adding assignment source

Makefile

+#!?bin/sh
+
+CC=gcc
+CFLAGS=-O3
+COPT=-std=c99
+OMPFLAGS=-fopenmp
+
+all: dgemm_seq dgemm_l1 dgemm_l2 dgemm_l3
+dgemm_seq: main.o dgemm_seq.o
+	$(CC) $(CFLAGS) -o $@ main.o dgemm_seq.o
+dgemm_l1: main.o dgemm_l1.o
+	$(CC) $(CFLAGS) $(OMPFLAGS) -o $@ main.o dgemm_l1.o
+dgemm_l2: main.o dgemm_l2.o
+	$(CC) $(CFLAGS) $(OMPFLAGS) -o $@ main.o dgemm_l2.o
+dgemm_l3: main.o dgemm_l3.o
+	$(CC) $(CFLAGS) $(OMPFLAGS) -o $@ main.o dgemm_l3.o
+dgemm_seq.o: ./src/dgemm_seq.c
+	$(CC) $(CFLAGS) $(COPT) -c -o $@ $<
+dgemm_l1.o: ./src/dgemm_l1.c
+	$(CC) $(CFLAGS) $(COPT) $(OMPFLAGS) -c -o $@ $<
+dgemm_l2.o: ./src/dgemm_l1.c
+	$(CC) $(CFLAGS) $(COPT) $(OMPFLAGS) -c -o $@ $<
+dgemm_l3.o: ./src/dgemm_l1.c
+	$(CC) $(CFLAGS) $(COPT) $(OMPFLAGS) -c -o $@ $<
+main.o: ./src/main.c
+	$(CC) $(CFLAGS) -c -o $@ $<
+clean:
+	rm -Rf dgemm_seq dgemm_l1 dgemm_l2 dgemm_l3 ./*.o

README

+CS5332 OpenMP Assignment
+-----------------------------------
+john.leidel@ttu.edu
+http://discl.cs.ttu.edu/gitlab/jleidel/cs5332_omp.git
+-----------------------------------
+
+The goal of this assignment is two fold. 
+First, we seek to utilize our lecture notes 
+from lecture 10 and implement a parallel 
+matrix-matrix multiplication kernel.  
+Second, we seek to observe the performance 
+differences depending upon where the directives 
+are implemented.  
+
+The source directory contains several files
+- main.c : driver
+- dgemm_seq.c : sequential DGEMM
+- dgemm_l1.c : L1 parallel solver
+- dgemm_l2.c : L2 parallel solver 
+- dgemm_l3.c : L3 parallel solver
+
+For each of the {L1,L2,L3} parallel solvers, 
+implement the necessary OpenMP clauses on 
+the 1st, 2nd and 3rd loop nests, respectively.
+Make sure that the private and shared data 
+variables are correctly specified.
+
+Execute the sequential version (dgemm_seq) 
+and record the timing and performance.
+Execute each of the parallel versions 
+using 1-8 threads and record the timing 
+and performance for each.  The driver 
+will print the timing and the performance.
+
+Use the included makefile to build the 
+different versions.  (Just type 'make').
+
+-----------------------------------
+EOF

src/dgemm_l1.c

+/*
+ * _DGMEMM_L1_C_
+ *
+ * L1 OpenMP DGEMM
+ *
+ * CS5332 OpenMP Programming Assignment
+ *
+ * John Leidel: john.leidel@ttu.edu
+ * http://discl.cs.ttu.edu/gitlab/jleidel/cs5332_omp.git*
+ *
+ */
+
+
+extern void naive_dgemm( double **restrict Result,
+                         double **restrict A,
+                         double **restrict B,
+                         int N ){
+  /* vars */
+  int i = 0;
+  int j = 0;
+  int k = 0;
+  double sum = 0.;
+  /* --- */
+
+  for( i=0; i<N; i++ ){
+    for( j=0; j<N; j++ ){
+      sum = 0.;
+      for( k=0; k<N; k++ ){
+        sum = sum + A[i][k] * B[k][j];
+      }
+      Result[i][j] = sum;
+    }
+  }
+}
+
+/* EOF */

src/dgemm_l2.c

+/*
+ * _DGMEMM_L2_C_
+ *
+ * L2 OpenMP DGEMM
+ *
+ * CS5332 OpenMP Programming Assignment
+ *
+ * John Leidel: john.leidel@ttu.edu
+ * http://discl.cs.ttu.edu/gitlab/jleidel/cs5332_omp.git*
+ *
+ */
+
+
+extern void naive_dgemm( double **restrict Result,
+                         double **restrict A,
+                         double **restrict B,
+                         int N ){
+  /* vars */
+  int i = 0;
+  int j = 0;
+  int k = 0;
+  double sum = 0.;
+  /* --- */
+
+  for( i=0; i<N; i++ ){
+    for( j=0; j<N; j++ ){
+      sum = 0.;
+      for( k=0; k<N; k++ ){
+        sum = sum + A[i][k] * B[k][j];
+      }
+      Result[i][j] = sum;
+    }
+  }
+}
+
+/* EOF */

src/dgemm_l3.c

+/*
+ * _DGMEMM_L3_C_
+ *
+ * L3 OpenMP DGEMM
+ *
+ * CS5332 OpenMP Programming Assignment
+ *
+ * John Leidel: john.leidel@ttu.edu
+ * http://discl.cs.ttu.edu/gitlab/jleidel/cs5332_omp.git*
+ *
+ */
+
+
+extern void naive_dgemm( double **restrict Result,
+                         double **restrict A,
+                         double **restrict B,
+                         int N ){
+  /* vars */
+  int i = 0;
+  int j = 0;
+  int k = 0;
+  double sum = 0.;
+  /* --- */
+
+  for( i=0; i<N; i++ ){
+    for( j=0; j<N; j++ ){
+      sum = 0.;
+      for( k=0; k<N; k++ ){
+        sum = sum + A[i][k] * B[k][j];
+      }
+      Result[i][j] = sum;
+    }
+  }
+}
+
+/* EOF */

src/dgemm_seq.c

+/*
+ * _DGMEMM_SEQ_C_
+ *
+ * Sequential DGEMM
+ *
+ * CS5332 OpenMP Programming Assignment
+ *
+ * John Leidel: john.leidel@ttu.edu
+ * http://discl.cs.ttu.edu/gitlab/jleidel/cs5332_omp.git*
+ *
+ */
+
+
+extern void naive_dgemm( double **restrict Result,
+                         double **restrict A,
+                         double **restrict B,
+                         int N ){
+  /* vars */
+  int i = 0;
+  int j = 0;
+  int k = 0;
+  double sum = 0.;
+  /* --- */
+
+  for( i=0; i<N; i++ ){
+    for( j=0; j<N; j++ ){
+      sum = 0.;
+      for( k=0; k<N; k++ ){
+        sum = sum + A[i][k] * B[k][j];
+      }
+      Result[i][j] = sum;
+    }
+  }
+}
+
+/* EOF */

src/main.c

+/*
+ * _MAIN_C_
+ *
+ * CS5332 OpenMP Programming Assignment
+ *
+ * John Leidel: john.leidel@ttu.edu
+ * http://discl.cs.ttu.edu/gitlab/jleidel/cs5332_omp.git
+ *
+ */
+
+#include <stdio.h>
+#include <stdlib.h>
+#include <unistd.h>
+#include <sys/types.h>
+#include <sys/time.h>
+
+/* --------------------------------- MACROS */
+#define X_LEN 1024
+#define Y_LEN 1024
+
+/* --------------------------------- FUNCTION PROTOTYPES */
+extern void naive_dgemm( double **Result,
+                         double **A,
+                         double **B,
+                         int N );
+
+/* --------------------------------- MYSECOND */
+double mysecond(){
+  struct timeval tp;
+  struct timezone tzp;
+
+  gettimeofday( &tp, &tzp );
+  return ( (double) tp.tv_sec + (double) tp.tv_usec * 1.e-6 );
+}
+
+/* --------------------------------- MAIN */
+int main( int argc, char **argv ){
+  /* vars */
+  double *Ap  = NULL; /* A storage */
+  double *Bp  = NULL; /* B storage */
+  double *Rp  = NULL; /* R storage */
+  double **A  = NULL; /* A pointer storage */
+  double **B  = NULL; /* B pointer storage */
+  double **R  = NULL; /* R pointer storage */
+  double start_t = 0.;
+  double end_t = 0.;
+  double flop = 0.;
+  unsigned long long i = 0;
+  pid_t pid = -1;
+  /* ---- */
+
+  flop = ((double)(X_LEN)*(double)(Y_LEN))*((2.0*(double)(X_LEN))-1.0);
+
+  /* -- allocate memory */
+  Ap = malloc( sizeof( double ) * X_LEN * Y_LEN );
+  if( Ap == NULL ){
+    printf( "Error : could not allocate memory for Ap\n" );
+    return -1;
+  }
+
+  Bp = malloc( sizeof( double ) * X_LEN * Y_LEN );
+  if( Bp == NULL ){
+    printf( "Error : could not allocate memory for Bp\n" );
+    return -1;
+  }
+
+  Rp = malloc( sizeof( double ) * X_LEN * Y_LEN );
+  if( Rp == NULL ){
+    printf( "Error : could not allocate memory for Rp\n" );
+    return -1;
+  }
+
+  A = malloc( sizeof( double * ) * X_LEN );
+  if( A == NULL ){
+    printf( "Error : could not allocate memory for A\n" );
+    return -1;
+  }
+
+  B = malloc( sizeof( double * ) * X_LEN );
+  if( B == NULL ){
+    printf( "Error : could not allocate memory for B\n" );
+    return -1;
+  }
+
+  R = malloc( sizeof( double * ) * X_LEN );
+  if( R == NULL ){
+    printf( "Error : could not allocate memory for R\n" );
+    return -1;
+  }
+
+  /* init the memory with nonzero values */
+  pid = getpid();
+  for( i=0; i<(X_LEN*Y_LEN); i++ ){
+    Ap[i] = (double)(pid)*(double)(i);
+    Bp[i] = Ap[i]*2.0;
+    Rp[i] = 0.;
+  }
+
+  /* init our 2D pointers */
+  for( i=0; i<X_LEN; i++ ){
+    A[i] = &(Ap[i*Y_LEN]);
+    B[i] = &(Bp[i*Y_LEN]);
+    R[i] = &(Rp[i*Y_LEN]);
+  }
+
+  /* execute the benchmark */
+  start_t = mysecond();
+  naive_dgemm( R, A, B, X_LEN );
+  end_t = mysecond();
+
+  printf( "Executed DGMEM in %f seconds; %f Gflops\n",
+          (end_t-start_t),
+          (flop/1000000000.0)/(end_t-start_t) );
+
+  /* free the memory */
+  free( A );
+  free( B );
+  free( R );
+  free( Ap );
+  free( Bp );
+  free( Rp );
+
+  return 0;
+}
+
+/* EOF */