Merge pull request #24 from bab2min/develop

direct-managing memory by using mi_stl_allocator
bab2min · Jan 19, 2021 · 971c646 · 971c646
2 parents 06ffe4e + bc1d628
commit 971c646
Show file tree

Hide file tree

Showing 8 changed files with 83 additions and 43 deletions.
diff --git a/evaluator.vcxproj b/evaluator.vcxproj
@@ -100,21 +100,27 @@
   <PropertyGroup Label="UserMacros" />
   <PropertyGroup Condition="'$(Configuration)|$(Platform)'=='Release|x64'">
     <LinkIncremental>false</LinkIncremental>
+    <IncludePath>$(ProjectDir)mimalloc/include/;$(VC_IncludePath);$(WindowsSDK_IncludePath);</IncludePath>
   </PropertyGroup>
   <PropertyGroup Condition="'$(Configuration)|$(Platform)'=='Release(LODA_TXT)|x64'">
     <LinkIncremental>false</LinkIncremental>
+    <IncludePath>$(ProjectDir)mimalloc/include/;$(VC_IncludePath);$(WindowsSDK_IncludePath);</IncludePath>
   </PropertyGroup>
   <PropertyGroup Condition="'$(Configuration)|$(Platform)'=='Debug|Win32'">
     <LinkIncremental>true</LinkIncremental>
+    <IncludePath>$(ProjectDir)mimalloc/include/;$(VC_IncludePath);$(WindowsSDK_IncludePath);</IncludePath>
   </PropertyGroup>
   <PropertyGroup Condition="'$(Configuration)|$(Platform)'=='Debug|x64'">
     <LinkIncremental>true</LinkIncremental>
+    <IncludePath>$(ProjectDir)mimalloc/include/;$(VC_IncludePath);$(WindowsSDK_IncludePath);</IncludePath>
   </PropertyGroup>
   <PropertyGroup Condition="'$(Configuration)|$(Platform)'=='Release|Win32'">
     <LinkIncremental>false</LinkIncremental>
+    <IncludePath>$(ProjectDir)mimalloc/include/;$(VC_IncludePath);$(WindowsSDK_IncludePath);</IncludePath>
   </PropertyGroup>
   <PropertyGroup Condition="'$(Configuration)|$(Platform)'=='Release(LODA_TXT)|Win32'">
     <LinkIncremental>false</LinkIncremental>
+    <IncludePath>$(ProjectDir)mimalloc/include/;$(VC_IncludePath);$(WindowsSDK_IncludePath);</IncludePath>
   </PropertyGroup>
   <ItemDefinitionGroup Condition="'$(Configuration)|$(Platform)'=='Release|x64'">
     <ClCompile>
@@ -123,7 +129,7 @@
       <Optimization>MaxSpeed</Optimization>
       <FunctionLevelLinking>true</FunctionLevelLinking>
       <IntrinsicFunctions>true</IntrinsicFunctions>
-      <PreprocessorDefinitions>NDEBUG;_CONSOLE;%(PreprocessorDefinitions)</PreprocessorDefinitions>
+      <PreprocessorDefinitions>USE_MIMALLOC;NDEBUG;_CONSOLE;%(PreprocessorDefinitions)</PreprocessorDefinitions>
       <RuntimeLibrary>MultiThreaded</RuntimeLibrary>
       <AdditionalOptions>/Qvec-report:1 %(AdditionalOptions)</AdditionalOptions>
       <EnableEnhancedInstructionSet>AdvancedVectorExtensions</EnableEnhancedInstructionSet>
@@ -142,7 +148,7 @@
       <Optimization>MaxSpeed</Optimization>
       <FunctionLevelLinking>true</FunctionLevelLinking>
       <IntrinsicFunctions>true</IntrinsicFunctions>
-      <PreprocessorDefinitions>LOAD_TXT;NDEBUG;_CONSOLE;%(PreprocessorDefinitions)</PreprocessorDefinitions>
+      <PreprocessorDefinitions>USE_MIMALLOC;LOAD_TXT;NDEBUG;_CONSOLE;%(PreprocessorDefinitions)</PreprocessorDefinitions>
       <RuntimeLibrary>MultiThreaded</RuntimeLibrary>
       <AdditionalOptions>/Qvec-report:1 %(AdditionalOptions)</AdditionalOptions>
       <EnableEnhancedInstructionSet>AdvancedVectorExtensions</EnableEnhancedInstructionSet>
@@ -159,7 +165,7 @@
       <PrecompiledHeader>NotUsing</PrecompiledHeader>
       <WarningLevel>Level3</WarningLevel>
       <Optimization>Disabled</Optimization>
-      <PreprocessorDefinitions>WIN32;_DEBUG;_CONSOLE;%(PreprocessorDefinitions)</PreprocessorDefinitions>
+      <PreprocessorDefinitions>USE_MIMALLOC;WIN32;_DEBUG;_CONSOLE;%(PreprocessorDefinitions)</PreprocessorDefinitions>
     </ClCompile>
     <Link>
       <SubSystem>Console</SubSystem>
@@ -170,7 +176,7 @@
       <PrecompiledHeader>NotUsing</PrecompiledHeader>
       <WarningLevel>Level3</WarningLevel>
       <Optimization>Disabled</Optimization>
-      <PreprocessorDefinitions>_DEBUG;_CONSOLE;%(PreprocessorDefinitions)</PreprocessorDefinitions>
+      <PreprocessorDefinitions>USE_MIMALLOC;_DEBUG;_CONSOLE;%(PreprocessorDefinitions)</PreprocessorDefinitions>
     </ClCompile>
     <Link>
       <SubSystem>Console</SubSystem>
@@ -184,7 +190,7 @@
       <Optimization>MaxSpeed</Optimization>
       <FunctionLevelLinking>true</FunctionLevelLinking>
       <IntrinsicFunctions>true</IntrinsicFunctions>
-      <PreprocessorDefinitions>WIN32;NDEBUG;_CONSOLE;%(PreprocessorDefinitions)</PreprocessorDefinitions>
+      <PreprocessorDefinitions>USE_MIMALLOC;WIN32;NDEBUG;_CONSOLE;%(PreprocessorDefinitions)</PreprocessorDefinitions>
       <RuntimeLibrary>MultiThreaded</RuntimeLibrary>
     </ClCompile>
     <Link>
@@ -200,7 +206,7 @@
       <Optimization>MaxSpeed</Optimization>
       <FunctionLevelLinking>true</FunctionLevelLinking>
       <IntrinsicFunctions>true</IntrinsicFunctions>
-      <PreprocessorDefinitions>LOAD_TXT;WIN32;NDEBUG;_CONSOLE;%(PreprocessorDefinitions)</PreprocessorDefinitions>
+      <PreprocessorDefinitions>USE_MIMALLOC;LOAD_TXT;WIN32;NDEBUG;_CONSOLE;%(PreprocessorDefinitions)</PreprocessorDefinitions>
       <RuntimeLibrary>MultiThreaded</RuntimeLibrary>
     </ClCompile>
     <Link>

diff --git a/src/core/KModelMgr.cpp b/src/core/KModelMgr.cpp
@@ -366,7 +366,7 @@ void KModelMgr::loadMorphBin(_Istream& is)
 	forms.resize(formSize);
 	morphemes.resize(morphemeSize);
 
-	auto mapper = [this](size_t p)->const KMorpheme*
+	auto mapper = [this](size_t p)
 	{
 		return (const KMorpheme*)p;
 	};

diff --git a/src/core/KTrie.cpp b/src/core/KTrie.cpp
@@ -8,15 +8,15 @@
 using namespace std;
 using namespace kiwi;
 
-vector<KGraphNode> KTrie::split(const k_string& str, const PatternMatcher* pm, size_t matchOptions) const
+mvector<KGraphNode> KTrie::split(const k_string& str, const PatternMatcher* pm, size_t matchOptions) const
 {
-	vector<KGraphNode> ret;
+	mvector<KGraphNode> ret;
 	ret.reserve(8);
 	ret.emplace_back();
 	size_t n = 0;
-	vector<const KForm*> candidates;
+	mvector<const KForm*> candidates;
 	const KTrie* curTrie = this;
-	unordered_map<uint32_t, int> spacePos;
+	munordered_map<uint32_t, int> spacePos;
 	size_t lastSpecialEndPos = 0, specialStartPos = 0;
 	KPOSTag chrType, lastChrType = KPOSTag::UNKNOWN;
 	auto branchOut = [&](bool makeLongMatch = false)
@@ -317,9 +317,9 @@ KTrie KTrie::loadFromBin(std::istream & is, const KForm* base)
 	return t;
 }
 
-vector<KGraphNode> KGraphNode::removeUnconnected(const vector<KGraphNode>& graph)
+mvector<KGraphNode> KGraphNode::removeUnconnected(const mvector<KGraphNode>& graph)
 {
-	vector<uint16_t> connectedList(graph.size()), newIndexDiff(graph.size());
+	mvector<uint16_t> connectedList(graph.size()), newIndexDiff(graph.size());
 	connectedList[graph.size() - 1] = true;
 	connectedList[0] = true;
 	// forward searching
@@ -368,7 +368,7 @@ vector<KGraphNode> KGraphNode::removeUnconnected(const vector<KGraphNode>& graph
 		newIndexDiff[i] = i + 1 - newIndexDiff[i];
 	}
 
-	vector<KGraphNode> ret;
+	mvector<KGraphNode> ret;
 	ret.reserve(connectedCnt);
 	for (size_t i = 0; i < graph.size(); ++i)
 	{

diff --git a/src/core/KTrie.h b/src/core/KTrie.h
@@ -23,7 +23,7 @@ namespace kiwi
 
 		KGraphNode* getPrev(size_t idx) const { return prevs[idx] ? (KGraphNode*)this - prevs[idx] : nullptr; }
 
-		static std::vector<KGraphNode> removeUnconnected(const std::vector<KGraphNode>& graph);
+		static mvector<KGraphNode> removeUnconnected(const mvector<KGraphNode>& graph);
 
 		void addPrev(size_t distance)
 		{
@@ -39,7 +39,7 @@ namespace kiwi
 
 	struct KTrie : public Trie<char16_t, const KForm*, OverriddenMap<std::map<char16_t, int32_t>>>
 	{
-		std::vector<KGraphNode> split(const k_string& str, const PatternMatcher* pm, size_t matchOptions) const;
+		mvector<KGraphNode> split(const k_string& str, const PatternMatcher* pm, size_t matchOptions) const;
 		const KForm* findForm(const k_string& str) const;
 		KTrie* getNext(k_char i) const { return (KTrie*)Trie::getNext(i); }
 		KTrie* getFail() const { return (KTrie*)Trie::getFail(); }

diff --git a/src/core/Kiwi.cpp b/src/core/Kiwi.cpp
@@ -1,8 +1,4 @@
-#ifdef USE_MIMALLOC
-#include <mimalloc-new-delete.h>
-#endif
-
-#include <future>
+#include <future>
 #include "Kiwi.h"
 #include "Utils.h"
 #include "KFeatureTestor.h"
@@ -287,8 +283,8 @@ struct MInfo
 };
 
 struct WordLL;
-using MInfos = vector<MInfo>;
-using WordLLs = vector<WordLL>;
+using MInfos = mvector<MInfo>;
+using WordLLs = mvector<WordLL>;
 
 struct WordLL
 {
@@ -327,7 +323,7 @@ auto findNthLargest(_Iter first, _Iter last, size_t nth, _Key&& fn, _Filter&& fi
 {
 	using KeyType = decltype(fn(*first));
 
-	std::vector<KeyType> v;
+	mvector<KeyType> v;
 	for (; first != last; ++first)
 	{
 		if(filter(*first)) v.emplace_back(fn(*first));
@@ -338,7 +334,7 @@ auto findNthLargest(_Iter first, _Iter last, size_t nth, _Key&& fn, _Filter&& fi
 }
 
 template<class _Type>
-void evalTrigram(const KNLangModel::Node* rootNode, const KMorpheme* morphBase, const vector<k_string>& ownForms, const WordLL** wBegin, const WordLL** wEnd, 
+void evalTrigram(const KNLangModel::Node* rootNode, const KMorpheme* morphBase, const mvector<k_string>& ownForms, const WordLL** wBegin, const WordLL** wEnd, 
 	array<WID, 4> seq, size_t chSize, const KMorpheme* curMorph, const KGraphNode* node, _Type& maxWidLL)
 {
 	for (; wBegin != wEnd; ++wBegin)
@@ -384,12 +380,12 @@ void evalTrigram(const KNLangModel::Node* rootNode, const KMorpheme* morphBase,
 	}
 }
 
-vector<pair<Kiwi::path, float>> Kiwi::findBestPath(const vector<KGraphNode>& graph, const KNLangModel * knlm, const KMorpheme* morphBase, size_t topN) const
+mvector<pair<Kiwi::path, float>> Kiwi::findBestPath(const mvector<KGraphNode>& graph, const KNLangModel * knlm, const KMorpheme* morphBase, size_t topN) const
 {
-	vector<WordLLs> cache(graph.size());
+	mvector<WordLLs> cache(graph.size());
 	const KGraphNode* startNode = &graph.front();
 	const KGraphNode* endNode = &graph.back();
-	vector<k_string> ownFormList;
+	mvector<k_string> ownFormList;
 
 	vector<const KMorpheme*> unknownNodeCands, unknownNodeLCands;
 	unknownNodeCands.emplace_back(morphBase + (size_t)KPOSTag::NNG + 1);
@@ -435,8 +431,8 @@ vector<pair<Kiwi::path, float>> Kiwi::findBestPath(const vector<KGraphNode>& gra
 			condV = curMorph->vowel;
 			condP = curMorph->polar;
 
-			unordered_map<WID, vector<WordLLP>> maxWidLL;
-			vector<const WordLL*> works;
+			munordered_map<WID, mvector<WordLLP>> maxWidLL;
+			mvector<const WordLL*> works;
 			works.reserve(8);
 			float discountForCombining = 0;
 
@@ -645,7 +641,7 @@ vector<pair<Kiwi::path, float>> Kiwi::findBestPath(const vector<KGraphNode>& gra
 
 #endif
 
-	vector<pair<path, float>> ret;
+	mvector<pair<path, float>> ret;
 	for (size_t i = 0; i < min(topN, cand.size()); ++i)
 	{
 		path mv(cand[i].morphs.size() - 1);
@@ -757,7 +753,7 @@ void Kiwi::perform(size_t topN, const function<u16string(size_t)>& reader, const
 std::vector<KResult> Kiwi::analyzeSent(const std::u16string::const_iterator & sBegin, const std::u16string::const_iterator & sEnd, size_t topN, size_t matchOptions) const
 {
 	auto nstr = normalizeHangul({ sBegin, sEnd });
-	vector<uint32_t> posMap(nstr.size() + 1);
+	mvector<uint32_t> posMap(nstr.size() + 1);
 	for (size_t i = 0; i < nstr.size(); ++i)
 	{
 		posMap[i + 1] = posMap[i] + (isHangulCoda(nstr[i]) ? 0 : 1);
@@ -820,7 +816,7 @@ vector<KResult> Kiwi::analyze(const u16string & str, size_t topN, size_t matchOp
 {
 	if (!mdl->getTrie()) throw KiwiException("Model should be prepared before analyzing.");
 	auto chunk = str.begin();
-	vector<u16string::const_iterator> sents;
+	mvector<u16string::const_iterator> sents;
 	sents.emplace_back(chunk);
 	while (chunk != str.end())
 	{

diff --git a/src/core/Kiwi.h b/src/core/Kiwi.h
@@ -45,7 +45,7 @@ namespace kiwi
 		friend std::ostream& (::operator<<) (std::ostream& os, const KWordPair& kp);
 	};
 
-	typedef std::pair<std::vector<KWordPair>, float> KResult;
+	using KResult = std::pair<std::vector<KWordPair>, float>;
 
 	class KModelMgr;
 
@@ -59,8 +59,8 @@ namespace kiwi
 		bool integrateAllomorph;
 		std::unique_ptr<PatternMatcher> pm;
 		KWordDetector detector;
-		typedef std::vector<std::tuple<const KMorpheme*, k_string, uint32_t>> path;
-		std::vector<std::pair<path, float>> findBestPath(const std::vector<KGraphNode>& graph, const KNLangModel * knlm, const KMorpheme* morphBase, size_t topN) const;
+		using path = mvector<std::tuple<const KMorpheme*, k_string, uint32_t>>;
+		mvector<std::pair<path, float>> findBestPath(const mvector<KGraphNode>& graph, const KNLangModel * knlm, const KMorpheme* morphBase, size_t topN) const;
 		std::vector<KResult> analyzeSent(const std::u16string::const_iterator& sBegin, const std::u16string::const_iterator& sEnd, size_t topN, size_t matchOptions) const;
 	public:
 		enum

diff --git a/src/core/KiwiHeader.h b/src/core/KiwiHeader.h
@@ -32,6 +32,10 @@
 #define TRIE_ALLOC_ARRAY
 #define KSTR(x) u##x
 
+#ifdef USE_MIMALLOC
+#include <mimalloc.h>
+#endif
+
 namespace kiwi
 {
 	template<typename T, typename... Args>
@@ -54,9 +58,42 @@ namespace kiwi
 		using std::runtime_error::runtime_error;
 	};
 
+#ifdef USE_MIMALLOC
+	template<typename _Ty>
+	using mvector = std::vector<_Ty, mi_stl_allocator<_Ty>>;
+
+	template<typename _K, typename _V>
+	using munordered_map = std::unordered_map<_K, _V, std::hash<_K>, std::equal_to<_K>, mi_stl_allocator<std::pair<const _K, _V>>>;
+
+	using k_string = std::basic_string<k_char, std::char_traits<k_char>, mi_stl_allocator<k_char>>;
+	using k_stringstream = std::basic_stringstream<k_char, std::char_traits<k_char>, mi_stl_allocator<k_char>>;
+	using k_vchar = mvector<k_char>;
+	using k_vpcf = mvector<std::pair<k_vchar, float>>;
+#else
+	template<typename _Ty>
+	using mvector = std::vector<_Ty>;
+
+	template<typename _K, typename _V>
+	using munordered_map = std::unordered_map<_K, _V>;
 
 	using k_string = std::basic_string<k_char>;
 	using k_stringstream = std::basic_stringstream<k_char>;
-	using k_vchar = std::vector<k_char>;
-	using k_vpcf = std::vector<std::pair<k_vchar, float>>;
+	using k_vchar = mvector<k_char>;
+	using k_vpcf = mvector<std::pair<k_vchar, float>>;
+#endif
+}
+
+#ifdef USE_MIMALLOC
+namespace std
+{
+	template<>
+	struct hash<kiwi::k_string>
+	{
+		size_t operator()(const kiwi::k_string& s) const
+		{
+			return hash<basic_string<kiwi::k_char>>{}({ s.begin(), s.end() });
+		}
+	};
+
 }
+#endif
diff --git a/src/core/serializer.hpp b/src/core/serializer.hpp
@@ -83,17 +83,18 @@ namespace kiwi
 			if (!is.read((char*)&v, sizeof(_Ty))) throw std::ios_base::failure(std::string{ "reading type '" } +typeid(_Ty).name() + "' failed");
 		}
 
-		inline void writeToBinStreamImpl(std::ostream& os, const std::u16string& v)
+		template<class CharTy, class T, class Al>
+		inline void writeToBinStreamImpl(std::ostream& os, const std::basic_string<CharTy, T, Al>& v)
 		{
 			writeToBinStream<uint32_t>(os, v.size());
-			if (!os.write((const char*)&v[0], v.size() * sizeof(char16_t))) throw std::ios_base::failure(std::string{ "writing type '" } +typeid(k_string).name() + "' failed");
+			if (!os.write((const char*)&v[0], v.size() * sizeof(CharTy))) throw std::ios_base::failure("writing type 'string' failed");
 		}
 
-		template<class _Istream>
-		inline void readFromBinStreamImpl(_Istream& is, std::u16string& v)
+		template<class _Istream, class CharTy, class T, class Al>
+		inline void readFromBinStreamImpl(_Istream& is, std::basic_string<CharTy, T, Al>& v)
 		{
 			v.resize(readFromBinStream<uint32_t>(is));
-			if (!is.read((char*)&v[0], v.size() * sizeof(char16_t))) throw std::ios_base::failure(std::string{ "reading type '" } +typeid(k_string).name() + "' failed");
+			if (!is.read((char*)&v[0], v.size() * sizeof(CharTy))) throw std::ios_base::failure("reading type 'string' failed");
 		}
 
 		template<class _Ty1, class _Ty2>