| Mohan Srinivasan | d5f50d8 | 2017-01-26 11:07:14 -0800 | [diff] [blame] | 1 | #include <fcntl.h> |
| 2 | #include <sys/ioctl.h> |
| 3 | #include <stdio.h> |
| 4 | #include <sys/time.h> |
| 5 | #include <sys/types.h> |
| 6 | #include <sys/mman.h> |
| 7 | #include <unistd.h> |
| 8 | #include <stdlib.h> |
| 9 | #include <math.h> |
| 10 | |
| 11 | char *pname; |
| 12 | char *in_file; |
| 13 | |
| 14 | #define DATA_COUNT (1024*1024) |
| 15 | double data_items[DATA_COUNT]; |
| 16 | |
| 17 | int num_data_items = 0; |
| 18 | |
| 19 | #define BUFSIZE 1024 |
| 20 | char in_buf[BUFSIZE]; |
| 21 | |
| 22 | static int |
| 23 | compare_double(const void *p1, const void *p2) |
| 24 | { |
| 25 | double val1 = *(u_int64_t *)p1; |
| 26 | double val2 = *(u_int64_t *)p2; |
| 27 | |
| 28 | if (val1 == val2) |
| 29 | return 0; |
| 30 | if (val1 < val2) |
| 31 | return -1; |
| 32 | return 1; |
| 33 | } |
| 34 | |
| 35 | int |
| 36 | main(int argc, char **argv) |
| 37 | { |
| 38 | FILE *in_fp; |
| 39 | double sum_x = 0; |
| 40 | double sum_sq_x = 0; |
| 41 | double mean; |
| 42 | double std_dev; |
| 43 | int i; |
| 44 | int one_sd = 0; |
| 45 | int two_sd = 0; |
| 46 | int three_sd = 0; |
| 47 | double one_sd_low, one_sd_high; |
| 48 | double two_sd_low, two_sd_high; |
| 49 | double three_sd_low, three_sd_high; |
| 50 | |
| 51 | pname = argv[0]; |
| 52 | if (argc == 1) |
| 53 | in_fp = stdin; |
| 54 | else { |
| 55 | in_file = argv[1]; |
| 56 | in_fp = fopen(in_file, "r"); |
| 57 | } |
| 58 | while (fgets(in_buf, BUFSIZE, in_fp)) { |
| 59 | if (num_data_items == DATA_COUNT) { |
| 60 | fprintf(stderr, |
| 61 | "DATA overflow, increase size of data_items array\n"); |
| 62 | exit(1); |
| 63 | } |
| 64 | sscanf(in_buf, "%lf", &data_items[num_data_items]); |
| 65 | #if 0 |
| 66 | printf("%lf\n", data_items[num_data_items]); |
| 67 | #endif |
| 68 | num_data_items++; |
| 69 | } |
| 70 | if (num_data_items == 0) { |
| 71 | fprintf(stderr, "Empty input file ?\n"); |
| 72 | exit(1); |
| 73 | } |
| 74 | #if 0 |
| 75 | printf("Total items %lu\n", num_data_items); |
| 76 | #endif |
| 77 | for (i = 0 ; i < num_data_items ; i++) { |
| 78 | sum_x += data_items[i]; |
| 79 | sum_sq_x += data_items[i] * data_items[i]; |
| 80 | } |
| 81 | mean = sum_x / num_data_items; |
| 82 | printf("\tMean %.4f\n", mean); |
| 83 | std_dev = sqrt((sum_sq_x / num_data_items) - (mean * mean)); |
| 84 | printf("\tStd Dev %.4f (%.4f%% of mean)\n", |
| 85 | std_dev, (std_dev * 100.0) / mean); |
| 86 | one_sd_low = mean - std_dev; |
| 87 | one_sd_high = mean + std_dev; |
| 88 | two_sd_low = mean - (2 * std_dev); |
| 89 | two_sd_high = mean + (2 * std_dev); |
| 90 | three_sd_low = mean - (3 * std_dev); |
| 91 | three_sd_high = mean + (3 * std_dev); |
| 92 | for (i = 0 ; i < num_data_items ; i++) { |
| 93 | if (data_items[i] >= one_sd_low && |
| 94 | data_items[i] <= one_sd_high) |
| 95 | one_sd++; |
| 96 | if (data_items[i] >= two_sd_low && |
| 97 | data_items[i] <= two_sd_high) |
| 98 | two_sd++; |
| 99 | if (data_items[i] >= three_sd_low && |
| 100 | data_items[i] <= three_sd_high) |
| 101 | three_sd++; |
| 102 | } |
| 103 | printf("\tWithin 1 SD %.2f%%\n", |
| 104 | ((double)one_sd * 100) / num_data_items); |
| 105 | printf("\tWithin 2 SD %.2f%%\n", |
| 106 | ((double)two_sd * 100) / num_data_items); |
| 107 | printf("\tWithin 3 SD %.2f%%\n", |
| 108 | ((double)three_sd* 100) / num_data_items); |
| 109 | printf("\tOutside 3 SD %.2f%%\n", |
| 110 | ((double)(num_data_items - three_sd) * 100) / num_data_items); |
| 111 | /* Sort the data to get percentiles */ |
| 112 | qsort(data_items, num_data_items, sizeof(u_int64_t), compare_double); |
| 113 | printf("\t50th percentile %lf\n", data_items[num_data_items / 2]); |
| 114 | printf("\t75th percentile %lf\n", data_items[(3 * num_data_items) / 4]); |
| 115 | printf("\t90th percentile %lf\n", data_items[(9 * num_data_items) / 10]); |
| 116 | printf("\t99th percentile %lf\n", data_items[(99 * num_data_items) / 100]); |
| 117 | } |
| 118 | |