|
| 1 | +#include <iostream> |
| 2 | +#include <limits> |
| 3 | +#include <sdsl/vectors.hpp> |
| 4 | +#include <sdsl/coder.hpp> |
| 5 | + |
| 6 | +/**** Benchmark for self - delimiting codes *********************************** |
| 7 | +For information about usage of this benchmark, see displayUsage - function. |
| 8 | +
|
| 9 | +To compile this benchmark, the following macros have to be defined |
| 10 | +(e.g. by passing them to compiler): |
| 11 | +- VTYPES: a comma - separated list of sdsl vector types to be testet, |
| 12 | + e.g. vlc_vector<coder::elias_gamma>,vlc_vector<coder::elias_delta> |
| 13 | +- VNAMES: symbolic names of the corresponding vector types, in same order |
| 14 | + as in macro VTYPES, defined as a character array. |
| 15 | + According to the upper sample on macro VTYPES, VNAMES could be defined as |
| 16 | + {"VLC Vector with Elias Gamma Coder","VLC Vector with Elias Delta Coder"} |
| 17 | +*/ |
| 18 | + |
| 19 | + |
| 20 | +//assert that needed macros are defined |
| 21 | +#ifndef VTYPES |
| 22 | +#error "Macro VTYPES with comma - separated list of vector types has to be \ |
| 23 | + defined for compiling benchmark" |
| 24 | +#endif |
| 25 | +
|
| 26 | +#ifndef VNAMES |
| 27 | +#error "Macro VNAMES with an array of characters has to be \ |
| 28 | + defined for compiling benchmark" |
| 29 | +#endif |
| 30 | +
|
| 31 | +using namespace std; |
| 32 | +using namespace sdsl; |
| 33 | +using namespace std::chrono; |
| 34 | +using timer = std::chrono::high_resolution_clock; |
| 35 | +
|
| 36 | +const char *(vectornames[]) = VNAMES; |
| 37 | +const size_t vectorcount = sizeof(vectornames) / sizeof(vectornames[0]); |
| 38 | +
|
| 39 | +struct iv_testresult { //testcase for one defined int vector |
| 40 | + double enc_MBperSec; //encoding rate: megabytes per second |
| 41 | + double dec_MBperSec; //decoding rate: megabytes per second |
| 42 | + double comp_percent; //compression rate: needed space in percentage compared |
| 43 | + //to original integer vector |
| 44 | +}; |
| 45 | +
|
| 46 | +//benchmark method declaration |
| 47 | +template<class... Vectors> //used vectors for benchmark |
| 48 | +bool runTestcase( const int_vector<> &iv, iv_testresult *result ); |
| 49 | +
|
| 50 | +//stuff for nice printing |
| 51 | +void displayUsage(const char *pname); |
| 52 | +void displayHeading(); |
| 53 | +void displayResult( const char *testcase, const iv_testresult *result ); |
| 54 | +
|
| 55 | +int main(const int argc, const char **argv) |
| 56 | +{ |
| 57 | + //check args |
| 58 | + if ((argc - 1) % 3 != 0) { |
| 59 | + displayUsage(argv[0]); |
| 60 | + return 1; |
| 61 | + } |
| 62 | +
|
| 63 | + //set up needed structures |
| 64 | + const size_t testcasecount = (argc - 1) / 3; |
| 65 | + iv_testresult overallresult[vectorcount]; |
| 66 | + |
| 67 | + //prepare overall result |
| 68 | + for (size_t i = 0; i < vectorcount; i++) { |
| 69 | + overallresult[i].enc_MBperSec = 0.0; |
| 70 | + overallresult[i].dec_MBperSec = 0.0; |
| 71 | + overallresult[i].comp_percent = 0.0; |
| 72 | + } |
| 73 | +
|
| 74 | + //start fetching test cases and run benchmark |
| 75 | + displayHeading(); |
| 76 | + for (size_t i = 0; i < testcasecount; i++) { |
| 77 | + const char *testcase = argv[3*i + 1]; |
| 78 | + const char *file = argv[3*i + 2]; //file of saved vector |
| 79 | + const char *type = argv[3*i + 3]; //type of saved vector |
| 80 | + uint8_t v_type = type[0]=='d' ? 'd' : type[0] - '0'; |
| 81 | +
|
| 82 | + //load vector |
| 83 | + int_vector<> iv; |
| 84 | + if (!load_vector_from_file(iv, file, v_type)) { |
| 85 | + cerr << "ERROR: vector from file " << file |
| 86 | + << " with type " << type << " could not be loaded" |
| 87 | + << endl; |
| 88 | + displayUsage(argv[0]); |
| 89 | + return 1; |
| 90 | + } |
| 91 | +
|
| 92 | + //run test |
| 93 | + iv_testresult result[vectorcount]; |
| 94 | + if (!runTestcase<VTYPES>( iv, result )) { |
| 95 | + cerr << "Testcase " << testcase << "failed" << endl; |
| 96 | + return 1; |
| 97 | + } |
| 98 | +
|
| 99 | + //print result |
| 100 | + displayResult( testcase, result ); |
| 101 | +
|
| 102 | + //and sum up results for overall result |
| 103 | + for (size_t j = 0; j < vectorcount; j++) { |
| 104 | + overallresult[j].enc_MBperSec += result[j].enc_MBperSec; |
| 105 | + overallresult[j].dec_MBperSec += result[j].dec_MBperSec; |
| 106 | + overallresult[j].comp_percent += result[j].comp_percent; |
| 107 | + } |
| 108 | + } |
| 109 | +
|
| 110 | + //build average for overall result |
| 111 | + for (size_t i = 0; i < vectorcount; i++) { |
| 112 | + overallresult[i].enc_MBperSec /= testcasecount; |
| 113 | + overallresult[i].dec_MBperSec /= testcasecount; |
| 114 | + overallresult[i].comp_percent /= testcasecount; |
| 115 | + } |
| 116 | +
|
| 117 | + //and display overall results |
| 118 | + displayResult( "Overall", overallresult ); |
| 119 | + return 0; |
| 120 | +} |
| 121 | +
|
| 122 | +//// BENCHMARK METHODS //////////////////////////////////////////////////////// |
| 123 | +template<class Vector> //used compression vector type |
| 124 | +bool runSingleTest( const int_vector<> &testcase, iv_testresult &result ) { |
| 125 | + //test encoding rate by constructing Vector |
| 126 | + auto start = timer::now(); |
| 127 | + Vector test( testcase ); |
| 128 | + auto stop = timer::now(); |
| 129 | + result.enc_MBperSec = size_in_mega_bytes( testcase ) |
| 130 | + / duration_cast<seconds>(stop-start).count(); |
| 131 | +
|
| 132 | + //care for compression rate |
| 133 | + result.comp_percent = size_in_mega_bytes(test) |
| 134 | + / size_in_mega_bytes(testcase) * 100.0; |
| 135 | +
|
| 136 | + //and finally for decoding rate |
| 137 | + //use a trick to decode all values: since (currently) all vectors are |
| 138 | + //using sample tables, access the element right before the next sampling |
| 139 | + //entry, so everything between 2 samples has to be decoded. |
| 140 | + size_t sample_dens = test.get_sample_dens(); |
| 141 | + start = timer::now(); |
| 142 | + //repeat test 5 times to avoid infinite decoding rates |
| 143 | + for (size_t j = 0; j < 5; j++) { |
| 144 | + size_t i = sample_dens - 1; |
| 145 | + for (; i < test.size(); i += sample_dens) { |
| 146 | + test[i]; //acess element right before next sample entry |
| 147 | + } |
| 148 | + //and finally access last element if not done yet |
| 149 | + if (i != test.size() + sample_dens - 1) |
| 150 | + test[test.size() - 1]; |
| 151 | + } |
| 152 | + stop = timer::now(); |
| 153 | + result.dec_MBperSec = size_in_mega_bytes( testcase ) |
| 154 | + / duration_cast<seconds>(stop-start).count() |
| 155 | + * 5.0; //multiply with 5 since vector was decoded 5 times |
| 156 | +
|
| 157 | + return true; //may use this return type for error detection in future |
| 158 | +} |
| 159 | +
|
| 160 | +template<class... Vectors> //used vectors for benchmark |
| 161 | +bool runTestcase( const int_vector<> &testcase, iv_testresult *result ) { |
| 162 | + size_t i = 0; |
| 163 | + //do variadic template pack expansion |
| 164 | + bool testfine[] = { runSingleTest<Vectors>( testcase, result[i++] )... }; |
| 165 | + bool testsfine = true; |
| 166 | + for (i = 0; i < vectorcount; i++) { |
| 167 | + if (!testfine[i]) { |
| 168 | + cerr << "Test on Vector " << vectornames[i] |
| 169 | + << "failed" << endl; |
| 170 | + testsfine = false; |
| 171 | + } |
| 172 | + } |
| 173 | + return testsfine; |
| 174 | +} |
| 175 | +
|
| 176 | +//// DISPLAYING OF RESULTS //////////////////////////////////////////////////// |
| 177 | +
|
| 178 | +void displayUsage(const char *pname) { |
| 179 | + cerr << "USAGE: " << pname << " [testcase file vectortype]*" |
| 180 | + << endl; |
| 181 | + cerr << "DESCRIPTION:" << endl; |
| 182 | + cerr << "\tThis Program runs a benchmark on self-delimiting " |
| 183 | + << "Codes." << endl; |
| 184 | + cerr << "\tProgram needs triples of parameters " |
| 185 | + << "for each test case, see Parameter section." << endl; |
| 186 | + cerr << "\tProgram will test a couple of compression vectors " |
| 187 | + << endl |
| 188 | + << "\ton measured encoding and decoding rates," << endl |
| 189 | + << "\tplus the compression rate in percent " |
| 190 | + << "(compared to the original integer vector)" << endl |
| 191 | + << "\tfor each testcase." |
| 192 | + << endl |
| 193 | + << "\tAdditionally, an overall result on different " |
| 194 | + << endl << "\tcompression vectors is printed." << endl; |
| 195 | + cerr << "\tThe generated output uses a CSV format, so " |
| 196 | + << "you may save it to a csv file for better visability" |
| 197 | + << endl << "\tand other utilites." << endl; |
| 198 | + cerr << "PARAMETERS: The parameters have to be passed as " |
| 199 | + << " triples for each test case." << endl |
| 200 | + << "\tA Triple consist of " << endl |
| 201 | + << "\t\t- testcase: A name for the test case" << endl |
| 202 | + << "\t\t- file: a path to the file where the test case" << endl |
| 203 | + << "\t\t\t(an integer vector) is contained" << endl |
| 204 | + << "\t\t- vectortype: type of saved integer vector" << endl |
| 205 | + << "\t\t\t0: serialized int_vector<>" << endl |
| 206 | + << "\t\t\t1: byte sequence" << endl |
| 207 | + << "\t\t\t2: 16-bit word sequence" << endl |
| 208 | + << "\t\t\t4: 32-bit word sequence" << endl |
| 209 | + << "\t\t\t8: 64-bit word sequence" << endl |
| 210 | + << "\t\t\td: Parse decimal numbers" << endl; |
| 211 | + cerr << "TESTET COMPRESSION VECTORS:" << endl; |
| 212 | + for (size_t i = 0; i < vectorcount; i++) { |
| 213 | + cerr << "\t- " << vectornames[i] << endl; |
| 214 | + } |
| 215 | +} |
| 216 | +void displayHeading() { |
| 217 | + cout << left; //left justify |
| 218 | + //add a comment how to read values |
| 219 | + cout << "# encoding / decoding rate unit: MB/s" << endl; |
| 220 | + cout << "# compression : percentage of needed space " |
| 221 | + << " compared to original vector" << endl; |
| 222 | + //and print a header for csv output |
| 223 | + cout << setw(20) << "testcase" |
| 224 | + << setw(1) << ";" << setw(20) << "vector" |
| 225 | + << setw(1) << ";" << setw(20) << "encodingrate" |
| 226 | + << setw(1) << ";" << setw(20) << "decodingrate" |
| 227 | + << setw(1) << ";" << "compressionrate" << endl; |
| 228 | +} |
| 229 | +
|
| 230 | +void displayResult( const char *testcase, const iv_testresult *result ) { |
| 231 | + cout << left << fixed; //prepare cout |
| 232 | + for (size_t i = 0; i < vectorcount; i++) { |
| 233 | + cout << setw(20) << testcase |
| 234 | + << setw(1) << ";" << setw(20) << vectornames[i] |
| 235 | + << setw(1) << ";" << setw(20) << result[i].enc_MBperSec |
| 236 | + << setw(1) << ";" << setw(20) << result[i].dec_MBperSec |
| 237 | + << setw(1) << ";" << result[i].comp_percent << endl; |
| 238 | + } |
| 239 | +} |
0 commit comments