代做DTS202TC Foundation of Parallel Computing Lab 5: MPI代写Processing
- 首页 >> C/C++编程DTS202TC Foundation of Parallel Computing
Lab 5: MPI
Task 1, OpenMPI
OpenMPI is one of the Message Passing Interface (MPI) implementations. Turn on the virtual machine from Lab 1, then verify the installation:
1 mpicc --version
Task 2, Simple OpenMPI Program
Type the following source code in a mpi hello.c file.
Compile the source code by:
1 mpicc -g -Wall -o mpi_hello mpi_hello.c
Run the program by:
1 mpiexec -n 4 ./mpi_hello
Figure 1: For task 2
Task 3
The following code estimates the value of the mathematical constant PI. Write an MPI version of estimate_pi that uses all available processes to do the work. Use MPI Send and MPI_Recv to communicate between processes.
1 /*Estimate PI*/
2
3 #include <stdio.h>
4 #include <math.h>
5 #include <stdlib.h>
6 #include <sys/time.h>
7
8 /* The argument now should be a double (not a pointer to a double) */
9 #define GET_TIME(now) { \
10 struct timeval t; \
11 gettimeofday(&t, NULL); \
12 now = t.tv_sec + t.tv_usec /1000000.0; \
13 }
14
15 double serial_pi(long long n) {
16 double sum = 0.0;
17 long long i;
18 double factor = 1.0;
19
20 for (i = 0; i < n; i++, factor = -factor) {
21 sum += factor/(2*i+1);
22 }
23 return 4.0*sum;
24
25 }
26
27 int main(int argc , char** argv) {
28 double start , finish;
29 GET_TIME(start);
30 double estimate_of_pi = serial_pi (1000000000) ;
31 printf("\nEstimated of pi: %1.10f.\n", estimate_of_pi);
32 GET_TIME(finish);
33
34 printf("\nAcutal value of pi: %1.10f.\n\n", atan(1) *4);
35 printf("The elapsed time is %e seconds\n", finish -start);
36 }
Analyse the performance of the serial and MPI versions by simply printing out the total execution time. Also compare the performance with different number of processes.
Task 4 (Optional)
Ask AI for more MPI practices if you have spare time.