Merge remote-tracking branch 'origin/master' into user/dwayner/DML1.8…

…forORT1.10

cmake/CMakeLists.txt

@@ -108,6 +108,7 @@ option(onnxruntime_USE_ROCM "Build with AMD GPU support" OFF)
 option(onnxruntime_DISABLE_CONTRIB_OPS "Disable contrib ops" OFF)
 option(onnxruntime_DISABLE_ML_OPS "Disable traditional ML ops" OFF)
 option(onnxruntime_DISABLE_SPARSE_TENSORS "Disable sparse tensors data types" OFF)
+option(onnxruntime_DISABLE_OPTIONAL_TYPE "Disable optional type" OFF)
 option(onnxruntime_MINIMAL_BUILD "Exclude as much as possible from the build. Support ORT format models. No support for ONNX format models." OFF)
 cmake_dependent_option(onnxruntime_DISABLE_RTTI "Disable RTTI" ON "NOT onnxruntime_ENABLE_PYTHON" OFF)
 # For now onnxruntime_DISABLE_EXCEPTIONS will only work with onnxruntime_MINIMAL_BUILD, more changes (ONNX, non-CPU EP, ...) are required to run this standalone
@@ -647,16 +648,6 @@ if (${CMAKE_SYSTEM_NAME} MATCHES "Darwin")
     set(CMAKE_CXX_FLAGS "${CMAKE_CXX_FLAGS} -fstack-protector-strong")
 endif()
 
-if (${CMAKE_SYSTEM_NAME} MATCHES "iOSCross")
-    #For ios compliance
-    message("Adding flags for ios builds")
-    if (CMAKE_OSX_ARCHITECTURES STREQUAL "arm64")
-      set(CMAKE_CXX_FLAGS "${CMAKE_CXX_FLAGS} -target arm64-apple-darwin-macho")
-    elseif (CMAKE_OSX_ARCHITECTURES STREQUAL "arm")
-      set(CMAKE_CXX_FLAGS "${CMAKE_CXX_FLAGS} -target armv7a-apple-darwin-macho")
-    endif()
-endif()
-
 #Dependencies begin
 if (onnxruntime_BUILD_UNIT_TESTS)
   if (onnxruntime_PREFER_SYSTEM_LIB)
@@ -817,6 +808,10 @@ if (onnxruntime_DISABLE_SPARSE_TENSORS)
   add_compile_definitions(DISABLE_SPARSE_TENSORS)
 endif()
 
+if (onnxruntime_DISABLE_OPTIONAL_TYPE)
+  add_compile_definitions(DISABLE_OPTIONAL_TYPE)
+endif()
+
 if (onnxruntime_USE_CUDA AND "${onnxruntime_CUDNN_HOME}" STREQUAL "")
   message(FATAL_ERROR "onnxruntime_CUDNN_HOME required for onnxruntime_USE_CUDA")
 endif()
@@ -1216,7 +1211,7 @@ endfunction()
 
 #For plugins that are not linked into other targets but may be loaded dynamically at runtime using dlopen-like functionality.
 function(onnxruntime_add_shared_library_module target_name)
-  if ((${CMAKE_SYSTEM_NAME} MATCHES "Darwin") OR (${CMAKE_SYSTEM_NAME} MATCHES "iOSCross") OR (${CMAKE_SYSTEM_NAME} MATCHES "iOS"))
+  if ((${CMAKE_SYSTEM_NAME} MATCHES "Darwin") OR (${CMAKE_SYSTEM_NAME} MATCHES "iOS"))
     add_library(${target_name} SHARED ${ARGN})
   else()
     #On Windows, this target shouldn't generate an import lib, but I don't know how to disable it.
@@ -1229,10 +1224,7 @@ function(onnxruntime_add_shared_library_module target_name)
   endif()
 endfunction()
 
-function(onnxruntime_add_executable target_name)
-  if (${CMAKE_SYSTEM_NAME} MATCHES "iOSCross")
-    message(FATAL_ERROR "iOS doesn't support commmand line tool")
-  endif()
+function(onnxruntime_add_executable target_name)  
   add_executable(${target_name} ${ARGN})
   onnxruntime_configure_target(${target_name})
   if (onnxruntime_target_platform STREQUAL "x86" AND NOT onnxruntime_BUILD_WEBASSEMBLY)

cmake/onnxruntime_mlas.cmake

@@ -26,6 +26,7 @@ onnxruntime_add_static_library(onnxruntime_mlas
   ${MLAS_SRC_DIR}/qlmul.cpp
   ${MLAS_SRC_DIR}/qpostprocessor.cpp
   ${MLAS_SRC_DIR}/qlgavgpool.cpp
+  ${MLAS_SRC_DIR}/qdwconv_kernelsize.cpp
 )
 
 set(ONNXRUNTIME_MLAS_LIBS onnxruntime_mlas)
@@ -46,6 +47,7 @@ function(setup_mlas_source_for_windows)
 
       set(mlas_platform_preprocess_srcs
         ${MLAS_SRC_DIR}/arm64/DepthwiseConvsymKernelNeon.asm
+        ${MLAS_SRC_DIR}/arm64/DepthwiseQConvKernelSize9Neon.asm
         ${MLAS_SRC_DIR}/arm64/QgemmU8X8KernelNeon.asm
         ${MLAS_SRC_DIR}/arm64/QgemmS8S8KernelNeon.asm
         ${MLAS_SRC_DIR}/arm64/QgemmU8X8KernelUdot.asm
@@ -265,6 +267,7 @@ else()
         enable_language(ASM)
         set(mlas_platform_srcs
           ${MLAS_SRC_DIR}/aarch64/DepthwiseConvSymKernelNeon.S
+          ${MLAS_SRC_DIR}/aarch64/DepthwiseQConvKernelSize9Neon.S
           ${MLAS_SRC_DIR}/aarch64/QgemmU8X8KernelNeon.S
           ${MLAS_SRC_DIR}/aarch64/QgemmS8S8KernelNeon.S
           ${MLAS_SRC_DIR}/aarch64/QgemmU8X8KernelUdot.S

docs/OperatorKernels.md

@@ -600,6 +600,10 @@ Do not modify directly.*
 |QuantizeLinear|*in* x:**T1**<br> *in* y_scale:**tensor(float)**<br> *in* y_zero_point:**T2**<br> *out* y:**T2**|10+|**T1** = tensor(float)<br/> **T2** = tensor(int8), tensor(uint8)|
 |RNN|*in* X:**T**<br> *in* W:**T**<br> *in* R:**T**<br> *in* B:**T**<br> *in* sequence_lens:**T1**<br> *in* initial_h:**T**<br> *out* Y:**T**<br> *out* Y_h:**T**|14+|**T** = tensor(double), tensor(float), tensor(float16)<br/> **T1** = tensor(int32)|
 |||[7, 13]|**T** = tensor(double), tensor(float), tensor(float16)<br/> **T1** = tensor(int32)|
+|RandomNormal|*out* output:**T**|1+|**T** = tensor(double), tensor(float), tensor(float16)|
+|RandomNormalLike|*in* input:**T1**<br> *out* output:**T2**|1+|**T1** = tensor(bfloat16), tensor(bool), tensor(double), tensor(float), tensor(float16), tensor(int16), tensor(int32), tensor(int64), tensor(int8), tensor(string), tensor(uint16), tensor(uint32), tensor(uint64), tensor(uint8)<br/> **T2** = tensor(double), tensor(float), tensor(float16)|
+|RandomUniform|*out* output:**T**|1+|**T** = tensor(double), tensor(float), tensor(float16)|
+|RandomUniformLike|*in* input:**T1**<br> *out* output:**T2**|1+|**T1** = tensor(bfloat16), tensor(bool), tensor(double), tensor(float), tensor(float16), tensor(int16), tensor(int32), tensor(int64), tensor(int8), tensor(string), tensor(uint16), tensor(uint32), tensor(uint64), tensor(uint8)<br/> **T2** = tensor(double), tensor(float), tensor(float16)|
 |Range|*in* start:**T**<br> *in* limit:**T**<br> *in* delta:**T**<br> *out* output:**T**|11+|**T** = tensor(double), tensor(float), tensor(int16), tensor(int32), tensor(int64)|
 |Reciprocal|*in* X:**T**<br> *out* Y:**T**|13+|**T** = tensor(double), tensor(float), tensor(float16)|
 |||[6, 12]|**T** = tensor(double), tensor(float), tensor(float16)|

include/onnxruntime/core/common/status.h

@@ -47,7 +47,7 @@ enum StatusCode {
   EP_FAIL = 11
 };
 
-inline const char* StatusCodeToString(StatusCode status) noexcept {
+constexpr const char* StatusCodeToString(StatusCode status) noexcept {
   switch (status) {
     case StatusCode::OK:
       return "SUCCESS";
@@ -79,7 +79,7 @@ inline const char* StatusCodeToString(StatusCode status) noexcept {
 }
 
 #ifdef _WIN32
-inline HRESULT StatusCodeToHRESULT(StatusCode status) noexcept {
+constexpr HRESULT StatusCodeToHRESULT(StatusCode status) noexcept {
     switch (status)
     {
     case StatusCode::OK:
@@ -89,23 +89,23 @@ inline HRESULT StatusCodeToHRESULT(StatusCode status) noexcept {
     case StatusCode::INVALID_ARGUMENT:
         return E_INVALIDARG;
     case StatusCode::NO_SUCHFILE:
-        return __HRESULT_FROM_WIN32(ERROR_FILE_NOT_FOUND);
+        return HRESULT_FROM_WIN32(ERROR_FILE_NOT_FOUND);
     case StatusCode::NO_MODEL:
-        return __HRESULT_FROM_WIN32(ERROR_FILE_NOT_FOUND);
+        return HRESULT_FROM_WIN32(ERROR_FILE_NOT_FOUND);
     case StatusCode::ENGINE_ERROR:
         return E_FAIL;
     case StatusCode::RUNTIME_EXCEPTION:
         return E_FAIL;
     case StatusCode::INVALID_PROTOBUF:
-        return __HRESULT_FROM_WIN32(ERROR_FILE_CORRUPT);
+        return HRESULT_FROM_WIN32(ERROR_FILE_CORRUPT);
     case StatusCode::MODEL_LOADED:
-        return __HRESULT_FROM_WIN32(ERROR_INTERNAL_ERROR);
+        return HRESULT_FROM_WIN32(ERROR_INTERNAL_ERROR);
     case StatusCode::NOT_IMPLEMENTED:
         return E_NOTIMPL;
     case StatusCode::INVALID_GRAPH:
-        return __HRESULT_FROM_WIN32(ERROR_FILE_CORRUPT);
+        return HRESULT_FROM_WIN32(ERROR_FILE_CORRUPT);
     case StatusCode::EP_FAIL:
-        return __HRESULT_FROM_WIN32(ERROR_INTERNAL_ERROR);
+        return HRESULT_FROM_WIN32(ERROR_INTERNAL_ERROR);
     default:
         return E_FAIL;
     }