Refactor math simd

1k/1kiss.ps1

@@ -895,6 +895,7 @@ function setup_cmake($skipOS = $false, $scope = 'local') {
             else {
                 & "$cmake_pkg_path" '--skip-license' '--prefix=/usr/local' 1>$null 2>$null
             }
+            if (!$?) { Remove-Item $cmake_pkg_path -Force }
         }
 
         $cmake_prog, $_ = find_prog -name 'cmake' -path $cmake_bin -silent $true

1k/fetch.cmake

@@ -20,10 +20,16 @@ function(_1kfetch_init)
         set(_1kfetch_manifest "${_1kfetch_manifest}" CACHE STRING "" FORCE)
     endif()
 
+    if(NOT EXISTS ${PWSH_PROG}) # try again
+        unset(PWSH_PROG CACHE)
+        find_program(PWSH_PROG NAMES pwsh powershell NO_PACKAGE_ROOT_PATH NO_CMAKE_PATH NO_CMAKE_ENVIRONMENT_PATH NO_CMAKE_SYSTEM_PATH NO_CMAKE_FIND_ROOT_PATH)
+    endif()
+
     execute_process(COMMAND ${PWSH_PROG} ${CMAKE_CURRENT_FUNCTION_LIST_DIR}/resolv-uri.ps1
         -name "1kdist"
         -manifest ${_1kfetch_manifest}
         OUTPUT_VARIABLE _1kdist_url
+        RESULT_VARIABLE _1kdist_error
     )
 
     if(_1kdist_url)
@@ -33,7 +39,7 @@ function(_1kfetch_init)
         set(_1kdist_base_url "${_1kdist_base_url}/${_1kdist_ver}" PARENT_SCOPE)
         set(_1kdist_ver ${_1kdist_ver} PARENT_SCOPE)
     else()
-        message(WARNING "Resolve 1kdist uri fail, the _1kfetch_dist will not work")
+        message(WARNING "Resolve 1kdist uri fail, ${_1kdist_error}, the _1kfetch_dist will not work")
     endif()
 endfunction()
 

3rdparty/README.md

@@ -248,7 +248,7 @@
 
 ## yasio
 - [![Upstream](https://img.shields.io/github/v/release/yasio/yasio?label=Upstream)](https://github.com/yasio/yasio)
-- Version: 4.2.3
+- Version: 4.2.4
 - License: MIT WITH Anti-996
 
 ## zlib

3rdparty/yasio/yasio/bindings/yasio_ni.cpp

@@ -60,14 +60,14 @@ YASIO_NI_API void yasio_init_globals(void(YASIO_INTEROP_DECL* pfn)(int level, co
 YASIO_NI_API void yasio_cleanup_globals() { io_service::cleanup_globals(); }
 
 struct yasio_io_event {
-  int kind; //
-  int channel;
-  void* thandle;
+  int kind; // event kind
+  int channel; // channel index
+  void* thandle; // transport
   union {
-    void* msg;
-    int status; //
+    void* hmsg; // io_packet*
+    int ec; // error code
   };
-  void* user;
+  void* user; // user data
 };
 
 YASIO_NI_API void* yasio_create_service(int channel_count, void(YASIO_INTEROP_DECL* event_cb)(yasio_io_event* event), void* user)
@@ -82,9 +82,9 @@ YASIO_NI_API void* yasio_create_service(int channel_count, void(YASIO_INTEROP_DE
     event.thandle = e->transport();
     event.user    = user;
     if (event.kind == yasio::YEK_ON_PACKET)
-      event.msg = !is_packet_empty(pkt) ? &pkt : nullptr;
+      event.hmsg = !is_packet_empty(pkt) ? &pkt : nullptr;
     else
-      event.status = e->status();
+      event.ec = e->status();
     event_cb(&event);
   });
   return service;
@@ -157,8 +157,12 @@ YASIO_NI_API void yasio_set_option(void* service_ptr, int opt, const char* pszAr
   std::array<cxx17::string_view, YASIO_MAX_OPTION_ARGC> args;
   int argc = 0;
   yasio::split_if(&strArgs.front(), ';', [&](char* s, char* e) {
-    *e           = '\0'; // to c style string
-    args[argc++] = cxx17::string_view(s, e - s);
+    if (e) {
+        *e           = '\0'; // to c style string
+        args[argc++] = cxx17::string_view(s, e - s);
+    } else {
+        args[argc++] = cxx17::string_view{s};
+    }
     return (argc < YASIO_MAX_OPTION_ARGC);
   });
 

3rdparty/yasio/yasio/config.hpp

@@ -205,7 +205,7 @@ SOFTWARE.
 /*
 **  The yasio version macros
 */
-#define YASIO_VERSION_NUM 0x040203
+#define YASIO_VERSION_NUM 0x040204
 
 /*
 ** The macros used by io_service.

3rdparty/yasio/yasio/impl/eventfd_select_interrupter.hpp

@@ -20,11 +20,11 @@
 #include <sys/stat.h>
 #include <sys/types.h>
 #include <fcntl.h>
-#if defined(__GLIBC__) && (__GLIBC__ == 2 && __GLIBC_MINOR__ < 8)
-#  include <asm/unistd.h>
-#else // __GLIBC__ == 2 && __GLIBC_MINOR__ < 8
+#if defined(__GLIBC__) && (__GLIBC__ == 2 && __GLIBC_MINOR__ < 8) && !defined(__UCLIBC__)
+#  include <asm/unistd.h> // for syscall without API: eventfd
+#else
 #  include <sys/eventfd.h>
-#endif // __GLIBC__ == 2 && __GLIBC_MINOR__ < 8
+#endif
 
 #include <unistd.h>
 
@@ -105,7 +105,7 @@ class eventfd_select_interrupter {
   // Open the descriptors. Throws on error.
   inline void open_descriptors()
   {
-#if defined(__GLIBC__) && (__GLIBC__ == 2 && __GLIBC_MINOR__ < 8)
+#if defined(__GLIBC__) && (__GLIBC__ == 2 && __GLIBC_MINOR__ < 8) && !defined(__UCLIBC__)
     write_descriptor_ = read_descriptor_ = syscall(__NR_eventfd, 0);
     if (read_descriptor_ != -1)
     {

3rdparty/yasio/yasio/logging.hpp

@@ -40,6 +40,9 @@ inline void yasio__print(std::string&& message) { ::write(::fileno(stdout), mess
 #  include <android/log.h>
 #  include <jni.h>
 #  define YASIO_LOG_TAG(tag, format, ...) __android_log_print(ANDROID_LOG_INFO, "yasio", (tag format), ##__VA_ARGS__)
+#elif defined(__OHOS__)
+#  include <hilog/log.h>
+#  define YASIO_LOG_TAG(tag, format, ...) OH_LOG_INFO(LOG_APP, (tag format "\n"), ##__VA_ARGS__)
 #else
 #  define YASIO_LOG_TAG(tag, format, ...) printf((tag format "\n"), ##__VA_ARGS__)
 #endif

3rdparty/yasio/yasio/xxsocket.cpp

@@ -209,7 +209,7 @@ int xxsocket::pserve(const endpoint& ep)
   if (!this->reopen(ep.af()))
     return -1;
 
-  set_optval(SOL_SOCKET, SO_REUSEADDR, 1);
+  this->reuse_address(true);
 
   int n = this->bind(ep);
   if (n != 0)

CMakeOptions.md

@@ -52,6 +52,7 @@ default is: `navigator.hardwareConcurrency`
 - AX_WASM_SHELL_FILE: specify the wasm shell file, by default use `${_AX_ROOT}/core/platform/wasm/shell_minimal.html`
 - AX_WASM_ENABLE_DEVTOOLS: whether enable web devtools aka `pause`, `resume`, `step` buttons in webpage, default: `TRUE`
 - AX_WASM_INITIAL_MEMORY: set the wasm initial memory size, default `1024MB`
+- AX_WASM_ISA_SIMD: specify the wasm simd intrinsics type, default `none`, supports `sse`, `neon`, note the `wasm-simd` not support by axmol yet
 
 ## The options for axmol apps
 

cmake/Modules/AXConfigDefine.cmake

@@ -186,22 +186,21 @@ endfunction()
 
 if(EMSCRIPTEN)
     set(AX_WASM_THREADS "4" CACHE STRING "Wasm threads count")
-
-    set(_AX_WASM_THREADS_INT 0)
+    set(_threads_hint "")
     if (AX_WASM_THREADS STREQUAL "auto") # not empty string or not 0
         # Enable pthread support globally
+        set(_threads_hint "(auto)")
         include(ProcessorCount)
+        set(_AX_WASM_THREADS_INT 0)
         ProcessorCount(_AX_WASM_THREADS_INT)
-    elseif(AX_WASM_THREADS MATCHES "^([0-9]+)$" OR AX_WASM_THREADS STREQUAL "navigator.hardwareConcurrency")
-        set(_AX_WASM_THREADS_INT ${AX_WASM_THREADS})
+        set(AX_WASM_THREADS "${_AX_WASM_THREADS_INT}" CACHE STRING "Wasm threads count" FORCE)
     endif()
 
-    message(STATUS "AX_WASM_THREADS=${AX_WASM_THREADS}")
-    message(STATUS "_AX_WASM_THREADS_INT=${_AX_WASM_THREADS_INT}")
+    message(STATUS "AX_WASM_THREADS=${AX_WASM_THREADS}${_threads_hint}")
 
-    if (_AX_WASM_THREADS_INT)
+    if(AX_WASM_THREADS MATCHES "^([0-9]+)$" OR AX_WASM_THREADS STREQUAL "navigator.hardwareConcurrency")
         list(APPEND _ax_compile_options -pthread)
-        add_link_options(-pthread -sPTHREAD_POOL_SIZE=${_AX_WASM_THREADS_INT})
+        add_link_options(-pthread -sPTHREAD_POOL_SIZE=${AX_WASM_THREADS})
     endif()
 
     set(AX_WASM_INITIAL_MEMORY "1024MB" CACHE STRING "")

core/CMakeLists.txt

@@ -400,9 +400,43 @@ if(WINDOWS)
     endif()
 endif()
 
-# AX_USE_SSE
-if (AX_ISA_SIMD MATCHES "sse")
-    target_compile_definitions(${_AX_CORE_LIB} PUBLIC AX_USE_SSE=1)
+# axmol math simd intrinsics support
+set(_simdc_defines)
+set(_simdc_options)
+if (NOT WASM) # native platforms auto detect from cmake or preprocessor check
+    if (AX_ISA_SIMD MATCHES "sse")
+        list(APPEND _simdc_defines AX_SSE_INTRINSICS=1)
+        if (AX_ISA_SIMD MATCHES "sse4")
+            list(APPEND _simdc_defines __SSE4_1__=1)
+            if (LINUX)
+                list(APPEND _simdc_options -msse4.1)
+            endif()
+        endif()
+    endif()
+else() # wasm requires user specify SIMD intrinsics manually
+    set(AX_WASM_ISA_SIMD "none" CACHE STRING "")
+    string(TOLOWER ${AX_WASM_ISA_SIMD} AX_WASM_ISA_SIMD)
+    if(AX_WASM_ISA_SIMD MATCHES "sse")
+        message(AUTHOR_WARNING "Using SSE intrinsics for WASM ...")
+        list(APPEND _simdc_defines AX_SSE_INTRINSICS=1 __SSE__=1 __SSE2__=1)
+        list(APPEND _simdc_options -msse -msse2)
+        if(AX_ISA_LEVEL GREATER_EQUAL 2)
+            list(APPEND _simdc_defines __SSE4_1__=1)
+            list(APPEND _simdc_options -msse4.1)
+        endif()
+        list(APPEND _simdc_options -msimd128)
+    elseif(AX_WASM_ISA_SIMD MATCHES "neon")
+        message(AUTHOR_WARNING "Using NEON intrinsics for WASM ...")
+        list(APPEND _simdc_defines AX_NEON_INTRINSICS=1)
+        list(APPEND _simdc_options -mfpu=neon -msimd128)
+    endif()
+endif()
+
+if(_simdc_defines)
+    target_compile_definitions(${_AX_CORE_LIB} PUBLIC ${_simdc_defines})
+    if(_simdc_options)
+        target_compile_options(${_AX_CORE_LIB} PUBLIC ${_simdc_options})
+    endif()
 endif()
 
 # engine extensions

core/base/Configuration.cpp

@@ -71,7 +71,7 @@ bool Configuration::init()
 #if AX_ENABLE_PROFILERS
     _valueDict["axmol.compiled_with_profiler"] = Value(true);
 #else
-    _valueDict["axmol.compiled_with_profiler"]       = Value(false);
+    _valueDict["axmol.compiled_with_profiler"] = Value(false);
 #endif
 
 #if AX_ENABLE_GL_STATE_CACHE == 0
@@ -83,7 +83,17 @@ bool Configuration::init()
 #if _AX_DEBUG
     _valueDict["axmol.build_type"] = Value("DEBUG");
 #else
-    _valueDict["axmol.build_type"]                   = Value("RELEASE");
+    _valueDict["axmol.build_type"] = Value("RELEASE");
+#endif
+
+#if defined(AX_SSE_INTRINSICS)
+#    if defined(__SSE4_1__)
+    _valueDict["axmol.simd"] = Value("SSE41");
+#    else
+    _valueDict["axmol.simd"] = Value("SSE2");
+#    endif
+#elif defined(AX_NEON_INTRINSICS)
+    _valueDict["axmol.simd"] = Value("NEON");
 #endif
 
     return true;

core/base/Console.cpp

@@ -398,7 +398,7 @@ bool Console::listenOnTCP(int port)
     if (sock.pserve(ep) != 0)
     {
         int ec = xxsocket::get_last_errno();
-        AXLOGW("Console: open server failed, ec:{}", ec);
+        AXLOGW("Console: open server failed, ec:{}, {}", ec, xxsocket::strerror(ec));
         return false;
     }
 

core/math/Mat4.cpp

@@ -17,7 +17,7 @@
 
  Original file from GamePlay3D: http://gameplay3d.org
 
- This file was modified to fit the cocos2d-x project
+ This file was modified to fit the axmol project
  */
 
 #include "math/Mat4.h"
@@ -459,11 +459,7 @@ void Mat4::add(float scalar)
 void Mat4::add(float scalar, Mat4* dst)
 {
     GP_ASSERT(dst);
-#ifdef AX_USE_SSE
-    MathUtil::addMatrix(col, scalar, dst->col);
-#else
     MathUtil::addMatrix(m, scalar, dst->m);
-#endif
 }
 
 void Mat4::add(const Mat4& mat)
@@ -474,11 +470,7 @@ void Mat4::add(const Mat4& mat)
 void Mat4::add(const Mat4& m1, const Mat4& m2, Mat4* dst)
 {
     GP_ASSERT(dst);
-#ifdef AX_USE_SSE
-    MathUtil::addMatrix(m1.col, m2.col, dst->col);
-#else
     MathUtil::addMatrix(m1.m, m2.m, dst->m);
-#endif
 }
 
 bool Mat4::decompose(Vec3* scale, Quaternion* rotation, Vec3* translation) const
@@ -751,11 +743,7 @@ void Mat4::multiply(float scalar, Mat4* dst) const
 void Mat4::multiply(const Mat4& m, float scalar, Mat4* dst)
 {
     GP_ASSERT(dst);
-#ifdef AX_USE_SSE
-    MathUtil::multiplyMatrix(m.col, scalar, dst->col);
-#else
     MathUtil::multiplyMatrix(m.m, scalar, dst->m);
-#endif
 }
 
 void Mat4::multiply(const Mat4& mat)
@@ -766,20 +754,12 @@ void Mat4::multiply(const Mat4& mat)
 void Mat4::multiply(const Mat4& m1, const Mat4& m2, Mat4* dst)
 {
     GP_ASSERT(dst);
-#ifdef AX_USE_SSE
-    MathUtil::multiplyMatrix(m1.col, m2.col, dst->col);
-#else
     MathUtil::multiplyMatrix(m1.m, m2.m, dst->m);
-#endif
 }
 
 void Mat4::negate()
 {
-#ifdef AX_USE_SSE
-    MathUtil::negateMatrix(col, col);
-#else
     MathUtil::negateMatrix(m, m);
-#endif
 }
 
 Mat4 Mat4::getNegated() const
@@ -945,11 +925,7 @@ void Mat4::subtract(const Mat4& mat)
 void Mat4::subtract(const Mat4& m1, const Mat4& m2, Mat4* dst)
 {
     GP_ASSERT(dst);
-#ifdef AX_USE_SSE
-    MathUtil::subtractMatrix(m1.col, m2.col, dst->col);
-#else
     MathUtil::subtractMatrix(m1.m, m2.m, dst->m);
-#endif
 }
 
 void Mat4::transformVector(Vec3* vector) const
@@ -967,7 +943,7 @@ void Mat4::transformVector(float x, float y, float z, float w, Vec3* dst) const
 {
     GP_ASSERT(dst);
 
-    MathUtil::transformVec4(m, x, y, z, w, (float*)dst);
+    MathUtil::transformVec4(m, x, y, z, w, reinterpret_cast<float*>(dst));
 }
 
 void Mat4::transformVector(Vec4* vector) const
@@ -979,14 +955,7 @@ void Mat4::transformVector(Vec4* vector) const
 void Mat4::transformVector(const Vec4& vector, Vec4* dst) const
 {
     GP_ASSERT(dst);
-#ifdef AX_USE_SSE
-    alignas(16) Vec4 inVal{vector};
-    alignas(16) Vec4 outVal;
-    MathUtil::transformVec4(col, reinterpret_cast<const __m128&>(inVal), reinterpret_cast<__m128&>(outVal));
-    *dst = outVal;
-#else
-    MathUtil::transformVec4(m, (const float*)&vector, (float*)dst);
-#endif
+    MathUtil::transformVec4(m, reinterpret_cast<const float*>(&vector), reinterpret_cast<float*>(dst));
 }
 
 void Mat4::translate(float x, float y, float z)
@@ -1013,11 +982,7 @@ void Mat4::translate(const Vec3& t, Mat4* dst) const
 
 void Mat4::transpose()
 {
-#ifdef AX_USE_SSE
-    MathUtil::transposeMatrix(col, col);
-#else
     MathUtil::transposeMatrix(m, m);
-#endif
 }
 
 Mat4 Mat4::getTransposed() const