Merge branch 'microsoft:main' into aix-main

cgmanifests/generated/cgmanifest.json

@@ -26,7 +26,7 @@
       "component": {
         "type": "git",
         "git": {
-          "commitHash": "990217f043af7222348ca8f0301e17fa7b841781",
+          "commitHash": "595228d99e3977ac27cb79d5963adda262af99ad",
           "repositoryUrl": "https://github.com/onnx/onnx.git"
         },
         "comments": "git submodule at cmake/external/onnx"
@@ -216,7 +216,7 @@
       "component": {
         "type": "git",
         "git": {
-          "commitHash": "eb43908b02a296ea0594432f06e9d3fac288d672",
+          "commitHash": "06adf4461ac84035bee658c6cf5df39f7ab6071d",
           "repositoryUrl": "https://github.com/onnx/onnx-tensorrt.git"
         },
         "comments": "onnx_tensorrt"

cmake/CMakeLists.txt

@@ -59,8 +59,8 @@ if (NOT CMAKE_BUILD_TYPE)
   set(CMAKE_BUILD_TYPE "RelWithDebInfo" CACHE STRING "Choose build type: Debug Release RelWithDebInfo MinSizeRel." FORCE)
 endif()
 
-if("${CMAKE_C_COMPILER_ID}" STREQUAL "GNU" AND CMAKE_C_COMPILER_VERSION VERSION_LESS 8)
-  message(FATAL_ERROR  "GCC version must be greater than or equal to 8")
+if("${CMAKE_C_COMPILER_ID}" STREQUAL "GNU" AND CMAKE_C_COMPILER_VERSION VERSION_LESS 9)
+  message(FATAL_ERROR  "GCC version must be greater than or equal to 9")
 endif()
 
 # Options
@@ -1300,12 +1300,6 @@ if (onnxruntime_USE_TVM)
   list(APPEND onnxruntime_EXTERNAL_DEPENDENCIES tvm)
 endif()
 
-# needs to link with stdc++fs in Linux
-if (UNIX AND "${CMAKE_C_COMPILER_ID}" STREQUAL "GNU" AND CMAKE_C_COMPILER_VERSION VERSION_LESS 9)
-  set(FS_STDLIB stdc++fs)
-endif()
-list(APPEND onnxruntime_EXTERNAL_LIBRARIES ${FS_STDLIB})
-
 # onnxruntime-extensions
 if (onnxruntime_USE_EXTENSIONS)
   include(extensions)
@@ -1474,16 +1468,6 @@ if (onnxruntime_USE_CUDA)
   endif()
 endif()
 
-if (onnxruntime_USE_TENSORRT)
-  # needs to link with stdc++fs in Linux
-  if (UNIX)
-    if (NOT APPLE)
-      set(FS_STDLIB stdc++fs)
-    endif()
-  endif()
-  list(APPEND onnxruntime_EXTERNAL_LIBRARIES ${FS_STDLIB})
-endif()
-
 if (onnxruntime_USE_MIGRAPHX)
   if (WIN32)
     message(FATAL_ERROR "MIGraphX does not support build in Windows!")

cmake/deps.txt

@@ -36,7 +36,7 @@ microsoft_wil;https://github.com/microsoft/wil/archive/refs/tags/v1.0.230629.1.z
 mimalloc;https://github.com/microsoft/mimalloc/archive/refs/tags/v2.1.1.zip;d5ee7d34223d0567892db5179849939c8769dc41
 mp11;https://github.com/boostorg/mp11/archive/refs/tags/boost-1.82.0.zip;9bc9e01dffb64d9e0773b2e44d2f22c51aace063
 neural_speed;https://github.com/intel/neural-speed/archive/refs/tags/v0.3.zip;5ec64e3071edc7347ebd8a81679cf06e2bb9b851
-onnx;https://github.com/onnx/onnx/archive/refs/tags/v1.16.0.zip;a6d8b619459fb4657f8bec7d1c6d95ad6d4c069d
+onnx;https://github.com/onnx/onnx/archive/refs/tags/v1.16.1.zip;2eb9198bb352757d5ff13977cbe0634898e0837c
 #use the latest commit of 10.0-GA
 onnx_tensorrt;https://github.com/onnx/onnx-tensorrt/archive/06adf4461ac84035bee658c6cf5df39f7ab6071d.zip;46dceef659d75d276e7914a8057c2282269d5e7b
 protobuf;https://github.com/protocolbuffers/protobuf/archive/refs/tags/v21.12.zip;7cf2733949036c7d52fda017badcab093fe73bfa

cmake/onnxruntime_providers_migraphx.cmake

@@ -49,7 +49,7 @@
   target_compile_options(onnxruntime_providers_migraphx PRIVATE -Wno-error=sign-compare)
   set_property(TARGET onnxruntime_providers_migraphx APPEND_STRING PROPERTY COMPILE_FLAGS "-Wno-deprecated-declarations")
   set_property(TARGET onnxruntime_providers_migraphx APPEND_STRING PROPERTY LINK_FLAGS "-Xlinker --version-script=${ONNXRUNTIME_ROOT}/core/providers/migraphx/version_script.lds -Xlinker --gc-sections")
-  target_link_libraries(onnxruntime_providers_migraphx PRIVATE nsync::nsync_cpp stdc++fs)
+  target_link_libraries(onnxruntime_providers_migraphx PRIVATE nsync::nsync_cpp)
 
   include(CheckLibraryExists)
   check_library_exists(migraphx::c "migraphx_program_run_async" "/opt/rocm/migraphx/lib" HAS_STREAM_SYNC)

cmake/onnxruntime_providers_tensorrt.cmake

@@ -206,7 +206,7 @@
   elseif(UNIX)
     set_property(TARGET onnxruntime_providers_tensorrt APPEND_STRING PROPERTY COMPILE_FLAGS "-Wno-deprecated-declarations")
     set_property(TARGET onnxruntime_providers_tensorrt APPEND_STRING PROPERTY LINK_FLAGS "-Xlinker --version-script=${ONNXRUNTIME_ROOT}/core/providers/tensorrt/version_script.lds -Xlinker --gc-sections")
-    target_link_libraries(onnxruntime_providers_tensorrt PRIVATE nsync::nsync_cpp stdc++fs)
+    target_link_libraries(onnxruntime_providers_tensorrt PRIVATE nsync::nsync_cpp)
   elseif(WIN32)
     set_property(TARGET onnxruntime_providers_tensorrt APPEND_STRING PROPERTY LINK_FLAGS "-DEF:${ONNXRUNTIME_ROOT}/core/providers/tensorrt/symbols.def")
   else()

cmake/patches/onnx/onnx.patch

@@ -36,15 +36,15 @@ index b847798e..a6c31904 100644
 --- a/onnx/common/file_utils.h
 +++ b/onnx/common/file_utils.h
 @@ -6,7 +6,6 @@
- 
+
  #pragma once
- 
+
 -#include <filesystem>
  #include <fstream>
  #include <string>
- 
+
 @@ -17,8 +16,7 @@ namespace ONNX_NAMESPACE {
- 
+
  template <typename T>
  void LoadProtoFromPath(const std::string proto_path, T& proto) {
 -  std::filesystem::path proto_u8_path = std::filesystem::u8path(proto_path);
@@ -53,50 +53,14 @@ index b847798e..a6c31904 100644
    if (!proto_stream.good()) {
      fail_check("Unable to open proto file: ", proto_path, ". Please check if it is a valid proto. ");
    }
-diff --git a/onnx/defs/quantization/defs.cc b/onnx/defs/quantization/defs.cc
-index 70b4a4db..98c11545 100644
---- a/onnx/defs/quantization/defs.cc
-+++ b/onnx/defs/quantization/defs.cc
-@@ -200,6 +200,9 @@ ONNX_OPERATOR_SET_SCHEMA(
-         .SetDoc(DequantizeLinear_ver21_doc)
-         .TypeAndShapeInferenceFunction([](ONNX_NAMESPACE::InferenceContext& ctx) {
-           propagateElemTypeFromInputToOutput(ctx, 1, 0);
-+          if (!hasInputShape(ctx, 0)) {
-+            return;
-+          }
-           auto& input_shape = getInputShape(ctx, 0);
-           updateOutputShape(ctx, 0, input_shape);
-         }));
-diff --git a/onnx/defs/quantization/old.cc b/onnx/defs/quantization/old.cc
-index 3f2d6384..d2f7cfd8 100644
---- a/onnx/defs/quantization/old.cc
-+++ b/onnx/defs/quantization/old.cc
-@@ -130,6 +130,9 @@ ONNX_OPERATOR_SET_SCHEMA(
-         .SetDoc(DequantizeLinear_ver19_doc)
-         .TypeAndShapeInferenceFunction([](ONNX_NAMESPACE::InferenceContext& ctx) {
-           propagateElemTypeFromInputToOutput(ctx, 1, 0);
-+          if (!hasInputShape(ctx, 0)) {
-+            return;
-+          }
-           auto& input_shape = getInputShape(ctx, 0);
-           updateOutputShape(ctx, 0, input_shape);
-         }));
-@@ -181,7 +184,6 @@ ONNX_OPERATOR_SET_SCHEMA(
-           if (!hasInputShape(ctx, 0)) {
-             return;
-           }
--
-           auto& input_shape = getInputShape(ctx, 0);
-           updateOutputShape(ctx, 0, input_shape);
-         }));
 diff --git a/onnx/onnx_pb.h b/onnx/onnx_pb.h
 index 0aab3e26..398ac2d6 100644
 --- a/onnx/onnx_pb.h
 +++ b/onnx/onnx_pb.h
 @@ -47,10 +47,28 @@
  #define ONNX_API ONNX_IMPORT
  #endif
- 
+
 +#if defined(__GNUC__)
 +#pragma GCC diagnostic push
 +
@@ -116,61 +80,9 @@ index 0aab3e26..398ac2d6 100644
  #else
  #include "onnx/onnx.pb.h"
  #endif
- 
+
 +#if defined(__GNUC__)
 +#pragma GCC diagnostic pop
 +#endif
 +
  #endif // ! ONNX_ONNX_PB_H
-diff --git a/onnx/shape_inference/implementation.cc b/onnx/shape_inference/implementation.cc
-index fab1faf2..8723dcd4 100644
---- a/onnx/shape_inference/implementation.cc
-+++ b/onnx/shape_inference/implementation.cc
-@@ -488,29 +488,29 @@ class ShapeInferenceImplBase {
-           ProcessCall(n, *(iter->second), ctx);
-         } else {
-           has_unsupported_op = true;
-+          return;
-         }
-       } else {
-         has_unsupported_op = true;
-+        return;
-       }
--      if (!has_unsupported_op) {
--        for (int i = 0; i < n.output_size(); ++i) {
--          // skip type and shape propagation for missing optional outputs.
--          if (!n.output(i).empty())
--            UpdateType(n.output(i), ctx.getOutputType(i));
--        }
--        // Constant values are tracked to improve inference/checking for subsequent nodes.
--        ProcessConstant(n);
--        // If data-propagation is enabled, partial-evaluation (aka data-propagation) is performed
--        // to improve inference/checking for subsequent nodes.
--        if (options.enable_data_propagation && schema && schema->has_data_propagation_function()) {
--          if (generated_shape_data_by_name == nullptr) {
--            fail_shape_inference(
--                "Container for generated shape data cannot be nullptr when enable_data_propagation option is set.");
--          }
--          DataPropagationContextImpl data_propagation_ctx(
--              n, value_types_by_name, input_data_by_name, *generated_shape_data_by_name);
--          schema->GetDataPropagationFunction()(data_propagation_ctx);
-+      for (int i = 0; i < n.output_size(); ++i) {
-+        // skip type and shape propagation for missing optional outputs.
-+        if (!n.output(i).empty())
-+          UpdateType(n.output(i), ctx.getOutputType(i));
-+      }
-+      // Constant values are tracked to improve inference/checking for subsequent nodes.
-+      ProcessConstant(n);
-+      // If data-propagation is enabled, partial-evaluation (aka data-propagation) is performed
-+      // to improve inference/checking for subsequent nodes.
-+      if (options.enable_data_propagation && schema && schema->has_data_propagation_function()) {
-+        if (generated_shape_data_by_name == nullptr) {
-+          fail_shape_inference(
-+              "Container for generated shape data cannot be nullptr when enable_data_propagation option is set.");
-         }
-+        DataPropagationContextImpl data_propagation_ctx(
-+            n, value_types_by_name, input_data_by_name, *generated_shape_data_by_name);
-+        schema->GetDataPropagationFunction()(data_propagation_ctx);
-       }
-     }
-     ONNX_CATCH(const ONNX_NAMESPACE::InferenceError& ex) {

docs/OperatorKernels.md

@@ -421,7 +421,7 @@ Do not modify directly.*
 |Transpose|*in* data:**T**<br> *out* transposed:**T**|21+|**T** = tensor(bfloat16), tensor(bool), tensor(double), tensor(float), tensor(float16), tensor(float8e4m3fn), tensor(float8e4m3fnuz), tensor(float8e5m2), tensor(float8e5m2fnuz), tensor(int16), tensor(int32), tensor(int4), tensor(int64), tensor(int8), tensor(string), tensor(uint16), tensor(uint32), tensor(uint4), tensor(uint64), tensor(uint8)|
 |||[13, 20]|**T** = 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)|
 |||[1, 12]|**T** = 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)|
-|Trilu|*in* input:**T**<br> *in* k:**tensor(int64)**<br> *out* output:**T**|14+|**T** = tensor(double), tensor(float), tensor(int64)|
+|Trilu|*in* input:**T**<br> *in* k:**tensor(int64)**<br> *out* output:**T**|14+|**T** = tensor(bool), tensor(double), tensor(float), tensor(int64)|
 |Unique|*in* X:**T**<br> *out* Y:**T**<br> *out* indices:**tensor(int64)**<br> *out* inverse_indices:**tensor(int64)**<br> *out* counts:**tensor(int64)**|11+|**T** = tensor(double), tensor(float), tensor(int64), tensor(int8), tensor(string)|
 |Unsqueeze|*in* data:**T**<br> *in* axes:**tensor(int64)**<br> *out* expanded:**T**<br><br>or<br><br>*in* data:**T**<br> *out* expanded:**T**|21+|**T** = 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)|
 |||[13, 20]|**T** = 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)|