Added ./example & improved Simd usage

WillisMedwell · Feb 12, 2024 · f6d9d6e · f6d9d6e
1 parent 6a7b188
commit f6d9d6e
Show file tree

Hide file tree

Showing 15 changed files with 648 additions and 1,898 deletions.
diff --git a/.github/workflows/gcc.yml b/.github/workflows/gcc.yml
@@ -12,7 +12,7 @@ jobs:
 
     - name: Build With GCC
       run: |
-        cmake --preset gcc-ninja -DCMAKE_BUILD_TYPE=Release
+        cmake --preset gcc-ninja -DCMAKE_BUILD_TYPE=Release 
         cmake --build ./build/gcc-ninja --target UtilyTest --config Debug
         cmake --build ./build/gcc-ninja --target UtilyBenchmark --config Release
 

diff --git a/CMakeLists.txt b/CMakeLists.txt
@@ -21,7 +21,7 @@ if(DEFINED EMSCRIPTEN)
     message(STATUS "For Emscripten SIMD use flags: \"-msimd128 -mrelaxed-simd -msse -msse2 -msse3 -msse4.1\"")
 elseif(NOT MSVC)
     target_compile_options(Utily_Utily PRIVATE "-Wall" "-Wextra" "-Wpedantic" "-Wcast-align" "-Wcast-qual" "-Wctor-dtor-privacy" "-Wformat=2" "-Winit-self" "-Wmissing-declarations" "-Wmissing-include-dirs" "-Wold-style-cast" "-Woverloaded-virtual" "-Wredundant-decls" "-Wshadow" "-Wsign-conversion" "-Wsign-promo" "-Wstrict-overflow=5" "-Wundef" "-Wno-unused" "-Wconversion" "-Wsign-compare")
-    message(STATUS "For Native SIMD use flags: \"-mtune=native\"")
+    message(STATUS "For Native SIMD use flags: \"-march=native\"")
 endif()
 
 set_property(TARGET Utily_Utily PROPERTY EXPORT_NAME Utily)
@@ -57,12 +57,9 @@ if(BUILD_UTILY_TESTS)
         set(CMAKE_EXECUTABLE_SUFFIX ".html")
     elseif(NOT MSVC)
         target_compile_options(UtilyTest PRIVATE "-Wall" "-Wextra" "-Wpedantic" "-Wcast-align" "-Wcast-qual" "-Wctor-dtor-privacy" "-Wformat=2" "-Winit-self" "-Wmissing-declarations" "-Wmissing-include-dirs" "-Wold-style-cast" "-Woverloaded-virtual" "-Wredundant-decls" "-Wshadow" "-Wsign-conversion" "-Wsign-promo" "-Wstrict-overflow=5" "-Wundef" "-Wno-unused" "-Wconversion" "-Wsign-compare")
-        # target_compile_options(UtilyTest PRIVATE -msse -msse2 -msse3 -msse4.1 -mavx512f -mavx512bw)
-        # target_link_options(UtilyTest PRIVATE -msse -msse2 -msse3 -msse4.1 -mavx512f -mavx512bw)
-        target_compile_options(UtilyTest PRIVATE -mtune=native)
-        target_link_options(UtilyTest PRIVATE -mtune=native)
+        target_compile_options(UtilyTest PRIVATE -march=native)
+        target_link_options(UtilyTest PRIVATE -march=native)
 
-
         if (CMAKE_CXX_COMPILER_ID STREQUAL "GNU")
             target_compile_options(UtilyTest PRIVATE -Wuseless-cast)
         endif()
@@ -109,18 +106,22 @@ if(BUILD_UTILY_BENCHMARKS)
         target_compile_options(UtilyBenchmark PRIVATE -msimd128 -mrelaxed-simd -msse -msse2 -msse3 -msse4.1)
     elseif(NOT MSVC)
         target_compile_options(UtilyBenchmark PRIVATE "-Wall" "-Wextra" "-Wpedantic" "-Wcast-align" "-Wcast-qual" "-Wctor-dtor-privacy" "-Wformat=2" "-Winit-self" "-Wmissing-declarations" "-Wmissing-include-dirs" "-Wold-style-cast" "-Woverloaded-virtual" "-Wredundant-decls" "-Wshadow" "-Wsign-conversion" "-Wsign-promo" "-Wstrict-overflow=5" "-Wswitch-default" "-Wundef" "-Wno-unused" "-Wconversion" "-Wsign-compare")
-        # target_compile_options(UtilyBenchmark PRIVATE -msse -msse2 -msse3 -msse4.1 -mavx512f -mavx512bw)
-        # target_link_options(UtilyBenchmark PRIVATE -msse -msse2 -msse3 -msse4.1 -mavx512f -mavx512bw)
-        target_compile_options(UtilyTest PRIVATE -mtune=native)
-        target_link_options(UtilyTest PRIVATE -mtune=native)
+        target_compile_options(UtilyBenchmark PRIVATE -march=native)
+        target_link_options(UtilyBenchmark PRIVATE -march=native)
         target_compile_options(UtilyBenchmark PRIVATE
             $<$<CONFIG:Release>:-O3 -DNDEBUG>
         )
 
         if (CMAKE_CXX_COMPILER_ID STREQUAL "GNU")
             target_compile_options(UtilyBenchmark PRIVATE -Wuseless-cast)
         endif()
+    else()
+
     endif()
     target_link_libraries(UtilyBenchmark PRIVATE benchmark::benchmark benchmark::benchmark_main Utily::Utily)
 endif()
 
+# Example Setup
+if(BUILD_UTILY_EXAMPLES)
+    add_subdirectory(example)
+endif()
diff --git a/CMakePresets.json b/CMakePresets.json
@@ -15,6 +15,7 @@
                 "CMAKE_BUILD_TYPE": "Release",
                 "BUILD_UTILY_TESTS": "ON",
                 "BUILD_UTILY_BENCHMARKS": "ON",
+                "BUILD_UTILY_EXAMPLES": "ON",
                 "HAVE_STEADY_CLOCK": "0",
                 "HAVE_POSIX_REGEX": "0",
                 "HAVE_STD_REGEX": "0",
@@ -33,6 +34,7 @@
             "cacheVariables": {
                 "BUILD_UTILY_TESTS": "ON",
                 "BUILD_UTILY_BENCHMARKS": "ON",
+                "BUILD_UTILY_EXAMPLES": "ON",
                 "HAVE_STEADY_CLOCK": "0",
                 "HAVE_POSIX_REGEX": "0",
                 "HAVE_STD_REGEX": "0",
@@ -52,6 +54,7 @@
                 "CMAKE_BUILD_TYPE": "Release",
                 "BUILD_UTILY_TESTS": "ON",
                 "BUILD_UTILY_BENCHMARKS": "ON",
+                "BUILD_UTILY_EXAMPLES": "ON",
                 "HAVE_STEADY_CLOCK": "0",
                 "HAVE_POSIX_REGEX": "0",
                 "HAVE_STD_REGEX": "0",

diff --git a/benchmark/BenchSimd.cpp b/benchmark/BenchSimd.cpp
@@ -1,18 +1,15 @@
 #include "Utily/Utily.hpp"
 
+#include "Utily/Simd128.hpp"
+#include "Utily/Simd512.hpp"
+
 #include <benchmark/benchmark.h>
 
 #include <algorithm>
 #include <array>
 #include <numeric>
 #include <ranges>
 
-#ifndef EMSCRIPTEN
-extern "C" {
-#include "stringzilla.h"
-}
-#endif
-
 #if 1
 
 auto all_a(size_t max) -> std::string {
@@ -49,17 +46,6 @@ static void BM_Uty_find_char_512(benchmark::State& state) {
 }
 BENCHMARK(BM_Uty_find_char_512);
 
-#ifndef EMSCRIPTEN
-static void BM_Zil_find_char(benchmark::State& state) {
-    auto delim = 'z';
-    for (auto _ : state) {
-        volatile auto iter = sz_find_1char(LONG_STRING.data(), LONG_STRING.size(), &delim);
-        benchmark::DoNotOptimize(iter);
-    }
-}
-BENCHMARK(BM_Zil_find_char);
-#endif
-
 static void BM_Std_find_char(benchmark::State& state) {
     for (auto _ : state) {
         volatile auto iter = std::find(LONG_STRING.begin(), LONG_STRING.end(), 'z');
@@ -71,7 +57,9 @@ BENCHMARK(BM_Std_find_char);
 static void BM_Uty_find_first_of_char(benchmark::State& state) {
     const auto data = std::to_array({ 'z', 'o', 'n' });
     for (auto _ : state) {
-        volatile auto iter = Utily::Simd128::Char::find_first_of(LONG_STRING.data(), LONG_STRING.size(), data.data(), data.size());
+        // Alternate, less performant version.
+        // volatile auto iter = Utily::Simd128::Char::find_first_of(LONG_STRING.data(), LONG_STRING.size(), data.data(), data.size()); 
+        volatile auto iter = Utily::Simd128::Char::find_first_of<3>(LONG_STRING.data(), LONG_STRING.size(), data.data());
         benchmark::DoNotOptimize(iter);
     }
 }
@@ -112,21 +100,6 @@ static void BM_Uty_search_char_4letters_512(benchmark::State& state) {
 }
 BENCHMARK(BM_Uty_search_char_4letters_512);
 
-#ifndef EMSCRIPTEN
-static void BM_Zil_search_char_4letters(benchmark::State& state) {
-    std::string_view find = "stri";
-    for (auto _ : state) {
-        volatile auto iter = sz_find_substring(
-            LONG_STRING.data(),
-            LONG_STRING.size(),
-            find.data(),
-            find.size());
-        benchmark::DoNotOptimize(iter);
-    }
-}
-BENCHMARK(BM_Zil_search_char_4letters);
-#endif
-
 static void BM_Std_search_char_4letters(benchmark::State& state) {
     std::string_view find = "stri";
     for (auto _ : state) {
@@ -153,17 +126,6 @@ static void BM_Uty_search_char_8letters(benchmark::State& state) {
 }
 BENCHMARK(BM_Uty_search_char_8letters);
 
-#ifndef EMSCRIPTEN
-static void BM_Zil_search_char_8letters(benchmark::State& state) {
-    std::string_view find = "stringer";
-    for (auto _ : state) {
-        auto index = sz_find_substring(LONG_STRING.data(), LONG_STRING.size(), find.data(), find.size());
-        benchmark::DoNotOptimize(index);
-    }
-}
-BENCHMARK(BM_Zil_search_char_8letters);
-#endif
-
 static void BM_Std_search_char_8letters(benchmark::State& state) {
     std::string_view find = "stringer";
     for (auto _ : state) {