Added test for FFTW vs TOMs transposes -- FFTW is much faster

tests/compile_test_transpose.sh

@@ -0,0 +1 @@
+gcc -g -O3 -Wall -W -ffast-math -I../include/ -o test_transpose test_transpose.c ../src/transpose.o ../src/vectors.o -lfftw3f -lm

tests/test_transpose.c

@@ -0,0 +1,151 @@
+#include <time.h>
+#include <sys/times.h>
+#include <string.h>
+#include <stdio.h>
+#include <math.h>
+#include "clk_tck.h"
+#include "vectors.h"
+#include "fftw3.h"
+#include "assert.h"
+
+extern short transpose_float(float *a, int nx, int ny, unsigned char *move, 
+                             int move_size);
+
+fftwf_plan plan_transpose(int rows, int cols, float *in, float *out) {
+    const unsigned flags = FFTW_MEASURE; /* other flags are possible */
+    fftwf_iodim howmany_dims[2];
+
+    howmany_dims[0].n  = rows;
+    howmany_dims[0].is = cols;
+    howmany_dims[0].os = 1;
+    howmany_dims[1].n  = cols;
+    howmany_dims[1].is = 1;
+    howmany_dims[1].os = rows;
+    return fftwf_plan_guru_r2r(/*rank=*/ 0, /*dims=*/ NULL,
+                               /*howmany_rank=*/ 2, howmany_dims,
+                               in, out, /*kind=*/ NULL, flags);
+}
+
+void print_array(float *arr, int N, int M) {
+    int ii, jj;
+    for (ii=0; ii<N; ii++){
+        for (jj=0; jj<M; jj++){
+            printf("%6.4f ", arr[ii*M+jj]);
+        }
+        printf("\n");
+    }
+    printf("\n");
+}
+
+
+int main(int argc, char *argv[]) {
+    float *array1, *array2;
+    fftwf_plan tplan1, tplan2;
+    struct tms runtimes;
+    double ttim, stim, utim, tott;
+    int ii, N, M, numtimes, move_size;
+    unsigned char *tmpspace;
+
+    if (argc <= 1 || argc > 4) {
+        printf("\nUsage:  test_transpose N M #times (array[N][M])\n\n");
+        exit(0);
+    } else {
+        N = atoi(argv[1]);
+        M = atoi(argv[2]);
+        numtimes = atoi(argv[3]);
+    }
+
+    // Setup arrays
+    array1 = gen_fvect(N * M);
+    array2 = gen_fvect(N * M);
+    for (ii = 0; ii < N * M; ii++)
+        array1[ii] = array2[ii] = (float) rand() / (float) RAND_MAX;
+
+    // Setup things for TOMs transpose
+    move_size = N * M / 2;
+    tmpspace = gen_bvect(move_size);
+
+    // Start the timing for TOMs transpose
+    tott = times(&runtimes) / (double) CLK_TCK;
+    utim = runtimes.tms_utime / (double) CLK_TCK;
+    stim = runtimes.tms_stime / (double) CLK_TCK;
+
+    if (N * M <= 20)
+        print_array(array1, N, M);
+    for (ii = 0; ii < numtimes; ii++) {
+        if (ii % 2)
+            transpose_float(array1, M, N, tmpspace, move_size);
+        else
+            transpose_float(array1, N, M, tmpspace, move_size);
+    }
+    if (N * M <= 20) {
+        if (numtimes % 2)
+            print_array(array1, M, N);
+        else
+            print_array(array1, N, M);
+    }
+
+    tott = times(&runtimes) / (double) CLK_TCK - tott;
+    utim = runtimes.tms_utime / (double) CLK_TCK - utim;
+    stim = runtimes.tms_stime / (double) CLK_TCK - stim;
+    ttim = utim + stim;
+
+    // Check for correctness
+    if (numtimes % 2 == 0) {
+        for (ii = 0; ii < N * M; ii++)
+            assert(fabs(array1[ii] - array2[ii]) < 1e-6);
+    }
+
+    printf("Timing summary (TOMS) NxM = %dx%d:\n", N, M);
+    printf("CPU usage: %.3f sec total (%.3f sec user, %.3f sec system)\n", \
+	   ttim, utim, stim);
+    printf("Total time elapsed:  %.3f sec.\n\n", tott);
+
+    // Now do the FFTW transpose
+
+    // The plan messes up the input array!
+    tplan1 = plan_transpose(N, M, array1, array1);
+    tplan2 = plan_transpose(M, N, array1, array1);
+    memcpy(array1, array2, sizeof(float) * N * M);
+
+    tott = times(&runtimes) / (double) CLK_TCK;
+    utim = runtimes.tms_utime / (double) CLK_TCK;
+    stim = runtimes.tms_stime / (double) CLK_TCK;
+
+    if (N * M <= 20)
+        print_array(array1, N, M);
+    for (ii = 0; ii < numtimes; ii++) {
+        if (ii % 2)
+            fftwf_execute_r2r(tplan2, array1, array1);
+        else
+            fftwf_execute_r2r(tplan1, array1, array1);
+    }
+    if (N * M <= 20) {
+        if (ii % 2)
+            print_array(array1, M, N);
+        else
+            print_array(array1, N, M);
+    }
+
+    tott = times(&runtimes) / (double) CLK_TCK - tott;
+    utim = runtimes.tms_utime / (double) CLK_TCK - utim;
+    stim = runtimes.tms_stime / (double) CLK_TCK - stim;
+    ttim = utim + stim;
+
+    // Check for correctness
+    if (numtimes % 2 == 0) {
+        for (ii = 0; ii < N * M; ii++)
+            assert(fabs(array1[ii] - array2[ii]) < 1e-6);
+    }
+
+    printf("Timing summary (FFTW) NxM = %dx%d:\n", N, M);
+    printf("CPU usage: %.3f sec total (%.3f sec user, %.3f sec system)\n", \
+	   ttim, utim, stim);
+    printf("Total time elapsed:  %.3f sec.\n\n", tott);
+
+    vect_free(array1);
+    vect_free(array2);
+    vect_free(tmpspace);
+    return 0;
+}
+