motiejus/turbonss
1
Fork 0
Code Releases Activity

First tentative on the perfect hash design.

Browse Source
This commit is contained in:
Davi Reis 2012-03-14 23:23:48 -03:00
parent 7fe9527459
commit 3c127c7690
2 changed files with 13 additions and 10 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

4
cxxmph/bm_map.cc
View File

@ -93,8 +93,8 @@ int main(int argc, char** argv) {
Benchmark::Register(new BM_CreateUrls<unordered_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, 0)); Benchmark::Register(new BM_SearchUrls<mph_map<StringPiece, StringPiece>>("URLS100k", 10*1000 * 1000, 0));
Benchmark::Register(new BM_SearchUrls<unordered_map<StringPiece, StringPiece, Murmur3StringPiece>>("URLS100k", 10*1000 * 1000, 0)); Benchmark::Register(new BM_SearchUrls<unordered_map<StringPiece, StringPiece, Murmur3StringPiece>>("URLS100k", 10*1000 * 1000, 0));
// Benchmark::Register(new BM_SearchUrls<mph_map<StringPiece, StringPiece>>("URLS100k", 10*1000 * 1000, 0.9)); Benchmark::Register(new BM_SearchUrls<mph_map<StringPiece, StringPiece>>("URLS100k", 10*1000 * 1000, 0.9));
// Benchmark::Register(new BM_SearchUrls<unordered_map<StringPiece, StringPiece, Murmur3StringPiece>>("URLS100k", 10*1000 * 1000, 0.9)); Benchmark::Register(new BM_SearchUrls<unordered_map<StringPiece, StringPiece, Murmur3StringPiece>>("URLS100k", 10*1000 * 1000, 0.9));
Benchmark::Register(new BM_SearchUint64<mph_map<uint64_t, uint64_t>>); Benchmark::Register(new BM_SearchUint64<mph_map<uint64_t, uint64_t>>);
Benchmark::Register(new BM_SearchUint64<unordered_map<uint64_t, uint64_t>>); Benchmark::Register(new BM_SearchUint64<unordered_map<uint64_t, uint64_t>>);
Benchmark::RunAll(); Benchmark::RunAll();

19
cxxmph/mph_map.h
View File

@ -69,8 +69,8 @@ class mph_map {
void erase(iterator pos); void erase(iterator pos);
void erase(const key_type& k); void erase(const key_type& k);
pair<iterator, bool> insert(const value_type& x); pair<iterator, bool> insert(const value_type& x);
iterator find(const key_type& k); iterator find(const key_type& k) { return slow_find(k, index_.perfect_hash(k)); }
const_iterator find(const key_type& k) const; const_iterator find(const key_type& k) const { return slow_find(k, index_.perfect_hash(k)); };
typedef int32_t my_int32_t; // help macros typedef int32_t my_int32_t; // help macros
int32_t index(const key_type& k) const; int32_t index(const key_type& k) const;
data_type& operator[](const key_type &k); data_type& operator[](const key_type &k);
@ -103,6 +103,9 @@ class mph_map {
return hollow_const_iterator<std::vector<value_type>>(&values_, &present_, it); return hollow_const_iterator<std::vector<value_type>>(&values_, &present_, it);
} }
// Experimental functions, not always faster
iterator fast_find(const key_type& k);
const_iterator fast_find(const key_type& k) const;
iterator slow_find(const key_type& k, uint32_t perfect_hash); iterator slow_find(const key_type& k, uint32_t perfect_hash);
const_iterator slow_find(const key_type& k, uint32_t perfect_hash) const; const_iterator slow_find(const key_type& k, uint32_t perfect_hash) const;
static const uint8_t kNestCollision = 3; // biggest 2 bit value static const uint8_t kNestCollision = 3; // biggest 2 bit value
@ -153,7 +156,7 @@ MPH_MAP_TMPL_SPEC MPH_MAP_CLASS_SPEC::mph_map() : size_(0) {
} }
MPH_MAP_TMPL_SPEC MPH_MAP_CLASS_SPEC::~mph_map() { MPH_MAP_TMPL_SPEC MPH_MAP_CLASS_SPEC::~mph_map() {
fprintf(stderr, "Fast taken: %d Fast: %d Slow %d very_slow %d ratio %f\n", fast_taken_, fast_, slow_, very_slow_, fast_*1.0/slow_); // fprintf(stderr, "Fast taken: %d Fast: %d Slow %d very_slow %d ratio %f\n", fast_taken_, fast_, slow_, very_slow_, fast_*1.0/slow_);
} }
MPH_MAP_METHOD_DECL(insert_return_type, insert)(const value_type& x) { MPH_MAP_METHOD_DECL(insert_return_type, insert)(const value_type& x) {
@ -192,7 +195,7 @@ MPH_MAP_METHOD_DECL(void_type, pack)() {
new_values.reserve(new_values.size() * 2); new_values.reserve(new_values.size() * 2);
std::vector<bool> new_present(index_.perfect_hash_size(), false); std::vector<bool> new_present(index_.perfect_hash_size(), false);
new_present.reserve(new_present.size() * 2); new_present.reserve(new_present.size() * 2);
auto new_nests_size = nextpoweroftwo(ceil(new_values.size())*100 + 1); auto new_nests_size = nextpoweroftwo(ceil(new_values.size())*10000 + 1);
dynamic_2bitset(new_nests_size, true /* fill with 1s */).swap(nests_); dynamic_2bitset(new_nests_size, true /* fill with 1s */).swap(nests_);
vector<bool> used_nests(nests_.size()); vector<bool> used_nests(nests_.size());
uint32_t collisions = 0; uint32_t collisions = 0;
@ -267,7 +270,7 @@ MPH_MAP_METHOD_DECL(void_type, erase)(const key_type& k) {
erase(it); erase(it);
} }
MPH_MAP_METHOD_DECL(const_iterator, find)(const key_type& k) const { MPH_MAP_METHOD_DECL(const_iterator, fast_find)(const key_type& k) const {
uint32_t h[4]; uint32_t h[4];
index_.hash_vector(k, h); index_.hash_vector(k, h);
auto nest = get_nest_value(h); auto nest = get_nest_value(h);
@ -287,7 +290,7 @@ MPH_MAP_METHOD_DECL(const_iterator, find)(const key_type& k) const {
} }
MPH_MAP_METHOD_DECL(const_iterator, slow_find)(const key_type& k, uint32_t perfect_hash) const { MPH_MAP_METHOD_DECL(const_iterator, slow_find)(const key_type& k, uint32_t perfect_hash) const {
if (__builtin_expect(index_.perfect_hash_size(), 0)) { if (__builtin_expect(index_.perfect_hash_size(), 1)) {
if (__builtin_expect(present_[perfect_hash], true)) { if (__builtin_expect(present_[perfect_hash], true)) {
auto vit = values_.begin() + perfect_hash; auto vit = values_.begin() + perfect_hash;
if (equal_(k, vit->first)) return make_iterator(vit); if (equal_(k, vit->first)) return make_iterator(vit);
@ -301,7 +304,7 @@ MPH_MAP_METHOD_DECL(const_iterator, slow_find)(const key_type& k, uint32_t perfe
return end(); return end();
} }
MPH_MAP_METHOD_DECL(iterator, find)(const key_type& k) { MPH_MAP_METHOD_DECL(iterator, fast_find)(const key_type& k) {
uint32_t h[4]; uint32_t h[4];
index_.hash_vector(k, h); index_.hash_vector(k, h);
auto nest = get_nest_value(h); auto nest = get_nest_value(h);
@ -320,7 +323,7 @@ MPH_MAP_METHOD_DECL(iterator, find)(const key_type& k) {
} }
MPH_MAP_METHOD_DECL(iterator, slow_find)(const key_type& k, uint32_t perfect_hash) { MPH_MAP_METHOD_DECL(iterator, slow_find)(const key_type& k, uint32_t perfect_hash) {
if (__builtin_expect(index_.perfect_hash_size(), 0)) { if (__builtin_expect(index_.perfect_hash_size(), 1)) {
if (__builtin_expect(present_[perfect_hash], true)) { if (__builtin_expect(present_[perfect_hash], true)) {
auto vit = values_.begin() + perfect_hash; auto vit = values_.begin() + perfect_hash;
if (equal_(k, vit->first)) return make_iterator(vit); if (equal_(k, vit->first)) return make_iterator(vit);