The program: You should write a program to multiply two
matrices (C = A*B), and output the sum of all the elements
of C. Your code will take four command line arguments,
M, N, P, and T, which are the number of rows
of A, the number of columns of A = number of
rows of B, the number of columns of B, and
the number of threads respectively. You should use OpenMP to
parallelize the program.
In order to help me check your program, please create a header file called AB.h which defines the following two macros, and use them to define the matrices A and B:
When you run your program, let A be defined by
A(i,j) = a(i+1,j+1) and B by B(j,k) =
b(j+1,k+1), 0 < i < M-1, 0
< j < N-1, 0 < k < P-1
. When I check your program, I may change the macros and check
if your code performs correctly, for other matrices that I define.
Due date: Monday, 17 Feb 2003.
How and what to turn in: Email me a tar file containing your code and Makefile, at ashoks@modi4.ncsa.uiuc.edu, by 10 am on the due date. You should also plot two speed-up curves, (i) with automatic parallelization, applied to the program without OpenMP directives (here, you will need to use an environment variable to specify the number of threads) and (ii) with OpenMP parallelization. Please turn in hardcopies of these figures at the beginning of class 17 Feb 2003.
Note: (i) Your code should run on the Origin 2000 at NCSA,
and you must have a Makefile. (ii) The tar file should not contain any
executable or object files. (iii) The tar file should be called
hw1.tar. On untarring (tar xvf hw1.tar), a directory called
hw1 should be created, with the required files in
it. (iv) Typing make in the hw1 directory
should create the executable hw1. (v) You should actually implement
the matrix multiplication algorithm yourself, rather a using a library
created by someone else! (vi) Please do not use more than 2 hours of cpu time, when running programs for this homework.
Grading criteria: Correctness of your program and its parallelization, performance of your program, speed-up curves, and following the given instructions.
The program: You should write a program to multiply two
square matrices (C = A*B), where A is lower triangular, and output the sum of all the elements
of C. Your code will take one command line argument,
N, which is the number of rows (and columns)
of A and B. You should use MPI to
parallelize the program.
In order to help me check your program, please create a header file called AB.h which defines the following two macros, and use them to define the matrices A and B:
When you run your program, let A be defined by
A(i,j) = a(i+1,j+1) and B by B(j,k) =
b(j+1,k+1), 0 < i,j < N-1. When I check your program, I may change the macros and check
if your code performs correctly, for other matrices that I define.
Due date: Monday, 31 Mar 2003.
How and what to turn in: Email me a tar file containing your code and Makefile, at ashoks@modi4.ncsa.uiuc.edu, by 10 am on the due date. You should also plot a speed-up curve. Please turn in hardcopies of these figures at the beginning of class, 31 Mar 2003.
Note: (i) Your code should run on the Origin 2000 at NCSA,
and you must have a Makefile. (ii) The tar file should not contain any
executable or object files. (iii) The tar file should be called
hw2.tar. On untarring (tar xvf hw2.tar), a directory called
hw2 should be created, with the required files in
it. (iv) Typing make in the hw2 directory
should create the executable hw2. (v) You should actually implement
the matrix multiplication algorithm yourself, rather a using a library
created by someone else! (vi) Please do not use more than 2 hours of cpu time, when running programs for this homework.
Grading criteria: Correctness of your program and its parallelization, efficiency of the algorithm and its implementation, speed-up curves, and following the given instructions.