MPI_ALLREDUCE PROBLEM

Question

I have a code which has a 2D local array (cval).This local array is being calculated by every processor and in the end I call MPI_ALLREDUCE to sum this local array to a global array(gns). This local array has different sizes for different processors.The way I do a all reduce is as follows

k = n2spmax- n2spmin + 1 ! an arbitrary big value

do i = nmin, nmax
     call MPI_ALLREDUCE(cval(i,:),gns(i,:),k,MPI_DOUBLE_PRECISION,MPI_SUM,MPI_COMM_WORLD,ierr)
end do

Is this the correct way of writing it.I am not sure about it ?

Jonathan Dursi · Accepted Answer · 2011-03-31 14:43:00Z

1

No, you can't do it this way. MPI_Allreduce requires that all of the processes in the communicator are contributing the same amount of data. That's why there's a single count argument.

To give more guidance on what is the right way to do this, we'll need a bit more clarity on what you're trying to do. Is the idea that you're calculating gns(i,j) = the sum over all ranks of cval(i,j), but not all ranks have all the cval(i,j)s?

answered Mar 31, 2011 at 14:43

Jonathan Dursi

51.1k10 gold badges131 silver badges160 bronze badges

Sign up to request clarification or add additional context in comments.

6 Comments

VAIBHAW KUMAR Over a year ago

Thank you for the prompt reply.Let me explain this in detail.I have two global ranges say (nmin,nmax) and (n2min,n2max). Hence gns has to be calculated for the entire range i.e gns(nmin:nmax,n2min:n2max).While these two ranges are divided among processors in an unequal fashion.So cval is the local array on each processor working on the the local ranges.At the end all the values of cval need to clubbed together in a single array gns.

suszterpatt Over a year ago

@Vaibhaw: If you're summing the values, then you can pad smaller cval arrays with zeros so that they're all the same size, and your result will be the same.

VAIBHAW KUMAR Over a year ago

Actually I have the cval allocated as cval(nmin:nmax,n2min:n2max) i.e same as gns.Before MPI_ALLREDUCE call, each processor works on cval(nl:nh,nln2:nhn2) where [nl,nh] is the local range for [nmin:nmax] while [nln2,nhn2] is the local range for [n2min,n2max].[nln2,nhn2] is of different size for different processor.I give you an example:

VAIBHAW KUMAR Over a year ago

1st proc:[nl,nh],[nln2,nhn2]= [0,88],[0,88]

VAIBHAW KUMAR Over a year ago

2nd proc:[nl,nh],[nln2,nhn2]= [88,125],[0,125]

|

Collectives™ on Stack Overflow

MPI_ALLREDUCE PROBLEM

1 Answer 1

6 Comments

Your Answer

Hot Network Questions

Collectives™ on Stack Overflow

1 Answer 1

6 Comments

Your Answer

Sign up or log in

Post as a guest

Related