Facultyprofile1blog.files.wordpress.com



BHARATI VIDYAPEETH’s COLLEGE OF ENGINEERING FOR WOMEN, PUNECOMPUTER ENGINEERING DEPARTMENTBE COMPUTERHIGH PERFORMANCE COMPUTINGQUESTION BANKUNIT IIIWrite down principles of message passing programming.Explain send & receive primitives in details.Explain non-buffered blocking message passing operation.Define MPI. Also explain minimal set of MPI routines & starting & terminating the MPI library.Define communicator. Write down a basic MPI program using MPI routines to print number of processes and determining the label of calling process. How will you avoid a deadlock in MPI? Explain in details.What are different ways to map a set of processes to a two-dimensional grid?How will you overlap communication with computation?Or Explain following operation: MPI_send, MPI_recv, MPI_Isend, MPI_Irecv, MPI_Test. Write a program to remove deadlock using non-blocking operation. Explain following operation – MPI_Barrier, MPI_Bcast, MPI_reduce. Explain collective communication operation in details. Explain concept of a groups and communicators. Write a short note on:One dimensional matrix-vector multiplication.Two dimensional matrix-vector multiplications.Single source shortest path. How will you define/describe one dimensional matrix vector multiplication with help of MPI Explain buffered vs. buffered message passing primitives. Write a short note on topologies and embedding. Design pseudo code for parallel quick sort. Implement cannons algorithm for matrix multiplication with example. Explain message passing interface. Explain non-blocking communicators using MPI. Define MPP. Write principals of MPP.Explain collective communication and computation operations.Explain groups and Communications. ................
................

In order to avoid copyright disputes, this page is only a partial summary.

Google Online Preview   Download