Adding support for miss benchmarks. Need to fix myfind methods.

cxxmph/Makefile.am

@@ -14,7 +14,7 @@ mph_map_test_SOURCES = mph_map_test.cc
 mph_index_test_LDADD   = libcxxmph.la
 mph_index_test_SOURCES = mph_index_test.cc
 
-bm_index_LDADD = libcxxmph.la
+bm_index_LDADD = libcxxmph.la -lcmph
 bm_index_SOURCES = bm_common.cc bm_index.cc 
 
 trigraph_test_LDADD   = libcxxmph.la

cxxmph/bm_common.cc

@@ -1,4 +1,6 @@
+#include <cmath>
 #include <fstream>
+#include <limits>
 #include <iostream>
 #include <set>
 
@@ -32,9 +34,15 @@ bool UrlsBenchmark::SetUp() {
 
 bool SearchUrlsBenchmark::SetUp() {
   if (!UrlsBenchmark::SetUp()) return false;
+  int32_t miss_ratio_int32 = std::numeric_limits<int32_t>::max() * miss_ratio_;
+  forced_miss_urls_.resize(nsearches_);
   random_.resize(nsearches_);
   for (int i = 0; i < nsearches_; ++i) {
     random_[i] = urls_[random() % urls_.size()];
+    if (random() < miss_ratio_int32) {
+      forced_miss_urls_[i] = random_[i].as_string() + ".force_miss";
+      random_[i] = forced_miss_urls_[i];
+    }
   }
   return true;
 }

cxxmph/bm_common.h

@@ -32,10 +32,12 @@ class UrlsBenchmark : public Benchmark {
 class SearchUrlsBenchmark : public UrlsBenchmark {
  public:
   SearchUrlsBenchmark(const std::string& urls_file, uint32_t nsearches)
-      : UrlsBenchmark(urls_file), nsearches_(nsearches) {}
+      : UrlsBenchmark(urls_file), nsearches_(nsearches), miss_ratio_(0.2) {}
  protected:
   virtual bool SetUp();
   const uint32_t nsearches_; 
+  float miss_ratio_;
+  std::vector<std::string> forced_miss_urls_;
   std::vector<StringPiece> random_;
 };
 

cxxmph/bm_index.cc

@@ -1,3 +1,6 @@
+#include <cmph.h>
+
+#include <cstdio>
 #include <set>
 #include <string>
 #include <unordered_map>
@@ -56,6 +59,56 @@ class BM_MPHIndexSearch : public SearchUrlsBenchmark {
   SimpleMPHIndex<StringPiece> index_;
 };
 
+class BM_CmphIndexSearch : public SearchUrlsBenchmark {
+ public:
+  BM_CmphIndexSearch(const std::string& urls_file, int nsearches)
+      : SearchUrlsBenchmark(urls_file, nsearches) { }
+  ~BM_CmphIndexSearch() { if (index_) cmph_destroy(index_); }
+  virtual void Run() {
+    for (auto it = random_.begin(); it != random_.end(); ++it) {
+      auto idx = cmph_search(index_, it->data(), it->length());
+      // Collision check to be fair with STL
+      if (strcmp(urls_[idx].c_str(), it->data()) != 0) idx = -1;
+    }
+  }
+ protected:
+  virtual bool SetUp() {
+   if (!SearchUrlsBenchmark::SetUp()) {
+      cerr << "Parent class setup failed." << endl;
+      return false;
+    }
+    FILE* f = fopen(urls_file_.c_str(), "r");
+    if (!f) {
+      cerr << "Faied to open " << urls_file_ << endl; 
+      return false;
+    }
+    cmph_io_adapter_t* source = cmph_io_nlfile_adapter(f);
+    if (!source) {
+      cerr << "Faied to create io adapter for " << urls_file_ << endl; 
+      return false;
+    }
+    cmph_config_t* config = cmph_config_new(source);
+    if (!config) {
+      cerr << "Failed to create config" << endl;
+      return false;
+    }
+    cmph_config_set_algo(config, CMPH_BDZ);
+    cmph_t* mphf = cmph_new(config);
+    if (!mphf) {
+      cerr << "Failed to create mphf." << endl;
+      return false;
+    }
+
+    cmph_config_destroy(config);
+    cmph_io_nlfile_adapter_destroy(source);
+    fclose(f);
+    index_ = mphf;
+    return true;
+  }
+  cmph_t* index_;
+};
+    
+
 class BM_STLIndexSearch : public SearchUrlsBenchmark {
  public:
   BM_STLIndexSearch(const std::string& urls_file, int nsearches)
@@ -80,10 +133,13 @@ class BM_STLIndexSearch : public SearchUrlsBenchmark {
 };
 
 int main(int argc, char** argv) {
+/*
   Benchmark::Register(new BM_MPHIndexCreate("URLS100k"));
   Benchmark::Register(new BM_STLIndexCreate("URLS100k"));
   Benchmark::Register(new BM_MPHIndexSearch("URLS100k", 100*1000*1000));
   Benchmark::Register(new BM_STLIndexSearch("URLS100k", 100*1000*1000));
+*/
+  Benchmark::Register(new BM_CmphIndexSearch("URLS100k", 100*1000*1000));
   Benchmark::RunAll();
   return 0;
 }

cxxmph/bm_map.cc

@@ -11,7 +11,9 @@
   namespace cxxmph {
 
   uint64_t myfind(const unordered_map<uint64_t, uint64_t>& mymap, const uint64_t& k) {
-    return mymap.find(k)->second;
+    auto it = mymap.find(k);
+    if (it == mymap.end()) return -1;
+    return it->second;
   }
 
   uint64_t myfind(const mph_map<uint64_t, uint64_t>& mymap, const uint64_t& k) {
@@ -19,7 +21,9 @@
   }
 
   const StringPiece& myfind(const unordered_map<StringPiece, StringPiece, Murmur2StringPiece>& mymap, const StringPiece& k) {
-    return mymap.find(k)->second;
+    auto it = mymap.find(k);
+    if (it == mymap.end()) return ".force_miss";
+    return it->second;
   }
   StringPiece myfind(const mph_map<StringPiece, StringPiece>& mymap, const StringPiece& k) {
     auto it = mymap.find(k);
@@ -44,13 +48,22 @@ class BM_SearchUrls : public SearchUrlsBenchmark {
   BM_SearchUrls(const std::string& urls_file, int nsearches) 
       : SearchUrlsBenchmark(urls_file, nsearches) { }
   virtual void Run() {
+    fprintf(stderr, "Running benchmark\n");
     for (auto it = random_.begin(); it != random_.end(); ++it) {
+      if (it->ends_with(".force_miss")) {
+        fprintf(stderr, "About to miss\n");
+      } else {
+        fprintf(stderr, "No miss\n");
+      }
+      fprintf(stderr, "it: *%s\n", it->as_string().c_str());
       auto v = myfind(mymap_, *it);
-      if (v != *it) {
+      fprintf(stderr, "v: %s, it: *%s\n", v.as_string().c_str(), it->as_string().c_str());
+      if (v != *it && !it->ends_with(".force_miss")) {
         fprintf(stderr, "Looked for %s got %s\n", it->data(), v.data());
 	exit(-1);
       }
     }
+    fprintf(stderr, "Done running benchmark\n");
   }
  protected:
   virtual bool SetUp() {
@@ -102,8 +115,8 @@ int main(int argc, char** argv) {
   Benchmark::Register(new BM_CreateUrls<mph_map<StringPiece, StringPiece>>("URLS100k"));
   Benchmark::Register(new BM_CreateUrls<unordered_map<StringPiece, StringPiece>>("URLS100k"));
   */
-  Benchmark::Register(new BM_SearchUrls<mph_map<StringPiece, StringPiece>>("URLS100k", 10*1000* 1000));
+  // Benchmark::Register(new BM_SearchUrls<mph_map<StringPiece, StringPiece>>("URLS100k", 10*1000 * 1000));
-  Benchmark::Register(new BM_SearchUrls<unordered_map<StringPiece, StringPiece, Murmur2StringPiece>>("URLS100k", 10*1000* 1000));
+  Benchmark::Register(new BM_SearchUrls<unordered_map<StringPiece, StringPiece, Murmur2StringPiece>>("URLS100k", 10));
   /*
   Benchmark::Register(new BM_SearchUint64<unordered_map<uint64_t, uint64_t>>);
   Benchmark::Register(new BM_SearchUint64<mph_map<uint64_t, uint64_t>>);