains) and ycell = size_y_domain/(number of Y subdomains) . So, I am running the code with number of processes = (number of X subdomains)*(number of Y subdomains) The gain relatively to sequential version is that I communicate between each process representing the sub-domains. Here a figure illustrating my

问题 I am trying to implement a MPI of the filter code below, but I'm facing difficulties doing it. How should it be done?: Filter code: int A[100000][100000]; int B[100000][100000]; for (int i=1; i<(100000 - 1); i++) for (int i=1; j<(100000 - 1); j++) B[i][j] = A[i-1][j] + A[i+1][j] + A[i][j-1] + A[i][j+1] - 4*A[i][j]; This is what I have tried while following the six functions of MPI: int myrank; /* Rank of process */ int numprocs; /* Number of processes */ int source; /* Rank of sender */ int

问题 how to simply make an producer and consumer app. Producer make item, sends it to the consumer, while consumer waits until he has this item. He use it, item is gone and he sends request to create new one to the producer. And over a over again. I have mode some MPI_send and MPI_recv combination, but it just go only once. Producer make one item, consumer consume one item and app is deadlocked. Should I use non blocking recieve and send? int count=10; if(myrank==0){ //server for(i=0;i<10;i++){

问题 I am working on a project that uses MPI to create parallel processes, each process uses dlopen() to load a module that's been build as a shared object library. One of the modules that I'm writing uses a 3rd party library (HDF). When I run the program, dlopen throws an error: dlopen failed: /home/jwomble/QTProjects/SurrogateModule/libsurrogate.so: undefined symbol: H5T_NATIVE_INT32_g The undefined symbol is in the HDF library. How do I load the symbols from the HDF library? Currently, my make