From 6afc7cf1054f8f7602fa036d208af728aaf3c446 Mon Sep 17 00:00:00 2001
From: Davi Reis <davi@davi-macbookpro.local>
Date: Sat, 21 Apr 2012 21:48:32 +0200
Subject: [PATCH] Fastest true incarnation so far. Not much faster than
 unordered_map.

---
 cxxmph/bm_map.cc   |  7 ++++---
 cxxmph/mph_index.h | 14 +++++++-------
 cxxmph/mph_map.h   | 21 ++++++++++++++++-----
 3 files changed, 27 insertions(+), 15 deletions(-)

diff --git a/cxxmph/bm_map.cc b/cxxmph/bm_map.cc
index 039b5d6..0f8dace 100644
--- a/cxxmph/bm_map.cc
+++ b/cxxmph/bm_map.cc
@@ -41,11 +41,12 @@ class BM_SearchUrls : public SearchUrlsBenchmark {
       : SearchUrlsBenchmark(urls_file, nsearches, miss_ratio) { }
   virtual ~BM_SearchUrls() {}
   virtual void Run() {
+    uint32_t total = 1;
     for (auto it = random_.begin(); it != random_.end(); ++it) {
-      myfind(mymap_, *it);
-      assert(it->ends_with(".force_miss") ^ v != NULL);
-      assert(!v || *v == *it);
+      auto v = myfind(mymap_, *it);
+      if (v) total += v->length();
     }
+    fprintf(stderr, "Total: %u\n", total);
   }
  protected:
   virtual bool SetUp() {
diff --git a/cxxmph/mph_index.h b/cxxmph/mph_index.h
index 1f6ace1..84b7258 100644
--- a/cxxmph/mph_index.h
+++ b/cxxmph/mph_index.h
@@ -124,7 +124,7 @@ bool MPHIndex::Reset(
   if ((r_ % 2) == 0) r_ += 1;
   // This can be used to speed mods, but increases occupation too much. 
   // Needs to try http://gmplib.org/manual/Integer-Exponentiation.html instead
-  // r_ = nextpoweroftwo(r_);
+  r_ = nextpoweroftwo(r_);
   nest_displacement_[0] = 0;
   nest_displacement_[1] = r_;
   nest_displacement_[2] = (r_ << 1);
@@ -177,12 +177,12 @@ template <class SeededHashFcn, class Key>
 uint32_t MPHIndex::perfect_hash(const Key& key) const {
   if (!g_.size()) return 0;
   h128 h = SeededHashFcn().hash128(key, hash_seed_[0]);
-  h[0] = (h[0] % r_) + nest_displacement_[0];
-  h[1] = (h[1] % r_) + nest_displacement_[1];
-  h[2] = (h[2] % r_) + nest_displacement_[2];
-  // h[0] = (h[0] & (r_-1)) + nest_displacement_[0];
-  // h[1] = (h[1] & (r_-1)) + nest_displacement_[1];
-  // h[2] = (h[2] & (r_-1)) + nest_displacement_[2];
+  // h[0] = (h[0] % r_) + nest_displacement_[0];
+  // h[1] = (h[1] % r_) + nest_displacement_[1];
+  // h[2] = (h[2] % r_) + nest_displacement_[2];
+  h[0] = (h[0] & (r_-1)) + nest_displacement_[0];
+  h[1] = (h[1] & (r_-1)) + nest_displacement_[1];
+  h[2] = (h[2] & (r_-1)) + nest_displacement_[2];
   assert((h[0]) < g_.size());
   assert((h[1]) < g_.size());
   assert((h[2]) < g_.size());
diff --git a/cxxmph/mph_map.h b/cxxmph/mph_map.h
index 4d0276f..63095c5 100644
--- a/cxxmph/mph_map.h
+++ b/cxxmph/mph_map.h
@@ -79,6 +79,7 @@ class mph_map {
   inline const_iterator find(const key_type& k) const;
   typedef int32_t my_int32_t;  // help macros
   inline int32_t index(const key_type& k) const;
+  inline int32_t index2(const key_type& k) const;
   data_type& operator[](const key_type &k);
   const data_type& operator[](const key_type &k) const;
 
@@ -197,7 +198,7 @@ MPH_MAP_METHOD_DECL(void_type, erase)(const key_type& k) {
 }
 
 MPH_MAP_INLINE_METHOD_DECL(const_iterator, find)(const key_type& k) const {
-  auto idx = index(k);
+  auto idx = index2(k);
   typename vector<value_type>::const_iterator vit = values_.begin() + idx;
   if (idx == -1 || vit->first != k) return end();
   return make_solid(&values_, &present_, vit);;
@@ -210,17 +211,27 @@ MPH_MAP_INLINE_METHOD_DECL(iterator, find)(const key_type& k) {
   return make_solid(&values_, &present_, vit);;
 }
 
-MPH_MAP_INLINE_METHOD_DECL(my_int32_t, index)(const key_type& k) const {
-  if (__builtin_expect(!slack_.empty(), 0)) {
-     auto sit = slack_.find(hasher128_.hash128(k, 0));
-     if (sit != slack_.end()) return sit->second;
+MPH_MAP_INLINE_METHOD_DECL(my_int32_t, index2)(const key_type& k) const {
+  if (__builtin_expect(index_.perfect_hash_size(), 1)) {
+    auto perfect_hash = index_.perfect_hash(k);
+    return perfect_hash;
   }
+  return -1;
+}
+
+
+
+MPH_MAP_INLINE_METHOD_DECL(my_int32_t, index)(const key_type& k) const {
   if (__builtin_expect(index_.perfect_hash_size(), 1)) {
     auto perfect_hash = index_.perfect_hash(k);
     if (__builtin_expect(present_[perfect_hash], true)) { 
       return perfect_hash;
     }
   }
+  if (__builtin_expect(!slack_.empty(), 0)) {
+     auto sit = slack_.find(hasher128_.hash128(k, 0));
+     if (sit != slack_.end()) return sit->second;
+  }
   return -1;
 }