Text isn’t a great format for parallel processing exactly because you don’t know ahead of time where (say) line 25001 begins. So these sorts of problems are often dealt with ahead of time through some preprocessing step, either building an index or partitioning the file into the appropriate number of chunks for each process to … Read more
There’s a long description of this issue in my answer to this question: the fact that many people have these questions is proof that it’s not obvious and the ideas take some getting used to. The important thing to know is what memory layout the MPI datatype describes. The calling sequence to MPI_Type_vector is: int … Read more
You have to go through at least one extra step in MPI to do this. The problem is that the most general of the gather/scatter routines, MPI_Scatterv and MPI_Gatherv, allow you to pass a “vector” (v) of counts/displacements, rather than just one count for Scatter and Gather, but the types are all assumed to be … Read more
What you’ve got is pretty much “best practice”; it’s just a bit confusing until you get used to it. Two things, though: First, be careful with this: sizeof(MPI_CHAR) is, I assume, 4 bytes, not 1. MPI_CHAR is an (integer) constant that describes (to the MPI library) a character. You probably want sizeof(char), or SIZE/2*sizeof(char), or … Read more
Atomicity of writes less than PIPE_BUF applies only to pipes and FIFOs. For file writes, POSIX says: This volume of POSIX.1-2008 does not specify behavior of concurrent writes to a file from multiple processes. Applications should use some form of concurrency control. …which means that you’re on your own – different UNIX-likes will give different … Read more
There are a number of performance analysis tools specialized for Parallel/MPI Programs, such as: Score-P, which works with a number of different Analysis tools, e.g. Cube, Vampir HPCToolkit uses sampling only, so you do not have to recompile your application Tau At first they may not be as simple to use simple to use, but … Read more
Implementing a shared counter isn’t trivial, but once you do it and have it in a library somewhere you can do a lot with it. In the Using MPI-2 book, which you should have to hand if you’re going to implement this stuff, one of the examples (the code is available online) is a shared … Read more
The purpose of MPI_Type_create_struct() is, as you know, to provide a way to create user’s MPI_Datatypes mapping his structured types. These new types will subsequently be usable for MPI communications and other calls just as the default types, allowing for example to transfer arrays of structures the same way you would transfer arrays of ints … Read more
Yes, there’s lots of such tools. MPI defines a tools interface that allows other libraries to interject themselves at your MPI function calls, and do counts, timing, etc. A very small MPI profiling tool is mpiP – it gives a very short summary of MPI activity in your code. The IPM library is fairly easy … Read more
This is a common source of confusion for people new to MPI. You don’t use MPI_Recv() to receive data sent by a broadcast; you use MPI_Bcast(). Eg, what you want is this: #include <mpi.h> #include <stdio.h> int main(int argc, char** argv) { int rank; int buf; const int root=0; MPI_Init(&argc, &argv); MPI_Comm_rank(MPI_COMM_WORLD, &rank); if(rank == … Read more