并行计算程序

Descr iption: This program blurs an image file in a parallelized computation

The root process reads the file and distributes a part of the image file to

other processes. Each process makes the filter operation on the part of the image data which is received from the root. Then all the processes send their blurred image data to the root and root writes the result in a file-Distributed Image Processing Filter

Write a program that blurs an image file in a parallelized computation. Whenever possible, use collective

operations (MPI_Scatter, MPI_Gather). The root process reads the image file from the NFS

server, distributes the data of the file among all processes, and gathers the results into the output file.

Each processor should receive a part of the image only. The image file is provided as a raw image file.

It can be read using the C function fread() with a single call and written with the single call of the

function fwrite(). The image is given as an array of h rows, a row has 1280 pixels , where each pixel

is one unsigned char grey value. You may assume that h is lower than or equal to 8192.

a) Write an MPI program in which the root process prepairs the data, so that all other processes

get their data by exactly one MPI_Scatter operation.