source: doc/theses/mike_brooks_MMath/benchmarks/string/prog-find.cfa@ 10ef475

#if defined IMPL_STL_NA_NA
  #define IMPL_STL
#endif

#if defined IMPL_BUHR94_NA_NA
  #define IMPL_BUHR94
#endif

#if defined IMPL_STL
  #include <string>
  #include <iostream>
  #include <cstdio>
  using namespace std;
  #define IMPL_CXX

#elif defined IMPL_CFA_HL_SHARE
  #define IMPL_CFA_HL
  #define IMPL_CFA

#elif defined IMPL_CFA_LL_SHARE
  #define IMPL_CFA_LL
  #define IMPL_CFA

#elif defined IMPL_CFA_HL_NOSHARE
  #define IMPL_CFA_HL
  #define CFA_NOSHARE
  #define IMPL_CFA

#elif defined IMPL_CFA_LL_NOSHARE
  #define IMPL_CFA_LL
  #define CFA_NOSHARE
  #define IMPL_CFA

#elif defined IMPL_BUHR94
  #include <iostream>
  #include <cstdio>
  #include "/u0/mlbrooks/usys1/sm/string/StringSharing/src/string.h"
  #define IMPL_CXX

#else
  #error Bad IMPL_
#endif


#if defined IMPL_CFA_HL
  #include <string.hfa>
#elif defined IMPL_CFA_LL
  #include <string_res.hfa>
  #include "not_string_res.hfa"
#endif

#if defined CFA_NOSHARE
  #include <string_sharectx.hfa>
  #define STRING_SHARING_CONTROL \
    string_sharectx c = { NO_SHARING };
#else 
  #define STRING_SHARING_CONTROL
#endif

#if defined IMPL_CFA
  #include <math.hfa>
  extern "C" {
    void malloc_stats( void );
  }
#elif defined IMPL_CXX
  #include <algorithm>
  using std::min;
  #include <malloc.h>
#endif

#include <time.h>
#include <stdlib.h> // atoi
#include <string.h> // strlen, only during setup
#include <limits.h> // LONG_MAX

#if defined IMPL_STL || defined IMPL_BUHR94
    #define PRINT(s) std::cout << s << std::endl
#elif defined IMPL_CFA_HL || defined IMPL_CFA_LL
    #define PRINT(s) sout | s;
#else
    #error Unhandled print case
#endif

#if defined IMPL_CFA_LL
    #define STRING_T string_res
    #define ASSIGN_CHAR(str, idx, val) assignAt(str, idx, val)
#else
    #define STRING_T string
    #define ASSIGN_CHAR(str, idx, val) str[idx] = val
#endif

double meanLen(int N, char ** strings) {
    int totalLen = 0;
    for (int i = 0 ; i < N; i ++) {
        totalLen += strlen(strings[i]);
    }
    return (double)totalLen / (double)N;
}

volatile int checkthis = 0;
#define MAYBE( op ) if (checkthis) { op; }


int main( int argc, char ** argv ) {

    STRING_SHARING_CONTROL


    const char * usage_args[] = {"(Ignored)    ExecTimeSecs   Corpus...",
                                 "(Ignored) -w WorkAllocCount Corpus..."};

    const int static_arg_posns = 3;
    int used_arg_posns = static_arg_posns;

    int execTimeSecs = -1;

    long int iterationCountTarget = -1;

    switch (min(argc, static_arg_posns)) {
      case 3: if ( strcmp(argv[2], "-w") == 0 ) {
                used_arg_posns ++;
                execTimeSecs = 0;
                iterationCountTarget = atoi(argv[3]);
              } else {
                execTimeSecs = atoi(argv[2]);
                iterationCountTarget = LONG_MAX;
              }
    }

    int corpuslen = argc - used_arg_posns;
    char ** corpus = argv + used_arg_posns;

    if ((execTimeSecs < 1 && iterationCountTarget < 1)  || corpuslen < 1) {
      for (int u = 0; u < sizeof(usage_args) / sizeof(*usage_args); u++) {
        printf("usage: %s %s\n", argv[0], usage_args[u]);
      }
      printf("output:\nxxx,corpusItemCount,corpusMeanLenChars,callDoneActualCount,execTimeActualSec\n");
      exit(1);
    }

    double meanCorpusLen = meanLen(corpuslen, corpus);

    clock_t start, end_target, end_actual;

    STRING_T corpus_imported[corpuslen];

    for (int i = 0; i < corpuslen; i++) {
        corpus_imported[i] = corpus[i];
        // if the callee ever modifies, then we will drive GCs, which will visit the entire corpus
    }

    start = clock();
    if (execTimeSecs != 0) {
      end_target = start + CLOCKS_PER_SEC * execTimeSecs;
    } else {
      end_target = start + CLOCKS_PER_SEC * 3600;
    }

    unsigned int t = 0;
    for ( ; t < iterationCountTarget && ((t+1) % 10000 != 0 || clock() < end_target) ; t += 1 ) {
            #if defined OP_PFI
              size_t foundAt =
                #if defined IMPL_CFA
                   find( corpus_imported[t % corpuslen], ' ')
                #elif defined IMPL_STL
                   corpus_imported[t % corpuslen].find(' ')
                #else
                  #error Bad IMPL
                #endif
                ;
            #else
              #error Bad OP_
            #endif
            MAYBE( PRINT( foundAt ) );
    }
    end_actual = clock();
    unsigned int callsDone = t;
    double elapsed = ((double) (end_actual - start)) / CLOCKS_PER_SEC;
    printf("xxx,%d,%f,%d,%f\n", corpuslen, meanCorpusLen, t, elapsed);

    // malloc_stats();

    return 0;
}