[ORT 1.17.0 Release] Cherry pick 1st round

cgmanifests/generated/cgmanifest.json

@@ -36,7 +36,7 @@
       "component": {
         "type": "git",
         "git": {
-          "commitHash": "dcd5bd5fd593e31465af3d9ef291d26c646b0a4f",
+          "commitHash": "4a2c63365eff8823a5221db86ef490e828306f9d",
           "repositoryUrl": "https://github.com/abseil/abseil-cpp.git"
         },
         "comments": "abseil_cpp"
@@ -192,6 +192,16 @@
         "comments": "mp11"
       }
     },
+    {
+      "component": {
+        "type": "git",
+        "git": {
+          "commitHash": "c11386eb632eec7c1c2aa323142f73519f946e2a",
+          "repositoryUrl": "https://github.com/intel/neural-speed.git"
+        },
+        "comments": "neural_speed"
+      }
+    },
     {
       "component": {
         "type": "git",

cmake/CMakeLists.txt

@@ -87,7 +87,7 @@ option(onnxruntime_USE_QNN "Build with QNN support" OFF)
 option(onnxruntime_USE_SNPE "Build with SNPE support" OFF)
 option(onnxruntime_USE_RKNPU "Build with RKNPU support" OFF)
 option(onnxruntime_USE_DNNL "Build with DNNL support" OFF)
-option(onnxruntime_USE_JBLAS "Build MLAS with JBLAS support" ON)
+option(onnxruntime_USE_NEURAL_SPEED "Build with Neural Speed support" ON)
 option(onnxruntime_USE_JSEP "Build with JavaScript implemented kernels support" OFF)
 option(onnxruntime_BUILD_UNIT_TESTS "Build ONNXRuntime unit tests" ON)
 option(onnxruntime_BUILD_CSHARP "Build C# library" OFF)
@@ -96,7 +96,6 @@ option(onnxruntime_USE_PREINSTALLED_EIGEN "Use pre-installed EIGEN. Need to prov
 option(onnxruntime_BUILD_BENCHMARKS "Build ONNXRuntime micro-benchmarks" OFF)
 option(onnxruntime_USE_LLVM "Build TVM with LLVM" OFF)
 
-cmake_dependent_option(onnxruntime_USE_CUTLASS "Build with cutlass support" ON "onnxruntime_USE_CUDA" OFF)
 cmake_dependent_option(onnxruntime_USE_FLASH_ATTENTION "Build flash attention kernel for scaled dot product attention" ON "NOT WIN32; onnxruntime_USE_CUDA" OFF)
 option(onnxruntime_USE_MEMORY_EFFICIENT_ATTENTION "Build memory efficient attention kernel for scaled dot product attention" ON)
 
@@ -706,20 +705,16 @@ if (onnxruntime_USE_CUDA)
   enable_language(CUDA)
   message( STATUS "CMAKE_CUDA_COMPILER_VERSION: ${CMAKE_CUDA_COMPILER_VERSION}")
 
+  if (onnxruntime_DISABLE_CONTRIB_OPS)
+    set(onnxruntime_USE_FLASH_ATTENTION OFF)
+    set(onnxruntime_USE_MEMORY_EFFICIENT_ATTENTION OFF)
+  endif()
   if (CMAKE_CUDA_COMPILER_VERSION VERSION_LESS 11.6)
-    message( STATUS "Turn off cutlass since CUDA compiler version < 11.6")
-    set(onnxruntime_USE_CUTLASS OFF)
+    message( STATUS "Turn off flash attention since CUDA compiler version < 11.6")
+    set(onnxruntime_USE_FLASH_ATTENTION OFF)
+    set(onnxruntime_USE_MEMORY_EFFICIENT_ATTENTION OFF)
   endif()
 else()
-  set(onnxruntime_USE_CUTLASS OFF)
-endif()
-
-if (NOT onnxruntime_USE_CUTLASS OR onnxruntime_DISABLE_CONTRIB_OPS)
-    if (onnxruntime_DISABLE_CONTRIB_OPS)
-      message( STATUS "Turn off flash attention/memory efficient attention since contrib ops are disabled")
-    else()
-      message( STATUS "Turn off flash attention/memory efficient attention since cutlass is not enabled")
-    endif()
   set(onnxruntime_USE_FLASH_ATTENTION OFF)
   set(onnxruntime_USE_MEMORY_EFFICIENT_ATTENTION OFF)
 endif()
@@ -905,8 +900,8 @@ function(onnxruntime_set_compile_flags target_name)
       target_compile_definitions(${target_name} PRIVATE ENABLE_ATEN)
     endif()
 
-    if (onnxruntime_USE_CUTLASS)
-      target_compile_definitions(${target_name} PRIVATE USE_CUTLASS)
+    if(USE_NEURAL_SPEED)
+      target_compile_definitions(${target_name} PRIVATE ORT_NEURAL_SPEED)
     endif()
 
     set_target_properties(${target_name} PROPERTIES COMPILE_WARNING_AS_ERROR ON)
@@ -1193,14 +1188,10 @@ if (onnxruntime_USE_DNNL)
   add_compile_definitions(DNNL_OPENMP)
 endif()
 
-set(USE_JBLAS FALSE)
-if (onnxruntime_USE_JBLAS AND NOT onnxruntime_MINIMAL_BUILD)
-  if ("${CMAKE_CXX_COMPILER_ID}" STREQUAL "GNU" AND onnxruntime_target_platform STREQUAL "x86_64")
-    add_compile_definitions(MLAS_JBLAS)
-    set(USE_JBLAS TRUE)
-  elseif ("${CMAKE_CXX_COMPILER_ID}" STREQUAL "MSVC" AND onnxruntime_target_platform STREQUAL "x64")
-    add_compile_definitions(MLAS_JBLAS)
-    set(USE_JBLAS TRUE)
+if (onnxruntime_USE_NEURAL_SPEED AND NOT onnxruntime_MINIMAL_BUILD)
+  include(neural_speed)
+  if (USE_NEURAL_SPEED)
+    list(APPEND onnxruntime_EXTERNAL_LIBRARIES neural_speed::bestla)
   endif()
 endif()
 

cmake/deps.txt

@@ -12,7 +12,7 @@
 # NOTE: You must run deps_update_and_upload.py and generate_cgmanifest.py when ready to test your changes in a CI.
 # See https://microsoft.sharepoint.com/teams/ONNX2/_layouts/OneNote.aspx?id=%2Fteams%2FONNX2%2FShared%20Documents%2FNotebooks%2FONNX%20Ecosystem%20Team%20Notebook&wd=target%28Development.one%7C63D3AB47-51D1-4A62-9965-66882234BD44%2FAdd%20or%20update%20a%20dependency%20in%20deps.txt%7C0E9ED71D-89D5-40FA-B05F-C0123289C591%2F%29
 #
-abseil_cpp;https://github.com/abseil/abseil-cpp/archive/dcd5bd5fd593e31465af3d9ef291d26c646b0a4f.zip;6cc204586014e189f5c0fe3274f83162fa7c700c
+abseil_cpp;https://github.com/abseil/abseil-cpp/archive/refs/tags/20240116.0.zip;bc2cec6baaad67fcb6c0c38972b687d4797927e9
 cxxopts;https://github.com/jarro2783/cxxopts/archive/3c73d91c0b04e2b59462f0a741be8c07024c1bc0.zip;6c6ca7f8480b26c8d00476e0e24b7184717fe4f0
 date;https://github.com/HowardHinnant/date/archive/refs/tags/v3.0.1.zip;2dac0c81dc54ebdd8f8d073a75c053b04b56e159
 dlpack;https://github.com/dmlc/dlpack/archive/refs/tags/v0.6.zip;4d565dd2e5b31321e5549591d78aa7f377173445
@@ -34,6 +34,7 @@ microsoft_gsl;https://github.com/microsoft/GSL/archive/refs/tags/v4.0.0.zip;cf36
 microsoft_wil;https://github.com/microsoft/wil/archive/refs/tags/v1.0.230629.1.zip;e4a542a323c070376f7c2d1973d0f7ddbc1d2fa5
 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/bestlav0.1.1.zip;65b0f7a0d04f72f0d5a8d48af70f0366f2ab3939
 onnx;https://github.com/onnx/onnx/archive/refs/tags/v1.15.0.zip;54c3f960a0541c5d8d3e60c2933e11f5d3688a11
 #use the commit of supporting all the plugins and TRT 8.6-GA (https://github.com/onnx/onnx-tensorrt/commit/0462dc31ae78f48744b6141ae376df1f96d3f459)
 onnx_tensorrt;https://github.com/onnx/onnx-tensorrt/archive/a43ce67187bab219520fd80f21af8bbd4354bc8c.zip;572535aefef477050f86744dfab1fef840198035
@@ -54,4 +55,4 @@ tensorboard;https://github.com/tensorflow/tensorboard/archive/373eb09e4c5d2b3cc2
 cutlass;https://github.com/NVIDIA/cutlass/archive/refs/tags/v3.1.0.zip;757f90a795034a89d4f48a79d1f009f7a04c8dee
 utf8_range;https://github.com/protocolbuffers/utf8_range/archive/72c943dea2b9240cd09efde15191e144bc7c7d38.zip;9925739c9debc0efa2adcb194d371a35b6a03156
 extensions;https://github.com/microsoft/onnxruntime-extensions/archive/94142d8391c9791ec71c38336436319a2d4ac7a0.zip;4365ac5140338b4cb75a39944a4be276e3829b3c
-composable_kernel;https://github.com/ROCmSoftwarePlatform/composable_kernel/archive/5356c4a943a35e74d7cdc69486afcb8703b9a59a.zip;522382c2af437e09124287e5879ab64af5b2e299
+composable_kernel;https://github.com/ROCmSoftwarePlatform/composable_kernel/archive/5356c4a943a35e74d7cdc69486afcb8703b9a59a.zip;522382c2af437e09124287e5879ab64af5b2e299

cmake/external/abseil-cpp.cmake

@@ -19,7 +19,7 @@ if(WIN32 AND NOT Patch_FOUND)
   set(ABSL_ENABLE_INSTALL ON)
 endif()
 # NB! Advancing Abseil version changes its internal namespace,
-# currently absl::lts_20230125 which affects abseil-cpp.natvis debugger
+# currently absl::lts_20240116 which affects abseil-cpp.natvis debugger
 # visualization file, that must be adjusted accordingly, unless we eliminate
 # that namespace at build time.
 FetchContent_Declare(

cmake/external/abseil-cpp.natvis

@@ -1,6 +1,6 @@
 <?xml version="1.0" encoding="utf-8"?>
 <AutoVisualizer xmlns="http://schemas.microsoft.com/vstudio/debugger/natvis/2010">
-  <Type Name="absl::lts_20230802::InlinedVector&lt;*&gt;">
+  <Type Name="absl::lts_20240116::InlinedVector&lt;*&gt;">
     <Intrinsic Name="_size" Expression="storage_.metadata_.value >> 1"/>
     <Intrinsic Name="_is_allocated" Expression="(storage_.metadata_.value &amp; 1) == 1"/>
     <Intrinsic Name="_inlined_data" Expression="($T1*)storage_.data_.inlined.inlined_data"/>
@@ -24,7 +24,7 @@
     </Expand>
   </Type>
   <!-- Should handle both flat hash_set and hash_map -->
-  <Type Name="absl::lts_20230802::container_internal::raw_hash_set&lt;*&gt;">
+  <Type Name="absl::lts_20240116::container_internal::raw_hash_set&lt;*&gt;">
     <Intrinsic Name="_commonfields" Expression="settings_.value"/>
     <Intrinsic Name="_size" Expression="settings_.value.compressed_tuple_.value"/>
     <Intrinsic Name="_capacity" Expression="_commonfields().capacity_"/>
@@ -51,7 +51,7 @@
   </Type>
 
   <!-- Primitive types stored as a value -->
-  <Type Name="absl::lts_20230802::container_internal::Storage&lt;*,*,0&gt;">
+  <Type Name="absl::lts_20240116::container_internal::Storage&lt;*,*,0&gt;">
     <DisplayString IncludeView="noparens">*($T1 *){value}</DisplayString>
     <DisplayString ExcludeView="noparens">(*($T1 *){value})</DisplayString>
     <Expand>
@@ -60,15 +60,15 @@
   </Type>
 
   <!-- For storage inherited from the type -->
-  <Type Name="absl::lts_20230802::container_internal::Storage&lt;*,*,1&gt;">
+  <Type Name="absl::lts_20240116::container_internal::Storage&lt;*,*,1&gt;">
     <DisplayString IncludeView="noparens">*($T1 *)this</DisplayString>
     <DisplayString ExcludeView="noparens">(*($T1 *)this)</DisplayString>
     <Expand>
       <ExpandedItem>*($T1 *)this</ExpandedItem>
     </Expand>
   </Type>
 
-  <Type Name="absl::lts_20230802::container_internal::map_slot_type&lt;*&gt;">
+  <Type Name="absl::lts_20240116::container_internal::map_slot_type&lt;*&gt;">
     <DisplayString IncludeView="noparens">{value.first}, {value.second}</DisplayString>
     <DisplayString ExcludeView="noparens">({value.first}, {value.second})</DisplayString>
     <Expand>

cmake/external/cutlass.cmake

@@ -1,13 +1,11 @@
-if (onnxruntime_USE_CUTLASS)
-  include(FetchContent)
-  FetchContent_Declare(
-    cutlass
-    URL ${DEP_URL_cutlass}
-    URL_HASH SHA1=${DEP_SHA1_cutlass}
-  )
+include(FetchContent)
+FetchContent_Declare(
+  cutlass
+  URL ${DEP_URL_cutlass}
+  URL_HASH SHA1=${DEP_SHA1_cutlass}
+)
 
-  FetchContent_GetProperties(cutlass)
-  if(NOT cutlass_POPULATED)
-    FetchContent_Populate(cutlass)
-  endif()
+FetchContent_GetProperties(cutlass)
+if(NOT cutlass_POPULATED)
+  FetchContent_Populate(cutlass)
 endif()

cmake/external/neural_speed.cmake

@@ -0,0 +1,15 @@
+if ("${CMAKE_CXX_COMPILER_ID}" STREQUAL "GNU" AND onnxruntime_target_platform STREQUAL "x86_64")
+  set(USE_NEURAL_SPEED TRUE)
+elseif ("${CMAKE_CXX_COMPILER_ID}" STREQUAL "MSVC" AND onnxruntime_target_platform STREQUAL "x64")
+  set(USE_NEURAL_SPEED TRUE)
+endif()
+
+if(USE_NEURAL_SPEED)
+  FetchContent_Declare(
+      neural_speed
+      URL ${DEP_URL_neural_speed}
+      URL_HASH SHA1=${DEP_SHA1_neural_speed}
+  )
+  set(BTLA_USE_OPENMP OFF)
+  onnxruntime_fetchcontent_makeavailable(neural_speed)
+endif()

cmake/onnxruntime_mlas.cmake

@@ -57,15 +57,6 @@ endif()
 
 set(ONNXRUNTIME_MLAS_LIBS onnxruntime_mlas)
 
-function(add_jblas)
-    add_subdirectory(${MLAS_SRC_DIR}/x86_64/jblas jblas)
-    target_link_libraries(onnxruntime_mlas PRIVATE jblas::jblas)
-    target_sources(onnxruntime_mlas PRIVATE
-        ${MLAS_SRC_DIR}/jblas_gemm.cpp
-     )
-    set_target_properties(${target_name} PROPERTIES COMPILE_WARNING_AS_ERROR OFF)
-endfunction()
-
 #TODO: set MASM flags properly
 function(setup_mlas_source_for_windows)
 
@@ -364,19 +355,23 @@ else()
             ${MLAS_SRC_DIR}/aarch64/HalfGemmKernelNeon.S
             ${MLAS_SRC_DIR}/aarch64/QgemmS8S8KernelSmmla.S
             ${MLAS_SRC_DIR}/aarch64/QgemmU8X8KernelUmmla.S
+            ${MLAS_SRC_DIR}/aarch64/SbgemmKernelNeon.S
             ${MLAS_SRC_DIR}/activate_fp16.cpp
             ${MLAS_SRC_DIR}/dwconv.cpp
             ${MLAS_SRC_DIR}/halfgemm_kernel_neon.cpp
             ${MLAS_SRC_DIR}/pooling_fp16.cpp
             ${MLAS_SRC_DIR}/qgemm_kernel_smmla.cpp
             ${MLAS_SRC_DIR}/qgemm_kernel_ummla.cpp
+            ${MLAS_SRC_DIR}/sbgemm_kernel_neon.cpp
           )
           set_source_files_properties(${MLAS_SRC_DIR}/aarch64/HalfGemmKernelNeon.S PROPERTIES COMPILE_FLAGS " -march=armv8.2-a+fp16 ")
           set_source_files_properties(${MLAS_SRC_DIR}/aarch64/QgemmS8S8KernelSmmla.S PROPERTIES COMPILE_FLAGS " -march=armv8.2-a+i8mm ")
           set_source_files_properties(${MLAS_SRC_DIR}/aarch64/QgemmU8X8KernelUmmla.S PROPERTIES COMPILE_FLAGS " -march=armv8.2-a+i8mm ")
+          set_source_files_properties(${MLAS_SRC_DIR}/aarch64/SbgemmKernelNeon.S PROPERTIES COMPILE_FLAGS " -march=armv8.2-a+bf16 ")
           set_source_files_properties(${MLAS_SRC_DIR}/activate_fp16.cpp PROPERTIES COMPILE_FLAGS " -march=armv8.2-a+fp16 ")
           set_source_files_properties(${MLAS_SRC_DIR}/dwconv.cpp PROPERTIES COMPILE_FLAGS " -march=armv8.2-a+fp16 ")
           set_source_files_properties(${MLAS_SRC_DIR}/pooling_fp16.cpp PROPERTIES COMPILE_FLAGS " -march=armv8.2-a+fp16 ")
+          set_source_files_properties(${MLAS_SRC_DIR}/sbgemm_kernel_neon.cpp PROPERTIES COMPILE_FLAGS " -march=armv8.2-a+bf16 ")
         endif()
 
         if(ONNXRUNTIME_MLAS_MULTI_ARCH)
@@ -622,10 +617,6 @@ else()
     target_sources(onnxruntime_mlas PRIVATE ${mlas_platform_srcs})
 endif()
 
-if(USE_JBLAS)
-  add_jblas()
-endif()
-
 foreach(mlas_target ${ONNXRUNTIME_MLAS_LIBS})
     target_include_directories(${mlas_target} PRIVATE ${MLAS_INC_DIR} ${MLAS_SRC_DIR})
     onnxruntime_add_include_to_target(${mlas_target} ${GSL_TARGET})

cmake/onnxruntime_providers_cpu.cmake

@@ -60,6 +60,15 @@ if(NOT onnxruntime_DISABLE_CONTRIB_OPS)
       "${ONNXRUNTIME_ROOT}/contrib_ops/cpu/aten_ops/aten_op_executor.cc"
     )
   endif()
+  set(onnxruntime_cpu_neural_speed_srcs 
+    "${ONNXRUNTIME_ROOT}/contrib_ops/cpu/quantization/neural_speed_wrapper.h"
+    "${ONNXRUNTIME_ROOT}/contrib_ops/cpu/quantization/neural_speed_defs.h"
+    "${ONNXRUNTIME_ROOT}/contrib_ops/cpu/quantization/neural_speed_gemm.cc"
+    "${ONNXRUNTIME_ROOT}/contrib_ops/cpu/quantization/neural_speed_gemm.h"
+  )
+  if(NOT USE_NEURAL_SPEED)
+    list(REMOVE_ITEM onnxruntime_cpu_contrib_ops_srcs ${onnxruntime_cpu_neural_speed_srcs})
+  endif()
   # add using ONNXRUNTIME_ROOT so they show up under the 'contrib_ops' folder in Visual Studio
   source_group(TREE ${ONNXRUNTIME_ROOT} FILES ${onnxruntime_cpu_contrib_ops_srcs})
   list(APPEND onnxruntime_providers_src ${onnxruntime_cpu_contrib_ops_srcs})
@@ -144,6 +153,12 @@ if (HAS_BITWISE_INSTEAD_OF_LOGICAL)
   target_compile_options(onnxruntime_providers PRIVATE "-Wno-bitwise-instead-of-logical")
 endif()
 
+if(NOT onnxruntime_DISABLE_CONTRIB_OPS)
+  if(USE_NEURAL_SPEED)
+    onnxruntime_add_include_to_target(onnxruntime_providers neural_speed::bestla)
+  endif()
+endif()
+
 if (MSVC)
    target_compile_options(onnxruntime_providers PRIVATE "/bigobj")
 #   if(NOT CMAKE_SIZEOF_VOID_P EQUAL 8)

docs/ContribOperators.md

@@ -3031,6 +3031,8 @@ This version of the operator has been available since version 1 of the 'com.micr
 <dd>Number of attention heads</dd>
 <dt><tt>scale</tt> : float</dt>
 <dd>Custom scale will be used if specified. Default value is 1/sqrt(head_size)</dd>
+<dt><tt>unidirectional</tt> : int</dt>
+<dd>Whether every token can only attend to previous tokens. Default value is 0.</dd>
 </dl>
 
 #### Inputs (1 - 8)
@@ -5021,6 +5023,10 @@ This version of the operator has been available since version 1 of the 'com.micr
 <dl>
 <dt><tt>interleaved</tt> : int</dt>
 <dd>Rotate using interleaved pattern. Default value is 0 (False).</dd>
+<dt><tt>num_heads</tt> : int</dt>
+<dd>Number of attention heads. Default value is 0. Must use with rotary_embedding_dim</dd>
+<dt><tt>rotary_embedding_dim</tt> : int</dt>
+<dd>Rotary embedding dimension. Default value is 0.</dd>
 <dt><tt>scale</tt> : float</dt>
 <dd>Custom scale will be used if specified. Default value is 1.0</dd>
 </dl>
@@ -5033,9 +5039,9 @@ This version of the operator has been available since version 1 of the 'com.micr
 <dt><tt>position_ids</tt> : M</dt>
 <dd>1D tensor with shape (1) or 2D tensor with shape (batch_size, sequence_length)</dd>
 <dt><tt>cos_cache</tt> : T</dt>
-<dd>2D tensor with shape (max_sequence_length, head_size / 2).</dd>
+<dd>2D tensor with shape (max_sequence_length, head_size / 2) or (max_sequence_length, rotary_embedding_dim / 2)</dd>
 <dt><tt>sin_cache</tt> : T</dt>
-<dd>2D tensor with shape (max_sequence_length, head_size / 2).</dd>
+<dd>2D tensor with shape (max_sequence_length, head_size / 2) or (max_sequence_length, rotary_embedding_dim / 2)</dd>
 </dl>
 
 #### Outputs
@@ -5048,7 +5054,7 @@ This version of the operator has been available since version 1 of the 'com.micr
 #### Type Constraints
 
 <dl>
-<dt><tt>T</tt> : tensor(float), tensor(float16)</dt>
+<dt><tt>T</tt> : tensor(float), tensor(float16), tensor(bfloat16)</dt>
 <dd>Constrain input and output types to float tensors.</dd>
 <dt><tt>M</tt> : tensor(int64)</dt>
 <dd>Constrain input and output types to integer tensors</dd>

docs/OperatorKernels.md

@@ -868,7 +868,7 @@ Do not modify directly.*
 |RemovePadding|*in* input:**T**<br> *in* sequence_token_count:**M**<br> *out* output:**T**<br> *out* token_offset:**M**<br> *out* cumulated_seq_len:**M**<br> *out* max_seq_len:**M**|1+|**T** = tensor(float), tensor(float16)|
 |RestorePadding|*in* input:**T**<br> *in* token_offset:**M**<br> *out* output:**T**|1+|**T** = tensor(float), tensor(float16)|
 |Rfft|*in* X:**T**<br> *out* Y:**T**|1+|**T** = tensor(double), tensor(float), tensor(float16)|
-|RotaryEmbedding|*in* input:**T**<br> *in* position_ids:**M**<br> *in* cos_cache:**T**<br> *in* sin_cache:**T**<br> *out* output:**T**|1+|**M** = tensor(int64)<br/> **T** = tensor(float), tensor(float16)|
+|RotaryEmbedding|*in* input:**T**<br> *in* position_ids:**M**<br> *in* cos_cache:**T**<br> *in* sin_cache:**T**<br> *out* output:**T**|1+|**M** = tensor(int64)<br/> **T** = tensor(bfloat16), tensor(float), tensor(float16)|
 |Sampling|*in* input_ids:**I**<br> *in* max_length:**I**<br> *in* min_length:**I**<br> *in* repetition_penalty:**T**<br> *in* vocab_mask:**I**<br> *in* prefix_vocab_mask:**I**<br> *in* attention_mask:**I**<br> *in* presence_mask:**I**<br> *in* seed:**I**<br> *out* sequences:**I**<br> *out* filtered_logits:**T**|1+|**T** = tensor(float), tensor(float16)|
 |SkipGroupNorm|*in* X:**T**<br> *in* gamma:**M**<br> *in* beta:**M**<br> *in* skip:**T**<br> *in* bias:**T**<br> *out* Y:**T**<br> *out* S:**T**|1+|**T** = tensor(float), tensor(float16)|
 |SkipLayerNormalization|*in* input:**T**<br> *in* skip:**T**<br> *in* gamma:**T**<br> *in* beta:**T**<br> *in* bias:**T**<br> *out* output:**T**<br> *out* mean:**U**<br> *out* inv_std_var:**U**<br> *out* input_skip_bias_sum:**T**|1+|**T** = tensor(float), tensor(float16)|

include/onnxruntime/core/framework/execution_provider.h

@@ -326,6 +326,15 @@ class IExecutionProvider {
    */
   virtual std::vector<AllocatorPtr> CreatePreferredAllocators() { return std::vector<AllocatorPtr>(); };
 
+  /**
+   * Get the array of pointers for EPContext nodes
+   * EP needs to implement this if has the requirement to generate the context cache model. Otherwise leave it.
+   * Default return an empty vector if not provided by the Execution Provider
+   */
+  virtual const InlinedVector<const Node*> GetEpContextNodes() const {
+    return InlinedVector<const Node*>();
+  }
+
  private:
   const std::string type_;