From d6905ab15992007c28589444a685b4a2878eb977 Mon Sep 17 00:00:00 2001
From: Peter Dimov <pdimov@gmail.com>
Date: Fri, 25 Nov 2022 20:52:56 +0200
Subject: [PATCH] Add benchmark/word_count.cpp

---
 benchmark/word_count.cpp | 397 +++++++++++++++++++++++++++++++++++++++
 1 file changed, 397 insertions(+)
 create mode 100644 benchmark/word_count.cpp

diff --git a/benchmark/word_count.cpp b/benchmark/word_count.cpp
new file mode 100644
index 0000000..2f09f04
--- /dev/null
+++ b/benchmark/word_count.cpp
@@ -0,0 +1,397 @@
+// Copyright 2022 Peter Dimov.
+// Distributed under the Boost Software License, Version 1.0.
+// https://www.boost.org/LICENSE_1_0.txt
+
+#define _SILENCE_CXX17_OLD_ALLOCATOR_MEMBERS_DEPRECATION_WARNING
+#define _SILENCE_CXX20_CISO646_REMOVED_WARNING
+
+#include <boost/unordered/unordered_flat_map.hpp>
+#ifdef HAVE_ABSEIL
+# include "absl/hash/hash.h"
+#endif
+#ifdef HAVE_MULXP_HASH
+# include "mulxp_hash.hpp"
+#endif
+#include <boost/regex.hpp>
+#include <vector>
+#include <memory>
+#include <cstdint>
+#include <iostream>
+#include <iomanip>
+#include <chrono>
+#include <fstream>
+#include <string_view>
+#include <string>
+
+using namespace std::chrono_literals;
+
+static void print_time( std::chrono::steady_clock::time_point & t1, char const* label, std::uint32_t s, std::size_t size )
+{
+    auto t2 = std::chrono::steady_clock::now();
+
+    std::cout << label << ": " << ( t2 - t1 ) / 1ms << " ms (s=" << s << ", size=" << size << ")\n";
+
+    t1 = t2;
+}
+
+static std::vector<std::string> words;
+
+static void init_words()
+{
+    char const* fn = "enwik9"; // http://mattmahoney.net/dc/textdata
+
+    auto t1 = std::chrono::steady_clock::now();
+
+    std::ifstream is( fn );
+    std::string in( std::istreambuf_iterator<char>( is ), std::istreambuf_iterator<char>{} );
+
+    boost::regex re( "[a-zA-Z]+");
+    boost::sregex_token_iterator it( in.begin(), in.end(), re, 0 ), end;
+
+    words.assign( it, end );
+
+    auto t2 = std::chrono::steady_clock::now();
+
+    std::cout << fn << ": " << words.size() << " words, " << ( t2 - t1 ) / 1ms << " ms\n\n";
+}
+
+template<class Map> BOOST_NOINLINE void test_word_count( Map& map, std::chrono::steady_clock::time_point & t1 )
+{
+    std::size_t s = 0;
+
+    for( auto const& word: words )
+    {
+        ++map[ word ];
+        ++s;
+    }
+
+    print_time( t1, "Word count", s, map.size() );
+
+    std::cout << std::endl;
+}
+
+template<class Map> BOOST_NOINLINE void test_contains( Map& map, std::chrono::steady_clock::time_point & t1 )
+{
+    std::size_t s = 0;
+
+    for( auto const& word: words )
+    {
+        std::string_view w2( word );
+        w2.remove_prefix( 1 );
+
+        s += map.contains( w2 );
+    }
+
+    print_time( t1, "Contains", s, map.size() );
+
+    std::cout << std::endl;
+}
+
+template<class Map> BOOST_NOINLINE void test_count( Map& map, std::chrono::steady_clock::time_point & t1 )
+{
+    std::size_t s = 0;
+
+    for( auto const& word: words )
+    {
+        std::string_view w2( word );
+        w2.remove_prefix( 1 );
+
+        s += map.count( w2 );
+    }
+
+    print_time( t1, "Count", s, map.size() );
+
+    std::cout << std::endl;
+}
+
+//
+
+struct record
+{
+    std::string label_;
+    long long time_;
+};
+
+static std::vector<record> times;
+
+template<class Hash> BOOST_NOINLINE void test( char const* label )
+{
+    std::cout << label << ":\n\n";
+
+    boost::unordered_flat_map<std::string_view, std::size_t, Hash> map;
+
+    auto t0 = std::chrono::steady_clock::now();
+    auto t1 = t0;
+
+    test_word_count( map, t1 );
+
+    record rec = { label, 0 };
+
+    test_contains( map, t1 );
+    test_count( map, t1 );
+
+    auto tN = std::chrono::steady_clock::now();
+    std::cout << "Total: " << ( tN - t0 ) / 1ms << " ms\n\n";
+
+    rec.time_ = ( tN - t0 ) / 1ms;
+    times.push_back( rec );
+}
+
+// mul31_hash
+
+struct mul31_hash
+{
+    // not avalanching
+
+    std::size_t operator()( std::string_view const& st ) const BOOST_NOEXCEPT
+    {
+        char const * p = st.data();
+        std::size_t n = st.size();
+
+#if SIZE_MAX > UINT32_MAX
+        std::size_t h = 0xCBF29CE484222325ull;
+#else
+        std::size_t h = 0x811C9DC5u;
+#endif
+
+        for( std::size_t i = 0; i < n; ++i )
+        {
+            h = h * 31 + static_cast<unsigned char>( p[i] );
+        }
+
+        return h;
+    }
+};
+
+// mul31_x4_hash
+
+struct mul31_x4_hash
+{
+    // not avalanching
+
+    std::size_t operator()( std::string_view const& st ) const BOOST_NOEXCEPT
+    {
+        char const * p = st.data();
+        std::size_t n = st.size();
+
+#if SIZE_MAX > UINT32_MAX
+        std::size_t h = 0xCBF29CE484222325ull;
+#else
+        std::size_t h = 0x811C9DC5u;
+#endif
+
+        while( n >= 4 )
+        {
+            h = h * (31u * 31u * 31u * 31u)
+                + static_cast<unsigned char>( p[0] ) * (31u * 31u * 31u)
+                + static_cast<unsigned char>( p[1] ) * (31u * 31u)
+                + static_cast<unsigned char>( p[2] ) * 31u
+                + static_cast<unsigned char>( p[3] );
+
+            p += 4;
+            n -= 4;
+        }
+
+        while( n > 0 )
+        {
+            h = h * 31u + static_cast<unsigned char>( *p );
+
+            ++p;
+            --n;
+        }
+
+        return h;
+    }
+};
+
+// mul31_x8_hash
+
+struct mul31_x8_hash
+{
+    // not avalanching
+
+    std::size_t operator()( std::string_view const& st ) const BOOST_NOEXCEPT
+    {
+        char const * p = st.data();
+        std::size_t n = st.size();
+
+        boost::uint64_t h = 0xCBF29CE484222325ull;
+
+        while( n >= 8 )
+        {
+            h = h * (31ull * 31ull * 31ull * 31ull * 31ull * 31ull * 31ull * 31ull)
+                + static_cast<unsigned char>( p[0] ) * (31ull * 31ull * 31ull * 31ull * 31ull * 31ull * 31ull)
+                + static_cast<unsigned char>( p[1] ) * (31ull * 31ull * 31ull * 31ull * 31ull * 31ull)
+                + static_cast<unsigned char>( p[2] ) * (31ull * 31ull * 31ull * 31ull * 31ull)
+                + static_cast<unsigned char>( p[3] ) * (31ull * 31ull * 31ull * 31ull)
+                + static_cast<unsigned char>( p[4] ) * (31ull * 31ull * 31ull)
+                + static_cast<unsigned char>( p[5] ) * (31ull * 31ull)
+                + static_cast<unsigned char>( p[6] ) * 31ull
+                + static_cast<unsigned char>( p[7] );
+
+            p += 8;
+            n -= 8;
+        }
+
+        while( n > 0 )
+        {
+            h = h * 31u + static_cast<unsigned char>( *p );
+
+            ++p;
+            --n;
+        }
+
+        return static_cast<std::size_t>( h );
+    }
+};
+
+// fnv1a_hash
+
+template<int Bits> struct fnv1a_hash_impl;
+
+template<> struct fnv1a_hash_impl<32>
+{
+    std::size_t operator()( std::string_view const& s ) const
+    {
+        std::size_t h = 0x811C9DC5u;
+
+        char const * first = s.data();
+        char const * last = first + s.size();
+
+        for( ; first != last; ++first )
+        {
+            h ^= static_cast<unsigned char>( *first );
+            h *= 0x01000193ul;
+        }
+
+        return h;
+    }
+};
+
+template<> struct fnv1a_hash_impl<64>
+{
+    std::size_t operator()( std::string_view const& s ) const
+    {
+        std::size_t h = 0xCBF29CE484222325ull;
+
+        char const * first = s.data();
+        char const * last = first + s.size();
+
+        for( ; first != last; ++first )
+        {
+            h ^= static_cast<unsigned char>( *first );
+            h *= 0x00000100000001B3ull;
+        }
+
+        return h;
+    }
+};
+
+struct fnv1a_hash: fnv1a_hash_impl< std::numeric_limits<std::size_t>::digits >
+{
+    using is_avalanching = void;
+};
+
+// std_hash
+
+struct std_hash: std::hash<std::string_view>
+{
+    using is_avalanching = void;
+};
+
+// absl_hash
+
+#ifdef HAVE_ABSEIL
+
+struct absl_hash: absl::Hash<std::string_view>
+{
+    using is_avalanching = void;
+};
+
+#endif
+
+#ifdef HAVE_MULXP_HASH
+
+struct mulxp0_hash_
+{
+    using is_avalanching = void;
+
+    std::size_t operator()( std::string_view const& st ) const BOOST_NOEXCEPT
+    {
+        return mulxp0_hash( (unsigned char const*)st.data(), st.size(), 0 );
+    }
+};
+
+struct mulxp1_hash_
+{
+    using is_avalanching = void;
+
+    std::size_t operator()( std::string_view const& st ) const BOOST_NOEXCEPT
+    {
+        return mulxp1_hash( (unsigned char const*)st.data(), st.size(), 0 );
+    }
+};
+
+struct mulxp2_hash_
+{
+    using is_avalanching = void;
+
+    std::size_t operator()( std::string_view const& st ) const BOOST_NOEXCEPT
+    {
+        return mulxp2_hash( (unsigned char const*)st.data(), st.size(), 0 );
+    }
+};
+
+struct mulxp3_hash_
+{
+    using is_avalanching = void;
+
+    std::size_t operator()( std::string_view const& st ) const BOOST_NOEXCEPT
+    {
+        return mulxp3_hash( (unsigned char const*)st.data(), st.size(), 0 );
+    }
+};
+
+#endif
+
+//
+
+int main()
+{
+    init_words();
+
+    test< boost::hash<std::string_view> >( "boost::hash" );
+    test< std_hash >( "std::hash" );
+    test< mul31_hash >( "mul31_hash" );
+    test< mul31_x4_hash >( "mul31_x4_hash" );
+    test< mul31_x8_hash >( "mul31_x8_hash" );
+    test< fnv1a_hash >( "fnv1a_hash" );
+
+#ifdef HAVE_ABSEIL
+
+    test< absl_hash >( "absl::Hash" );
+
+#endif
+
+#ifdef HAVE_MULXP_HASH
+
+    test< mulxp0_hash_ >( "mulxp0_hash" );
+    test< mulxp1_hash_ >( "mulxp1_hash" );
+    test< mulxp2_hash_ >( "mulxp2_hash" );
+    test< mulxp3_hash_ >( "mulxp3_hash" );
+
+#endif
+
+    std::cout << "---\n\n";
+
+    for( auto const& x: times )
+    {
+        std::cout << std::setw( 22 ) << ( x.label_ + ": " ) << std::setw( 5 ) << x.time_ << " ms\n";
+    }
+}
+
+#ifdef HAVE_ABSEIL
+# include "absl/hash/internal/hash.cc"
+# include "absl/hash/internal/low_level_hash.cc"
+# include "absl/hash/internal/city.cc"
+#endif