[Inference]OpenVINO submodule clear (#70313)

* test delete openvino thirty party * optimize openvino submodule * check
PaddlePaddle · Dec 19, 2024 · 0beb71a · 0beb71a
1 parent b22c396
commit 0beb71a
Show file tree

Hide file tree

Showing 3 changed files with 161 additions and 106 deletions.
diff --git a/cmake/external/openvino.cmake b/cmake/external/openvino.cmake
@@ -115,17 +115,40 @@ ExternalProject_Add(
   UPDATE_COMMAND ""
   #BUILD_ALWAYS        1
   PATCH_COMMAND ${OPENVINO_PATCH_COMMAND}
-  CMAKE_ARGS -DCMAKE_INSTALL_PREFIX=${OPENVINO_INSTALL_DIR}
+  CMAKE_ARGS -DCMAKE_CXX_COMPILER=${CMAKE_CXX_COMPILER}
+             -DCMAKE_C_COMPILER=${CMAKE_C_COMPILER}
+             -DCMAKE_CXX_FLAGS=${ONEDNN_CXXFLAG}
+             -DCMAKE_CXX_FLAGS_DEBUG=${CMAKE_CXX_FLAGS_DEBUG}
+             -DCMAKE_CXX_FLAGS_RELEASE=${ONEDNN_CXXFLAG_RELEASE}
+             -DCMAKE_C_FLAGS=${ONEDNN_CFLAG}
+             -DCMAKE_C_FLAGS_DEBUG=${CMAKE_C_FLAGS_DEBUG}
+             -DCMAKE_C_FLAGS_RELEASE=${ONEDNN_CFLAG_RELEASE}
+             -DCMAKE_INSTALL_PREFIX=${OPENVINO_INSTALL_DIR}
              -DCMAKE_BUILD_TYPE=${CMAKE_BUILD_TYPE}
              -DTHREADING=TBB
              -DENABLE_INTEL_CPU=ON
              -DENABLE_INTEL_GPU=OFF
-             -DENABLE_OV_JAX_FRONTEND=OFF
+             -DENABLE_INTEL_NPU=OFF
+             -DENABLE_HETERO=OFF
+             -DENABLE_MULTI=OFF
+             -DENABLE_AUTO=OFF
+             -DENABLE_TEMPLATE=OFF
+             -DENABLE_AUTO_BATCH=OFF
+             -DENABLE_PROXY=OFF
              -DENABLE_OV_ONNX_FRONTEND=OFF
-             -DENABLE_OV_PYTORCH_FRONTEND=OFF
              -DENABLE_OV_TF_FRONTEND=OFF
              -DENABLE_OV_TF_LITE_FRONTEND=OFF
-             -DENABLE_INTEL_NPU=OFF
+             -DENABLE_OV_PYTORCH_FRONTEND=OFF
+             -DENABLE_OV_JAX_FRONTEND=OFF
+             -DENABLE_OV_IR_FRONTEND=OFF
+             -DENABLE_SAMPLES=OFF
+             -DENABLE_TESTS=OFF
+             -DENABLE_PYTHON=OFF
+             -DENABLE_WHEEL=OFF
+             -DENABLE_DOCS=OFF
+             -DENABLE_CPPLINT=OFF
+             -DENABLE_CLANG_FORMAT=OFF
+             -DENABLE_NCC_STYLE=OFF
   CMAKE_CACHE_ARGS -DCMAKE_INSTALL_PREFIX:PATH=${OPENVINO_INSTALL_DIR}
   BUILD_BYPRODUCTS ${BUILD_BYPRODUCTS_ARGS})
 

diff --git a/cmake/third_party.cmake b/cmake/third_party.cmake
@@ -41,6 +41,39 @@ if(NOT WITH_SETUP_INSTALL)
       "Check submodules of paddle, and run 'git submodule sync --recursive && git submodule update --init --recursive'"
   )
 
+  execute_process(
+    COMMAND git submodule update --init third_party/openvino
+    WORKING_DIRECTORY ${PADDLE_SOURCE_DIR}
+    RESULT_VARIABLE result_var)
+  # List of modules to be deleted
+  set(delete_module
+      "thirdparty/zlib/zlib"
+      "thirdparty/gflags/gflags"
+      "thirdparty/gtest/gtest"
+      "thirdparty/ocl/icd_loader"
+      "thirdparty/ocl/cl_headers"
+      "thirdparty/ocl/clhpp_headers"
+      "thirdparty/onnx/onnx"
+      "src/bindings/python/thirdparty/pybind11"
+      "thirdparty/ittapi/ittapi"
+      "cmake/developer_package/ncc_naming_style/ncc"
+      "src/plugins/intel_gpu/thirdparty/onednn_gpu"
+      "thirdparty/open_model_zoo"
+      "thirdparty/json/nlohmann_json"
+      "thirdparty/flatbuffers/flatbuffers"
+      "thirdparty/snappy"
+      "thirdparty/level_zero/level-zero"
+      "src/plugins/intel_npu/thirdparty/level-zero-ext"
+      "src/plugins/intel_npu/thirdparty/yaml-cpp")
+  # Iterate over each module and perform actions
+  foreach(module IN LISTS delete_module)
+    # Remove the module from git cache
+    execute_process(
+      COMMAND git rm --cached ${module}
+      WORKING_DIRECTORY ${PADDLE_SOURCE_DIR}/third_party/openvino
+      RESULT_VARIABLE git_rm_result)
+  endforeach()
+
   # execute_process does not support sequential commands, so we execute echo command separately
   execute_process(
     COMMAND git submodule sync --recursive

diff --git a/patches/openvino/convert.patch b/patches/openvino/convert.patch
@@ -167,6 +167,61 @@ index 0000000000..c51a2af6f9
 +}  // namespace paddle
 +}  // namespace frontend
 +}  // namespace ov
+diff --git a/src/frontends/paddle/src/op/expand_as_v2.cpp b/src/frontends/paddle/src/op/expand_as_v2.cpp
+new file mode 100644
+index 0000000000..19cf05758b
+--- /dev/null
++++ b/src/frontends/paddle/src/op/expand_as_v2.cpp
+@@ -0,0 +1,49 @@
++// Copyright (C) 2018-2024 Intel Corporation
++// SPDX-License-Identifier: Apache-2.0
++//
++
++#include "default_opset.hpp"
++#include "openvino/frontend/paddle/node_context.hpp"
++
++namespace ov {
++namespace frontend {
++namespace paddle {
++namespace op {
++NamedOutputs expand_as_v2(const NodeContext& node) {
++    using namespace default_opset;
++    auto x = node.get_input("X");
++    Output<Node> shape_expected_node;
++    if (node.has_input("Y")) {
++        shape_expected_node = std::make_shared<ShapeOf>(node.get_input("Y"), element::i32);
++    } else {
++        std::vector<int32_t> shape_expected;
++        if (node.has_attribute("target_shape")) {
++            shape_expected = node.get_attribute<std::vector<int32_t>>("target_shape");
++        } else {
++            throw std::runtime_error("expand: has no target_shape attribute");
++        }
++        shape_expected_node = Constant::create(element::i32, {shape_expected.size()}, shape_expected);
++    }
++    // expected shape rank
++    const auto shape_expected_node_rank = std::make_shared<ShapeOf>(shape_expected_node, element::i32);
++    // input shape rank
++    const auto input_shape_node_shape = std::make_shared<ShapeOf>(x, element::i32);
++    const auto input_shape_node_rank = std::make_shared<ShapeOf>(input_shape_node_shape, element::i32);
++    // rank difference
++    const auto rank_diff = std::make_shared<Subtract>(shape_expected_node_rank, input_shape_node_rank);
++    // axis index needed to add
++    const auto rank_idx = std::make_shared<Broadcast>(Constant::create(element::i32, {1}, {1}), rank_diff);
++    // add axis
++    const auto fixed_input_shape_node = std::make_shared<Concat>(NodeVector{rank_idx, input_shape_node_shape}, 0);
++
++    // if -1 in shape we will copy the orginal value from input
++    auto zero_node = Constant::create(ov::element::i32, {1}, {0});
++    auto mask_node = std::make_shared<Greater>(shape_expected_node, zero_node);
++    auto fixed_shape_node = std::make_shared<Select>(mask_node, shape_expected_node, fixed_input_shape_node);
++    return node.default_single_output_mapping({std::make_shared<Broadcast>(x, fixed_shape_node)}, {"Out"});
++}
++
++}  // namespace op
++}  // namespace paddle
++}  // namespace frontend
++}  // namespace ov
 diff --git a/src/frontends/paddle/src/op/expand_v2.cpp b/src/frontends/paddle/src/op/expand_v2.cpp
 index d79e49db28..ea174efa3a 100644
 --- a/src/frontends/paddle/src/op/expand_v2.cpp
@@ -270,6 +325,29 @@ index e7b317f288..5ab551dc3b 100644
  }
 
  NamedOutputs linear_interp_v2(const NodeContext& node) {
+diff --git a/src/frontends/paddle/src/op/less_equal.cpp b/src/frontends/paddle/src/op/less_equal.cpp
+new file mode 100644
+index 0000000000..89c626c820
+--- /dev/null
++++ b/src/frontends/paddle/src/op/less_equal.cpp
+@@ -0,0 +1,17 @@
++// Copyright (C) 2018-2024 Intel Corporation
++// SPDX-License-Identifier: Apache-2.0
++//
++
++#include "elementwise_ops.hpp"
++
++namespace ov {
++namespace frontend {
++namespace paddle {
++namespace op {
++NamedOutputs less_equal(const NodeContext& node) {
++    return elementwise_ops<default_opset::LessEqual>(node);
++}
++}  // namespace op
++}  // namespace paddle
++}  // namespace frontend
++}  // namespace ov
 diff --git a/src/frontends/paddle/src/op/reduce_ops.hpp b/src/frontends/paddle/src/op/reduce_ops.hpp
 index 2b59516042..954d1de425 100644
 --- a/src/frontends/paddle/src/op/reduce_ops.hpp
@@ -402,22 +480,31 @@ index 0000000000..64bc75c66a
 +}  // namespace frontend
 +}  // namespace ov
 diff --git a/src/frontends/paddle/src/op_table.cpp b/src/frontends/paddle/src/op_table.cpp
-index 769492eb13..3a6b5dda9f 100644
+index 769492eb13..d7de21a481 100644
 --- a/src/frontends/paddle/src/op_table.cpp
 +++ b/src/frontends/paddle/src/op_table.cpp
-@@ -39,9 +39,11 @@ OP_CONVERTER(elementwise_sub);
+@@ -39,9 +39,12 @@ OP_CONVERTER(elementwise_sub);
  OP_CONVERTER(equal);
  OP_CONVERTER(greater_equal);
  OP_CONVERTER(not_equal);
 +OP_CONVERTER(elu);
  OP_CONVERTER(embedding);
  OP_CONVERTER(exp);
  OP_CONVERTER(expand_v2);
++OP_CONVERTER(expand_as_v2);
 +OP_CONVERTER(eye);
  OP_CONVERTER(flip);
  OP_CONVERTER(flatten_contiguous_range);
  OP_CONVERTER(floor);
-@@ -138,6 +140,12 @@ OP_CONVERTER(write_to_array);
+@@ -60,6 +63,7 @@ OP_CONVERTER(index_select);
+ OP_CONVERTER(layer_norm);
+ OP_CONVERTER(leaky_relu);
+ OP_CONVERTER(less_than);
++OP_CONVERTER(less_equal);
+ OP_CONVERTER(linear_interp_v2);
+ OP_CONVERTER(linspace);
+ OP_CONVERTER(lod_array_length);
+@@ -138,6 +142,12 @@ OP_CONVERTER(write_to_array);
  OP_CONVERTER(where_index);
  OP_CONVERTER(yolo_box);
  OP_CONVERTER(generate_proposals_v2);
@@ -430,19 +517,28 @@ index 769492eb13..3a6b5dda9f 100644
  }  // namespace op
  std::map<std::string, CreatorFunction> get_supported_ops() {
      return {{"arg_max", op::argmax},
-@@ -173,9 +181,11 @@ std::map<std::string, CreatorFunction> get_supported_ops() {
+@@ -173,9 +183,12 @@ std::map<std::string, CreatorFunction> get_supported_ops() {
              {"elementwise_sub", op::elementwise_sub},
              {"dropout", op::dropout},
              {"elementwise_pow", op::elementwise_pow},
 +            {"elu", op::elu},
              {"equal", op::equal},
              {"exp", op::exp},
              {"expand_v2", op::expand_v2},
++            {"expand_as_v2", op::expand_as_v2},
 +            {"eye", op::eye},
              {"fill_any_like", op::fill_any_like},
              {"fill_constant", op::fill_constant},
              {"fill_constant_batch_size_like", op::fill_constant_batch_size_like},
-@@ -277,7 +287,13 @@ std::map<std::string, CreatorFunction> get_supported_ops() {
+@@ -196,6 +209,7 @@ std::map<std::string, CreatorFunction> get_supported_ops() {
+             {"layer_norm", op::layer_norm},
+             {"leaky_relu", op::leaky_relu},
+             {"less_than", op::less_than},
++            {"less_equal", op::less_equal},
+             {"linear_interp_v2", op::linear_interp_v2},
+             {"linspace", op::linspace},
+             {"lod_array_length", op::lod_array_length},
+@@ -277,7 +291,13 @@ std::map<std::string, CreatorFunction> get_supported_ops() {
              {"while", op::while_},
              {"write_to_array", op::write_to_array},
              {"where_index", op::where_index},
@@ -479,100 +575,3 @@ index 99357a3a33..53ea785260 100644
      std::string("flip_1/flip_1.pdmodel"),
      std::string("flip_2/flip_2.pdmodel"),
      std::string("flip_3/flip_3.pdmodel"),
-diff --git a/src/frontends/paddle/tests/test_models/gen_scripts/generate_elu.py b/src/frontends/paddle/tests/test_models/gen_scripts/generate_elu.py
-new file mode 100644
-index 0000000000..4dc67b2051
---- /dev/null
-+++ b/src/frontends/paddle/tests/test_models/gen_scripts/generate_elu.py
-@@ -0,0 +1,44 @@
-+# Copyright (C) 2018-2024 Intel Corporation
-+# SPDX-License-Identifier: Apache-2.0
-+
-+#
-+# relu6 paddle model generator
-+#
-+import numpy as np
-+from save_model import saveModel
-+import paddle
-+import sys
-+
-+
-+def elu(name: str, x, alpha=None, data_type='float32'):
-+    paddle.enable_static()
-+
-+    with paddle.static.program_guard(paddle.static.Program(), paddle.static.Program()):
-+        node_x = paddle.static.data(name='x', shape=x.shape, dtype=data_type)
-+
-+        if paddle.__version__ >= '2.0.0':
-+            out = paddle.nn.functional.elu(node_x, alpha, name='elu')
-+        else:
-+            out = paddle.fluid.layers.elu(node_x, alpha, name='elu')
-+        cpu = paddle.static.cpu_places(1)
-+        exe = paddle.static.Executor(cpu[0])
-+        # startup program will call initializer to initialize the parameters.
-+        exe.run(paddle.static.default_startup_program())
-+
-+        outs = exe.run(
-+            feed={'x': x},
-+            fetch_list=[out])
-+
-+        saveModel(name, exe, feed_vars=[node_x], fetchlist=[out],
-+                  inputs=[x], outputs=[outs[0]], target_dir=sys.argv[1])
-+
-+    return outs[0]
-+
-+
-+def main():
-+    data_type = 'float32'
-+    data = np.random.randn(2, 3, 4).astype('float32')
-+    elu("elu", data)
-+
-+if __name__ == "__main__":
-+    main()
-diff --git a/src/frontends/paddle/tests/test_models/gen_scripts/generate_eye.py b/src/frontends/paddle/tests/test_models/gen_scripts/generate_eye.py
-new file mode 100644
-index 0000000000..9b1a4f668c
---- /dev/null
-+++ b/src/frontends/paddle/tests/test_models/gen_scripts/generate_eye.py
-@@ -0,0 +1,41 @@
-+# Copyright (C) 2018-2024 Intel Corporation
-+# SPDX-License-Identifier: Apache-2.0
-+
-+#
-+# fill_const paddle model generator
-+#
-+import numpy as np
-+from save_model import saveModel
-+import paddle
-+import sys
-+
-+
-+def eye(name : str, rows, cols = None, dtype = None):
-+    paddle.enable_static()
-+    with paddle.static.program_guard(paddle.static.Program(), paddle.static.Program()):
-+        if paddle.__version__ >= '2.0.0':
-+            x1 = paddle.eye(num_rows=rows, num_columns=cols, dtype=dtype, name='fill')
-+            x2 = paddle.eye(num_rows=rows, num_columns=cols, dtype=dtype, name='fill')
-+        else:
-+            x1 = paddle.fluid.layers.eye(num_rows=rows, num_columns=cols, dtype=dtype, name='fill_constant')
-+            x2 = paddle.fluid.layers.eye(num_rows=rows, num_columns=cols, dtype=dtype, name='fill_constant')
-+        out = paddle.add(x1, x2)
-+        cpu = paddle.static.cpu_places(1)
-+        exe = paddle.static.Executor(cpu[0])
-+        # startup program will call initializer to initialize the parameters.
-+        exe.run(paddle.static.default_startup_program())
-+
-+        outs = exe.run(
-+            fetch_list=[out])
-+
-+        saveModel(name, exe, feed_vars=[], fetchlist=[out], inputs=[], outputs=[outs[0]], target_dir=sys.argv[1])
-+
-+    return outs[0]
-+
-+def main():
-+    eye("eye", 3)
-+    eye("eye_int32", 2, 3, "int32")
-+    eye("eye_int64", 2, 3, "int64")
-+
-+if __name__ == "__main__":
-+    main()