Update benchmarks

bm/CMakeLists.txt

@@ -119,8 +119,13 @@ set(RYML_SAJSON_INC_DIR ${sajson_dir}/include)
 function(ryml_add_bm_comparison_exe name)
     c4_add_executable(ryml-bm-${name}
         SOURCES bm_common.hpp ${ARGN}
+            ../src_extra/c4/yml/extra/event_handler_ints.hpp
+            ../src_extra/c4/yml/extra/event_handler_ints.cpp
         LIBS ryml yaml yaml-cpp benchmark jsoncpp_static nlohmann_json c4fs
-        INC_DIRS ${RYML_RAPIDJSON_INC_DIR} ${RYML_SAJSON_INC_DIR}
+        INC_DIRS
+            ${RYML_RAPIDJSON_INC_DIR}
+            ${RYML_SAJSON_INC_DIR}
+            ../src_extra
         FOLDER bm)
     if(RYML_DBG)
         target_compile_definitions(ryml-bm-${name} PRIVATE RYML_DBG)
@@ -133,9 +138,8 @@ function(ryml_add_bm_comparison_exe name)
     _c4_set_target_folder(ryml-bm-${name}-all bm/run)
 endfunction()
 
-ryml_add_bm_comparison_exe(parse bm_parse.cpp)
 ryml_add_bm_comparison_exe(emit bm_emit.cpp)
-
+ryml_add_bm_comparison_exe(parse bm_parse.cpp)
 
 function(ryml_add_bm_comparison_case target name case_file)
     c4_dbg("adding benchmark case: ${case_file}")

bm/bm_common.hpp

@@ -4,6 +4,7 @@
 #include <ryml_std.hpp>
 #include <c4/fs/fs.hpp>
 #include "c4/yml/parse.hpp"
+#include "c4/yml/extra/event_handler_ints.hpp"
 
 #include <vector>
 #include <iostream>
@@ -82,14 +83,81 @@ namespace bm = benchmark;
 #   pragma clang diagnostic ignored "-Wdeprecated-declarations"
 #   pragma clang diagnostic ignored "-Wunused-variable"
 #   pragma clang diagnostic ignored "-Wsign-conversion"
+#   pragma clang diagnostic ignored "-Wold-style-cast"
 #elif defined(__GNUC__)
 #   pragma GCC diagnostic push
 #   pragma GCC diagnostic ignored "-Wdeprecated-declarations"
 #   pragma GCC diagnostic ignored "-Wunused-variable"
 #   pragma GCC diagnostic ignored "-Wsign-conversion"
+#   pragma GCC diagnostic ignored "-Wold-style-cast"
 #endif
 
 
+//-----------------------------------------------------------------------------
+//-----------------------------------------------------------------------------
+//-----------------------------------------------------------------------------
+
+using IntHandler = ryml::extra::EventHandlerInts;
+using IntParser = ryml::ParseEngine<IntHandler>;
+struct IntData
+{
+    std::vector<int> events;
+    std::vector<char> arena;
+    void resize(ryml::extra::EventHandlerInts const& handler)
+    {
+        resize((size_t)handler.required_size_events(),
+               handler.required_size_arena());
+    }
+    void resize(ryml::substr src)
+    {
+        resize((size_t)ryml::extra::estimate_events_ints_size(src),
+               src.len);
+    }
+    void resize(size_t evt_sz, size_t arn_sz)
+    {
+        events.resize(evt_sz);
+        events.resize(arn_sz);
+    }
+};
+struct IntObjects
+{
+    IntObjects(ryml::ParserOptions opts={})
+        : handler()
+        , parser(&handler, opts)
+        , data()
+    {
+    }
+    IntHandler handler;
+    IntParser parser;
+    IntData data;
+    bool again()
+    {
+        if(!handler.fits_buffers())
+        {
+            data.resize(handler);
+            return true;
+        }
+        return false;
+    }
+};
+inline void parse_yaml_inplace(ryml::csubstr filename, ryml::substr src, IntParser &parser, IntData *dst)
+{
+    parser.m_evt_handler->reset(src,
+                                ryml::to_substr(dst->arena),
+                                dst->events.data(),
+                                (int)dst->events.size());
+    parser.parse_in_place_ev(filename, src);
+}
+inline void parse_json_inplace(ryml::csubstr filename, ryml::substr src, IntParser &parser, IntData *dst)
+{
+    parser.m_evt_handler->reset(src,
+                                ryml::to_substr(dst->arena),
+                                dst->events.data(),
+                                (int)dst->events.size());
+    parser.parse_json_in_place_ev(filename, src);
+}
+
+
 //-----------------------------------------------------------------------------
 //-----------------------------------------------------------------------------
 //-----------------------------------------------------------------------------
@@ -99,6 +167,7 @@ enum : int {
       kClearTreeArena=2,
       kResetInPlace=4,
       kReserveTree=8,
+      kReserveInts=8,
       kAll=kClearTree|kClearTreeArena|kResetInPlace|kReserveTree,
 };
 
@@ -117,6 +186,9 @@ struct BmCase
     rapidjson::Document    rapidjson_doc;
     ryml::Tree             libyaml_tree;
 
+    IntObjects int_obj;
+    IntObjects int_obj_nofilter{ryml::ParserOptions().scalar_filtering(false)};
+
     void run(std::string name, int argc, char **argv)
     {
         bm_name = name;
@@ -173,6 +245,11 @@ USAGE: bm <case.yml>
             RYML_CHECK(capacity > 0);
             ryml_tree.reserve(capacity);
         }
+        if(what & kReserveTree)
+        {
+            RYML_CHECK(capacity > 0);
+            ryml_tree.reserve(capacity);
+        }
         if(what & kResetInPlace)
         {
             C4_ASSERT(in_place.size() == src.size());

bm/bm_parse.cpp

@@ -279,6 +279,168 @@ void bm_ryml_json_arena_nofilter_reserve(bm::State& st)
 }
 
 
+//-----------------------------------------------------------------------------
+
+void bm_rymlints_estimate(bm::State& st)
+{
+    c4::csubstr src = c4::to_csubstr(s_bm_case->in_place).trimr('\0');
+    for(auto _ : st)
+    {
+        auto sz = c4::yml::extra::estimate_events_ints_size(src);
+        bm::DoNotOptimize(sz);
+    }
+    s_bm_case->report(st);
+}
+
+
+void bm_rymlints_yaml_inplace(bm::State& st)
+{
+    c4::substr src = c4::to_substr(s_bm_case->in_place).trimr('\0');
+    for(auto _ : st)
+    {
+        IntObjects obj;
+        s_bm_case->prepare(st, kResetInPlace);
+        parse_yaml_inplace(s_bm_case->filename, src, obj.parser, &obj.data);
+        if(obj.again())
+        {
+            s_bm_case->prepare(st, kResetInPlace);
+            parse_yaml_inplace(s_bm_case->filename, src, obj.parser, &obj.data);
+        }
+        bm::DoNotOptimize(obj);
+    }
+    s_bm_case->report(st);
+}
+
+void bm_rymlints_yaml_inplace_reserve(bm::State& st)
+{
+    c4::substr src = c4::to_substr(s_bm_case->in_place).trimr('\0');
+    int sz = ryml::extra::estimate_events_ints_size(src);
+    for(auto _ : st)
+    {
+        st.PauseTiming();
+        s_bm_case->prepare(kResetInPlace);
+        IntObjects obj;
+        obj.data.resize(src);
+        st.ResumeTiming();
+        parse_yaml_inplace(s_bm_case->filename, src, obj.parser, &obj.data);
+        bm::DoNotOptimize(obj);
+    }
+    s_bm_case->report(st);
+}
+
+void bm_rymlints_yaml_inplace_nofilter(bm::State& st)
+{
+    c4::substr src = c4::to_substr(s_bm_case->in_place).trimr('\0');
+    for(auto _ : st)
+    {
+        IntObjects obj(ryml::ParserOptions().scalar_filtering(false));
+        s_bm_case->prepare(st, kResetInPlace);
+        parse_yaml_inplace(s_bm_case->filename, src, obj.parser, &obj.data);
+        if(obj.again())
+        {
+            s_bm_case->prepare(st, kResetInPlace);
+            parse_yaml_inplace(s_bm_case->filename, src, obj.parser, &obj.data);
+        }
+        bm::DoNotOptimize(obj);
+    }
+    s_bm_case->report(st);
+}
+
+void bm_rymlints_yaml_inplace_nofilter_reserve(bm::State& st)
+{
+    c4::substr src = c4::to_substr(s_bm_case->in_place).trimr('\0');
+    int sz = ryml::extra::estimate_events_ints_size(src);
+    for(auto _ : st)
+    {
+        st.PauseTiming();
+        s_bm_case->prepare(kResetInPlace);
+        IntObjects obj(ryml::ParserOptions().scalar_filtering(false));
+        obj.data.resize(src);
+        st.ResumeTiming();
+        parse_yaml_inplace(s_bm_case->filename, src, obj.parser, &obj.data);
+        bm::DoNotOptimize(obj);
+    }
+    s_bm_case->report(st);
+}
+
+
+//-----------------------------------------------------------------------------
+
+void bm_rymlints_json_inplace(bm::State& st)
+{
+    c4::substr src = c4::to_substr(s_bm_case->in_place).trimr('\0');
+    for(auto _ : st)
+    {
+        ONLY_FOR_JSON;
+        IntObjects obj;
+        s_bm_case->prepare(st, kResetInPlace);
+        parse_json_inplace(s_bm_case->filename, src, obj.parser, &obj.data);
+        if(obj.again())
+        {
+            s_bm_case->prepare(st, kResetInPlace);
+            parse_json_inplace(s_bm_case->filename, src, obj.parser, &obj.data);
+        }
+        bm::DoNotOptimize(obj);
+    }
+    s_bm_case->report(st);
+}
+
+void bm_rymlints_json_inplace_reserve(bm::State& st)
+{
+    c4::substr src = c4::to_substr(s_bm_case->in_place).trimr('\0');
+    int sz = ryml::extra::estimate_events_ints_size(src);
+    for(auto _ : st)
+    {
+        ONLY_FOR_JSON;
+        st.PauseTiming();
+        s_bm_case->prepare(kResetInPlace);
+        IntObjects obj;
+        obj.data.resize(src);
+        st.ResumeTiming();
+        parse_json_inplace(s_bm_case->filename, src, obj.parser, &obj.data);
+        bm::DoNotOptimize(obj);
+    }
+    s_bm_case->report(st);
+}
+
+void bm_rymlints_json_inplace_nofilter(bm::State& st)
+{
+    c4::substr src = c4::to_substr(s_bm_case->in_place).trimr('\0');
+    for(auto _ : st)
+    {
+        ONLY_FOR_JSON;
+        IntObjects obj(ryml::ParserOptions().scalar_filtering(false));
+        s_bm_case->prepare(st, kResetInPlace);
+        parse_json_inplace(s_bm_case->filename, src, obj.parser, &obj.data);
+        if(obj.again())
+        {
+            s_bm_case->prepare(st, kResetInPlace);
+            parse_json_inplace(s_bm_case->filename, src, obj.parser, &obj.data);
+        }
+        bm::DoNotOptimize(obj);
+    }
+    s_bm_case->report(st);
+}
+
+void bm_rymlints_json_inplace_nofilter_reserve(bm::State& st)
+{
+    c4::substr src = c4::to_substr(s_bm_case->in_place).trimr('\0');
+    int sz = ryml::extra::estimate_events_ints_size(src);
+    for(auto _ : st)
+    {
+        ONLY_FOR_JSON;
+        st.PauseTiming();
+        s_bm_case->prepare(kResetInPlace);
+        IntObjects obj(ryml::ParserOptions().scalar_filtering(false));
+        obj.data.resize(src);
+        st.ResumeTiming();
+        parse_json_inplace(s_bm_case->filename, src, obj.parser, &obj.data);
+        bm::DoNotOptimize(obj);
+    }
+    s_bm_case->report(st);
+}
+
+
 //-----------------------------------------------------------------------------
 
 void bm_ryml_yaml_inplace(bm::State& st)
@@ -513,6 +675,66 @@ void bm_ryml_json_arena_reuse_nofilter_reserve(bm::State& st)
 }
 
 
+//-----------------------------------------------------------------------------
+
+void bm_rymlints_yaml_inplace_reuse(bm::State& st)
+{
+    c4::substr src = c4::to_substr(s_bm_case->in_place).trimr('\0');
+    s_bm_case->int_obj.data.resize(src);
+    for(auto _ : st)
+    {
+        s_bm_case->prepare(st, kResetInPlace);
+        parse_yaml_inplace(s_bm_case->filename, src, s_bm_case->int_obj.parser, &s_bm_case->int_obj.data);
+        bm::DoNotOptimize(s_bm_case->int_obj);
+    }
+    s_bm_case->report(st);
+}
+
+void bm_rymlints_yaml_inplace_reuse_nofilter(bm::State& st)
+{
+    c4::substr src = c4::to_substr(s_bm_case->in_place).trimr('\0');
+    s_bm_case->int_obj_nofilter.data.resize(src);
+    for(auto _ : st)
+    {
+        s_bm_case->prepare(st, kResetInPlace);
+        parse_yaml_inplace(s_bm_case->filename, src, s_bm_case->int_obj_nofilter.parser, &s_bm_case->int_obj_nofilter.data);
+        bm::DoNotOptimize(s_bm_case->int_obj_nofilter);
+    }
+    s_bm_case->report(st);
+}
+
+
+//-----------------------------------------------------------------------------
+
+void bm_rymlints_json_inplace_reuse(bm::State& st)
+{
+    c4::substr src = c4::to_substr(s_bm_case->in_place).trimr('\0');
+    s_bm_case->int_obj.data.resize(src);
+    for(auto _ : st)
+    {
+        ONLY_FOR_JSON;
+        s_bm_case->prepare(st, kResetInPlace);
+        parse_json_inplace(s_bm_case->filename, src, s_bm_case->int_obj.parser, &s_bm_case->int_obj.data);
+        bm::DoNotOptimize(s_bm_case->int_obj);
+    }
+    s_bm_case->report(st);
+}
+
+void bm_rymlints_json_inplace_reuse_nofilter(bm::State& st)
+{
+    c4::substr src = c4::to_substr(s_bm_case->in_place).trimr('\0');
+    s_bm_case->int_obj_nofilter.data.resize(src);
+    for(auto _ : st)
+    {
+        ONLY_FOR_JSON;
+        s_bm_case->prepare(st, kResetInPlace);
+        parse_json_inplace(s_bm_case->filename, src, s_bm_case->int_obj_nofilter.parser, &s_bm_case->int_obj_nofilter.data);
+        bm::DoNotOptimize(s_bm_case->int_obj_nofilter);
+    }
+    s_bm_case->report(st);
+}
+
+
 //-----------------------------------------------------------------------------
 
 void bm_ryml_yaml_inplace_reuse(bm::State& st)
@@ -617,6 +839,16 @@ void bm_ryml_json_inplace_reuse_nofilter_reserve(bm::State& st)
 
 //-----------------------------------------------------------------------------
 
+BENCHMARK(bm_rymlints_estimate);
+
+
+BENCHMARK(bm_rymlints_json_inplace_reuse_nofilter);
+BENCHMARK(bm_rymlints_json_inplace_reuse);
+
+BENCHMARK(bm_rymlints_yaml_inplace_reuse_nofilter);
+BENCHMARK(bm_rymlints_yaml_inplace_reuse);
+
+
 BENCHMARK(bm_ryml_json_inplace_reuse_nofilter_reserve);
 BENCHMARK(bm_ryml_json_inplace_reuse_nofilter);
 BENCHMARK(bm_ryml_json_inplace_reuse_reserve);
@@ -639,6 +871,17 @@ BENCHMARK(bm_ryml_yaml_arena_reuse_reserve);
 BENCHMARK(bm_ryml_yaml_arena_reuse);
 
 
+BENCHMARK(bm_rymlints_json_inplace_nofilter_reserve);
+BENCHMARK(bm_rymlints_json_inplace_nofilter);
+BENCHMARK(bm_rymlints_json_inplace_reserve);
+BENCHMARK(bm_rymlints_json_inplace);
+
+BENCHMARK(bm_rymlints_yaml_inplace_nofilter_reserve);
+BENCHMARK(bm_rymlints_yaml_inplace_nofilter);
+BENCHMARK(bm_rymlints_yaml_inplace_reserve);
+BENCHMARK(bm_rymlints_yaml_inplace);
+
+
 BENCHMARK(bm_ryml_json_inplace_nofilter_reserve);
 BENCHMARK(bm_ryml_json_inplace_nofilter);
 BENCHMARK(bm_ryml_json_inplace_reserve);