The following a literal Fortran translation of this answer. I had thought this was unnecessary, but the multiple differences in array indexing and memory layout might mean that it is worth doing a Fortran version.
Let me start by saying that you generally don't really want to do this - scatter and gather huge chunks of data from some "master" process. Normally you want each task to be chugging away at its own piece of the puzzle, and you should aim to never have one processor need a "global view" of the whole data; as soon as you require that, you limit scalability and the problem size. If you're doing this for I/O - one process reads the data, then scatters it, then gathers it back for writing, you'll want eventually to look into MPI-IO.
Getting to your question, though, MPI has very nice ways of pulling arbitrary data out of memory, and scatter/gathering it to and from a set of processors. Unfortunately that requires a fair number of MPI concepts - MPI Types, extents, and collective operations. A lot of the basic ideas are discussed in the answer to this question -- MPI_Type_create_subarray and MPI_Gather .
Consider a 1d integer global array that task 0 has that you want to distribute to a number of MPI tasks, so that they each get a piece in their local array. Say you have 4 tasks, and the global array is [0,1,2,3,4,5,6,7]. You could have task 0 send four messages (including one to itself) to distribute this, and when it's time to re-assemble, receive four messages to bundle it back together; but that obviously gets very time consuming at large numbers of processes. There are optimized routines for these sorts of operations - scatter/gather operations. So in this 1d case you'd do something like this:
integer, dimension(8) :: global ! only root has this
integer, dimension(2) :: local ! everyone has this
integer, parameter :: root = 0
integer :: rank, comsize
integer :: i, ierr
call MPI_Init(ierr)
call MPI_Comm_size(MPI_COMM_WORLD, comsize, ierr)
call MPI_Comm_rank(MPI_COMM_WORLD, rank, ierr)
if (rank == root) then
global = [ (i, i=1,8) ]
endif
call MPI_Scatter(global, 2, MPI_INTEGER, & ! send everyone 2 ints from global
local, 2, MPI_INTEGER, & ! each proc recieves 2 into
root, & ! sending process is root,
MPI_COMM_WORLD, ierr) ! all procs in COMM_WORLD participate
After this, the processors' data would look like
task 0: local:[1,2] global: [1,2,3,4,5,6,7,8]
task 1: local:[3,4] global: [garbage]
task 2: local:[5,6] global: [garbage]
task 3: local:[7,8] global: [garbage]
That is, the scatter operation takes the global array and sends contiguous 2-int chunks to all the processors.
To re-assemble the array, we use the MPI_Gather() operation, which works exactly the same but in reverse:
local = local + rank
call MPI_Gather (local, 2, MPI_INTEGER, & ! everyone sends 2 ints from local
global, 2, MPI_INTEGER, & ! root receives 2 ints each proc into global
root, & ! receiving process is root,
MPI_COMM_WORLD, ierr) ! all procs in COMM_WORLD participate
And now the arrays look like:
task 0: local:[1,2] global: [1,2,4,5,7,8,10,11]
task 1: local:[4,5] global: [garbage-]
task 2: local:[7,8] global: [garbage-]
task 3: local:[10,11] global: [garbage-]
Gather brings all the data back.
What happens if the number of data points doesn't evenly divide the number of processes, and we need to send different numbers of items to each process? Then you need a generalized version of scatter, MPI_Scatterv, which lets you specify the counts for each processor, and displacements -- where in the global array that piece of data starts. So let's say with the same 4 tasks you had an array of characters [a,b,c,d,e,f,g,h,i] with 9 characters, and you were going to assign every process two characters except the last, that got three. Then you'd need
character, dimension(9) :: global
character, dimension(3) :: local
integer, dimension(4) :: counts
integer, dimension(4) :: displs
if (rank == root) then
global = [ (achar(i+ichar('a')), i=0,8) ]
endif
local = ['-','-','-']
counts = [2,2,2,3]
displs = [0,2,4,6]
mycounts = counts(rank+1)
call MPI_Scatterv(global, counts, displs, & ! proc i gets counts(i) chars from displs(i)
MPI_CHARACTER, &
local, mycounts, MPI_CHARACTER, & ! I get mycounts chars into
root, & ! root rank does sending
MPI_COMM_WORLD, ierr) ! all procs in COMM_WORLD participate
Now the data looks like
task 0: local:"ab-" global: "abcdefghi"
task 1: local:"cd-" global: *garbage*
task 2: local:"ef-" global: *garbage*
task 3: local:"ghi" global: *garbage*
You've now used scatterv to distribute the irregular amounts of data. The displacement in each case is two*rank (measured in characters; the displacement is in unit of the types being sent for a scatter or received for a gather; it's not generally in bytes or something) from the start of the array, and the counts are [2,2,2,3]. If it had been the first processor we wanted to have 3 characters, we would have set counts=[3,2,2,2] and displacements would have been [0,3,5,7]. Gatherv again works exactly the same but reverse; the counts and displs arrays would remain the same.
Now, for 2D, this is a bit trickier. If we want to send 2d sublocks of a 2d array, the data we're sending now no longer is contiguous. If we're sending (say) 3x3 subblocks of a 6x6 array to 4 processors, the data we're sending has holes in it:
2D Array
---------
|000|222|
|000|222|
|000|222|
|---+---|
|111|333|
|111|333|
|111|333|
---------
Actual layout in memory
[000111000111000111222333222333222333]
(Note that all high-performance computing comes down to understanding the layout of data in memory.)
If we want to send the data that is marked "1" to task 1, we need to skip three values, send three values, skip three values, send three values, skip three values, send three values. A second complication is where the subregions stop and start; note that region "1" doesn't start where region "0" stops; after the last element of region "0", the next location in memory is partway-way through region "1".
Let's tackle the first layout problem first - how to pull out just the data we want to send. We could always just copy out all the "0" region data to another, contiguous array, and send that; if we planned it out carefully enough, we could even do that in such a way that we could call MPI_Scatter on the results. But we'd rather not have to transpose our entire main data structure that way.
So far, all the MPI data types we've used are simple ones - MPI_INTEGER specifies (say) 4 bytes in a row. However, MPI lets you create your own data types that describe arbitrarily complex data layouts in memory. And this case -- rectangular subregions of an array -- is common enough that there's a specific call for that. For the 2-dimensional case we're describing above,
integer :: newtype;
integer, dimension(2) :: sizes, subsizes, starts
sizes = [6,6] ! size of global array
subsizes = [3,3] ! size of sub-region
starts = [0,0] ! let's say we're looking at region "0"
! which begins at offset [0,0]
call MPI_Type_create_subarray(2, sizes, subsizes, starts, MPI_ORDER_FORTRAN, MPI_INTEGER, newtype, ierr)
call MPI_Type_commit(newtype, ierr)
This creates a type which picks out just the region "0" from the global array. Note that even in Fortran, the start parameter is given as an offset (eg, 0-based) from the start of the array, not an index (eg, 1-based).
We could send just that piece of data now to another processor
call MPI_Send(global, 1, newtype, dest, tag, MPI_COMM_WORLD, ierr) ! send region "0"
and the receiving process could receive it into a local array. Note that the receiving process, if it's only receiving it into a 3x3 array, can not describe what it's receiving as a type of newtype; that no longer describes the memory layout, because there aren't big skips between the end of one row and the start of the next. Instead, it's just receiving a block of 3*3 = 9 integers:
call MPI_Recv(local, 3*3, MPI_INTEGER, 0, tag, MPI_COMM_WORLD, ierr)
Note that we could do this for other sub-regions, too, either by creating a different type (with different start array) for the other blocks, or just by sending starting from the first location of the particular block:
if (rank == root) then
call MPI_Send(global(4,1), 1, newtype, 1, tag, MPI_COMM_WORLD, ierr)
call MPI_Send(global(1,4), 1, newtype, 2, tag, MPI_COMM_WORLD, ierr)
call MPI_Send(global(4,4), 1, newtype, 3, tag, MPI_COMM_WORLD, ierr)
local = global(1:3, 1:3)
else
call MPI_Recv(local, 3*3, MPI_INTEGER, 0, tag, MPI_COMM_WORLD, rstatus, ierr)
endif
Now that we understand how to specify subregions, there's only one more thing to discuss before using scatter/gather operations, and that's the "size" of these types. We couldn't just use MPI_Scatter() (or even scatterv) with these types yet, because these types have an extent of 15 integers; that is, where they end is 15 integers after they start -- and where they end doesn't line up nicely with where the next block begins, so we can't just use scatter - it would pick the wrong place to start sending data to the next processor.
Of course, we could use MPI_Scatterv() and specify the displacements ourselves, and that's what we'll do - except the displacements are in units of the send-type size, and that doesn't help us either; the blocks start at offsets of (0,3,18,21) integers from the start of the global array, and the fact that a block ends 15 integers from where it starts doesn't let us express those displacements in integer multiples at all.
To deal with this, MPI lets you set the extent of the type for the purposes of these calculations. It doesn't truncate the type; it's just used for figuring out where the next element starts given the last element. For types like these with holes in them, it's frequently handy to set the extent to be something smaller than the distance in memory to the actual end of the type.
We can set the extent to be anything that's convenient to us. We could just make the extent 1 integer, and then set the displacements in units of integers. In this case, though, I like to set the extent to be 3 integers - the size of a sub-column - that way, block "1" starts immediately after block "0", and block "3" starts immediately after block "2". Unfortunately, it doesn't quite work as nicely when jumping from block "2" to block "3", but that can't be helped.
So to scatter the subblocks in this case, we'd do the following:
integer(kind=MPI_ADDRE
