0

I wrote a program in c using MPI (Message Passing Inteface) that compute recursively the inverse of a lower triangular matrix. Every cpu sends 2 submatrices to other two cpus, they compute them and they give them back to the cpu caller. When the cpu caller has its submatrices it has to perform a matrix multiplication. In the recurrence equation the bottle neck is matrix multiplication. I implemented parallel multiplication with mpi in c but i'm not able to embed it into a function. Is it possible?

thanks, Simone

Improve this question
1
  • Are you asking how you write a synchronous function to block on the computation you're trying to do? Or do you want to expose your code via an API and have it use MPI to do the work? Sorry, I'm having trouble understanding what you're asking. Commented Aug 5, 2009 at 0:16

1 Answer 1

Reset to default
1

You can use PDGEMM/PSGEMM functions for parallel mpi matrix multiplication from SCALAPACK/PBLAS.

http://www.netlib.org/scalapack/pblas_qref.html#PvGEMM

download here http://www.netlib.org/scalapack/

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

Comments

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.