All compiles in the mac.

cxxmph/Makefile.am

@@ -1,6 +1,6 @@
 AM_CXXFLAGS='-std=c++0x'
 TESTS = $(check_PROGRAMS)
-check_PROGRAMS = mph_map_test mph_index_test trigraph_test
+check_PROGRAMS = mph_map_test mph_index_test # trigraph_test
 noinst_PROGRAMS = bm_index bm_map
 bin_PROGRAMS = cxxmph
 lib_LTLIBRARIES = libcxxmph.la
@@ -15,8 +15,8 @@ mph_map_test_SOURCES = mph_map_test.cc
 mph_index_test_LDADD   = libcxxmph.la
 mph_index_test_SOURCES = mph_index_test.cc
 
-trigraph_test_LDADD   = libcxxmph.la
+# trigraph_test_LDADD   = libcxxmph.la
-trigraph_test_SOURCES = trigraph_test.cc
+# trigraph_test_SOURCES = trigraph_test.cc
 
 bm_index_LDADD = libcxxmph.la
 bm_index_SOURCES = bm_common.cc bm_index.cc 

cxxmph/benchmark.cc

@@ -3,15 +3,22 @@
 #include <cerrno>
 #include <cstring>
 #include <cstdio>
+#include <memory>
 #include <sys/time.h>
 #include <sys/resource.h>
 
+#include <iomanip>
 #include <iostream>
+#include <sstream>
 #include <vector>
 
 using std::cerr;
+using std::cout;
 using std::endl;
+using std::setfill;
+using std::setw;
 using std::string;
+using std::ostringstream;
 using std::vector;
 
 namespace {
@@ -42,6 +49,14 @@ int timeval_subtract (
   return x->tv_sec < y->tv_sec;
 }
 
+// C++ iostream is terrible for formatting.
+string timeval_to_string(timeval tv) {
+  ostringstream out;
+  out << setfill(' ') << setw(3) << tv.tv_sec << '.';
+  out << setfill('0') << setw(6) << tv.tv_usec;
+  return out.str();
+}
+
 struct rusage getrusage_or_die() {
   struct rusage rs;
   int ret = getrusage(RUSAGE_SELF, &rs);
@@ -92,11 +107,14 @@ namespace cxxmph {
 
 /* static */ void Benchmark::RunAll() {
   for (int i = 0; i < g_benchmarks.size(); ++i) {
-    Benchmark* bm = g_benchmarks[i];
+    std::auto_ptr<Benchmark> bm(g_benchmarks[i]);
-    bm->SetUp();
+    if (!bm->SetUp()) {
+      cerr << "Set up phase for benchmark "
+           << bm->name() << " failed." << endl;
+      continue;
+    }
     bm->MeasureRun();
     bm->TearDown(); 
-    delete bm;
   }
 }
 
@@ -114,11 +132,11 @@ void Benchmark::MeasureRun() {
   struct timeval wtime;
   timeval_subtract(&wtime, &walltime_end, &walltime_begin);
 
-  printf("Benchmark: %s\n", name().c_str());
+  cout << "Benchmark: " << name_ << endl;
-  printf("CPU User time  : %ld.%06ld\n", utime.tv_sec, utime.tv_usec);
+  cout << "CPU User time  : " << timeval_to_string(utime) << endl;
-  printf("CPU System time: %ld.%06ld\n", stime.tv_sec, stime.tv_usec);
+  cout << "CPU System time: " << timeval_to_string(stime) << endl;
-  printf("Wall clock time: %ld.%06ld\n", wtime.tv_sec, wtime.tv_usec);
+  cout << "Wall clock time: " << timeval_to_string(wtime) << endl;
-  printf("\n");
+  cout << endl;
 }
 
 }  // namespace cxxmph

cxxmph/bm_map.cc

@@ -29,11 +29,12 @@ class BM_MapSearch : public SearchUrlsBenchmark {
       : SearchUrlsBenchmark(urls_file, nsearches) { }
   virtual void Run() {
     for (auto it = random_.begin(); it != random_.end(); ++it) {
-      auto value = mymap_[it->ToString()];
+      mymap_.find(*it);
     }
   }
  protected:
   virtual bool SetUp() {
+    if (!SearchUrlsBenchmark::SetUp()) return false;
     for (auto it = urls_.begin(); it != urls_.end(); ++it) {
       mymap_[*it] = *it;
     }
@@ -48,9 +49,9 @@ class BM_MapSearch : public SearchUrlsBenchmark {
 using namespace cxxmph;
 
 int main(int argc, char** argv) {
-  Benchmark::Register(new BM_MapCreate<mph_map<string, string>>("URLS100k"));
+  Benchmark::Register(new BM_MapCreate<mph_map<StringPiece, StringPiece>>("URLS100k"));
-  Benchmark::Register(new BM_MapCreate<unordered_map<string, string>>("URLS100k"));
+  Benchmark::Register(new BM_MapCreate<unordered_map<StringPiece, StringPiece>>("URLS100k"));
-  Benchmark::Register(new BM_MapSearch<mph_map<string, string>>("URLS100k", 1000 * 1000));
+  Benchmark::Register(new BM_MapSearch<mph_map<StringPiece, StringPiece>>("URLS100k", 1000* 1000));
-  Benchmark::Register(new BM_MapSearch<unordered_map<string, string>>("URLS100k", 1000 * 1000));
+  Benchmark::Register(new BM_MapSearch<unordered_map<StringPiece, StringPiece>>("URLS100k", 1000* 1000));
   Benchmark::RunAll();
 }

cxxmph/mph_index.h

@@ -93,6 +93,10 @@ class MPHIndex {
 // Template method needs to go in the header file.
 template <class SeededHashFcn, class ForwardIterator>
 bool MPHIndex::Reset(ForwardIterator begin, ForwardIterator end) {
+  if (end == begin) {
+    clear();
+    return true;
+  }
   m_ = end - begin;
   r_ = static_cast<uint32_t>(ceil((c_*m_)/3));
   if ((r_ % 2) == 0) r_ += 1;

cxxmph/seeded_hash.h

@@ -60,6 +60,8 @@ template <> struct seeded_hash<std::tr1::hash<const char*> >
 { typedef seeded_hash_function<Murmur2StringPiece> hash_function; };
 template <> struct seeded_hash<std::tr1::hash<std::string> >
 { typedef seeded_hash_function<Murmur2StringPiece> hash_function; };
+template <> struct seeded_hash<std::tr1::hash<cxxmph::StringPiece> >
+{ typedef seeded_hash_function<Murmur2StringPiece> hash_function; };
 
 template <> struct seeded_hash<std::tr1::hash<char> >
 { typedef seeded_hash_function<Murmur2> hash_function; };

cxxmph/stringpiece.h

@@ -145,32 +145,34 @@ class StringPiece {
   StringPiece substr(size_type pos, size_type n = npos) const;
 };
 
-}  // namespace cxxmph
+inline bool operator==(const StringPiece& x, const StringPiece& y) {
+  return x.length() == y.length() && memcmp(x.data(), y.data(), x.length()) == 0;
+}
 
-bool operator==(const cxxmph::StringPiece& x, const cxxmph::StringPiece& y);
+inline bool operator!=(const StringPiece& x, const StringPiece& y) {
-
-inline bool operator!=(const cxxmph::StringPiece& x, const cxxmph::StringPiece& y) {
   return !(x == y);
 }
 
-inline bool operator<(const cxxmph::StringPiece& x, const cxxmph::StringPiece& y) {
+inline bool operator<(const StringPiece& x, const StringPiece& y) {
   const int r = memcmp(x.data(), y.data(),
                        std::min(x.size(), y.size()));
   return ((r < 0) || ((r == 0) && (x.size() < y.size())));
 }
 
-inline bool operator>(const cxxmph::StringPiece& x, const cxxmph::StringPiece& y) {
+inline bool operator>(const StringPiece& x, const StringPiece& y) {
   return y < x;
 }
 
-inline bool operator<=(const cxxmph::StringPiece& x, const cxxmph::StringPiece& y) {
+inline bool operator<=(const StringPiece& x, const StringPiece& y) {
   return !(x > y);
 }
 
-inline bool operator>=(const cxxmph::StringPiece& x, const cxxmph::StringPiece& y) {
+inline bool operator>=(const StringPiece& x, StringPiece& y) {
   return !(x < y);
 }
 
+}  // namespace cxxmph
+
 // allow StringPiece to be logged
 extern std::ostream& operator<<(std::ostream& o, const cxxmph::StringPiece& piece);
 

src/bm_numbers.c

@@ -1,10 +1,6 @@
 #include <stdlib.h>
 #include <string.h>
 
-#include <ext/hash_set>
-using __gnu_cxx::hash_set;
-static const char cxx_name = "__gnu_cxx::hash_set";
-
 #include "bitbool.h"
 #include "cmph.h"
 #include "cmph_benchmark.h"
@@ -71,12 +67,12 @@ void bm_create(CMPH_ALGO algo, int iters) {
 
 void bm_search(CMPH_ALGO algo, int iters) {
   int i = 0;
-  char mphf_name[128];
+  char *mphf_name;
   cmph_t* mphf = NULL; 
 
-  
+  mphf_name = create_lsmap_key(algo, iters);
-  snprintf(mphf_name, 128, "%s:%u", cxx_name, iters);
   mphf = (cmph_t*)lsmap_search(g_created_mphf, mphf_name);
+  free(mphf_name);
 
   cmph_uint32* count = (cmph_uint32*)malloc(sizeof(cmph_uint32)*iters);  
   cmph_uint32* hash_count = (cmph_uint32*)malloc(sizeof(cmph_uint32)*iters);  
@@ -106,49 +102,6 @@ DECLARE_ALGO(CMPH_BRZ);
 DECLARE_ALGO(CMPH_FCH);
 DECLARE_ALGO(CMPH_BDZ);
 
-void bm_create_ext_hash_set(int iters) {
-  cmph_uint32 i = 0;
-
-  if (iters > g_numbers_len) {
-    fprintf(stderr, "No input with proper size.");
-    exit(-1);
-  }
-
-  hash_set<cmph_uint32>* ext_hash_set = new hash_set<cmph_uint32>;
-  for (i = 0; i < iters; ++i) {
-    ext_hash_set->insert(g_numbers[i]);
-  }
-  lsmap_append(g_created_mphf, cxx_name, ext_hash_set);
-}
-
-void bm_search_ext_hash_set(int iters) {
-  cmph_uint32 i = 0;
-
-  if (iters > g_numbers_len) {
-    fprintf(stderr, "No input with proper size.");
-    exit(-1);
-  }
-
-  snprintf(mphf_name, 128, "%s:%u", hash_count, iters);
-  mphf = (__gnu_cxx::hash_set*)lsmap_search(g_created_mphf, mphf_name);
-
-  cmph_uint32* count = (cmph_uint32*)malloc(sizeof(cmph_uint32)*iters);  
-  cmph_uint32* hash_count = (cmph_uint32*)malloc(sizeof(cmph_uint32)*iters);  
-
-  for (i = 0; i < iters * 100; ++i) {
-    cmph_uint32 pos = random() % iters;
-    const char* buf = (const char*)(g_numbers + pos);
-    cmph_uint32 h = cmph_search(mphf, buf, sizeof(cmph_uint32));
-    ++count[pos];
-    ++hash_count[h];
-  }
-
-  // Verify correctness later.
-  lsmap_append(g_expected_probes, create_lsmap_key(algo, iters), count);
-  lsmap_append(g_mphf_probes, create_lsmap_key(algo, iters), hash_count);
-}
-}
-
 int main(int argc, char** argv) {
   g_numbers_len = 1000 * 1000;
   g_numbers = random_numbers_vector_new(g_numbers_len);
@@ -162,8 +115,8 @@ int main(int argc, char** argv) {
   BM_REGISTER(bm_search_CMPH_CHM, 1000 * 1000);
 //  BM_REGISTER(bm_create_CMPH_BRZ, 1000 * 1000);
 //  BM_REGISTER(bm_search_CMPH_BRZ, 1000 * 1000);
-  BM_REGISTER(bm_create_CMPH_FCH, 1000 * 1000);
+//  BM_REGISTER(bm_create_CMPH_FCH, 1000 * 1000);
-  BM_REGISTER(bm_search_CMPH_FCH, 1000 * 1000);
+//  BM_REGISTER(bm_search_CMPH_FCH, 1000 * 1000);
   BM_REGISTER(bm_create_CMPH_BDZ, 1000 * 1000);
   BM_REGISTER(bm_search_CMPH_BDZ, 1000 * 1000);
   run_benchmarks(argc, argv);

src/cmph_benchmark.c

@@ -111,8 +111,10 @@ void bm_end(const char* name) {
   timeval_subtract(&stime, &benchmark->end.ru_stime, &benchmark->begin.ru_stime);
   
   printf("Benchmark: %s\n", benchmark->name);
-  printf("User time used  : %ld.%06ld\n", utime.tv_sec, utime.tv_usec);
+  printf("User time used  : %ld.%06ld\n",
-  printf("System time used: %ld.%06ld\n", stime.tv_sec, stime.tv_usec);
+         utime.tv_sec, (long int)utime.tv_usec);
+  printf("System time used: %ld.%06ld\n",
+         stime.tv_sec, (long int)stime.tv_usec);
   printf("\n");
 }