inference-engine/tests/unit/engines/mkldnn/graph/layers/internal/graph_activation_test.cpp

   1 // Copyright (C) 2018 Intel Corporation
   2 //
   3 // SPDX-License-Identifier: Apache-2.0
   4 //
   5
   6 #include <gtest/gtest.h>
   7 #include <gmock/gmock-spec-builders.h>
   8 #include "mkldnn_plugin/mkldnn_graph.h"
   9 #include "mock_mkldnn_primitive.hpp"
  10 #include "test_graph.hpp"
  11 #include "single_layer_common.hpp"
  12 #include <mkldnn_plugin/mkldnn_extension_utils.h>
  13 #include <inference_engine/cnn_network_impl.hpp>
  14 #include "tests_common.hpp"
  15
  16 using namespace ::testing;
  17 using namespace std;
  18 using namespace mkldnn;
  19
  20 struct activation_test_params {
  21     mkldnn::algorithm alg;
  22     float alpha;
  23     float beta;
  24
  25     struct {
  26         size_t n;
  27         size_t c;
  28         size_t h;
  29         size_t w;
  30     } in;
  31
  32     size_t num_prim_desc;
  33
  34     MKLDNNPlugin::impl_desc_type selectedType;
  35     std::vector<MKLDNNPlugin::impl_desc_type> preferTypes;
  36
  37     std::vector<std::function<void(MKLDNNPlugin::PrimitiveDescInfo)>> comp;
  38 };
  39
  40 template <typename T, typename A> inline T relu_fwd(T s, A alpha) {
  41     return s > 0 ? s : static_cast<T>(s * alpha);
  42 }
  43
  44 template <typename T, typename A> T elu_fwd(T s, A alpha) {
  45     return s > 0 ? s : static_cast<T>(alpha * (::expf(s) - 1));
  46 }
  47
  48 template <typename T>
  49 T logistic_fwd(T s) {
  50     T v = ::expf(s);
  51     return v / (v + 1);
  52 }
  53
  54 template <typename T, typename A>
  55 T bounded_relu_fwd(T s, A alpha) {
  56     s = s > 0 ? s : 0;
  57     return s > alpha ? (T)(alpha) : s;
  58 }
  59
  60 template <typename data_t>
  61 void ref_activation(const InferenceEngine::TBlob<data_t> &src, InferenceEngine::TBlob<data_t> &dst, activation_test_params prm) {
  62     size_t IW = src.dims()[3];
  63     size_t IH = src.dims()[2];
  64     size_t IC = src.dims()[1];
  65     size_t MB = src.dims()[0];
  66
  67     const data_t *src_data = src.readOnly();
  68     data_t *dst_data = dst.data();
  69
  70     for(int mb = 0; mb < MB; mb++) {
  71         for(int c = 0; c < IC; c++) {
  72             for(int h = 0; h < IH; h++) {
  73                 for(int w = 0; w < IW; w++) {
  74                     int idx = mb * IC * IH * IW
  75                               + c * IH * IW
  76                               + h * IW + w;
  77
  78                     switch (prm.alg) {
  79                         case eltwise_relu:         dst_data[idx] = relu_fwd(src_data[idx], prm.alpha);         break;
  80                         case eltwise_elu:          dst_data[idx] = elu_fwd(src_data[idx], prm.alpha);          break;
  81                         case eltwise_logistic:     dst_data[idx] = logistic_fwd(src_data[idx]);                break;
  82                         case eltwise_bounded_relu: dst_data[idx] = bounded_relu_fwd(src_data[idx], prm.alpha); break;
  83                         default: assert(!"unknown alg_kind");
  84                     }
  85                 }
  86             }
  87         }
  88     }
  89 }
  90
  91 class MKLDNNGraphActivationTests: public TestsCommon,
  92                                      public WithParamInterface<activation_test_params> {
  93     std::string model_t = R"V0G0N(
  94 <Net Name="Activation" version="2" precision="FP32" batch="1">
  95     <layers>
  96         <layer name="in1" type="Input" precision="FP32" id="0">
  97             <output>
  98                 <port id="0">
  99                     <dim>_IN_</dim>
 100                     <dim>_IC_</dim>
 101                     <dim>_IH_</dim>
 102                     <dim>_IW_</dim>
 103                 </port>
 104             </output>
 105         </layer>
 106         <layer name="activation" id="1" type="_LT_" precision="FP32">
 107             <data _P1_NAME_="_P1_VAL_" _P2_NAME_="_P2_VAL_" PrimitivesPriority="_IMPLS_"/>
 108             <input>
 109                 <port id="1">
 110                     <dim>_IN_</dim>
 111                     <dim>_IC_</dim>
 112                     <dim>_IH_</dim>
 113                     <dim>_IW_</dim>
 114                 </port>
 115             </input>
 116             <output>
 117                 <port id="2">
 118                     <dim>_IN_</dim>
 119                     <dim>_IC_</dim>
 120                     <dim>_IH_</dim>
 121                     <dim>_IW_</dim>
 122                 </port>
 123             </output>
 124         </layer>
 125     </layers>
 126     <edges>
 127         <edge from-layer="0" from-port="0" to-layer="1" to-port="1"/>
 128     </edges>
 129 </Net>
 130 )V0G0N";
 131
 132 protected:
 133     virtual void TearDown() {
 134     }
 135
 136     std::string getModel(activation_test_params p) {
 137         std::string model = model_t;
 138
 139         switch (p.alg) {
 140             case eltwise_relu:         REPLACE_WITH_STR(model, "_LT_", "ReLU"); break;
 141             case eltwise_elu:          REPLACE_WITH_STR(model, "_LT_", "ELU"); break;
 142             case eltwise_logistic:     REPLACE_WITH_STR(model, "_LT_", "Sigmoid"); break;
 143             case eltwise_bounded_relu: REPLACE_WITH_STR(model, "_LT_", "ReLU6"); break;
 144             default: assert(!"unknown alg_kind");
 145         }
 146
 147         if (p.alg == eltwise_relu)
 148             REPLACE_WITH_STR(model, "_P1_NAME_", "negative_slope");
 149         else if (p.alg == eltwise_bounded_relu)
 150             REPLACE_WITH_STR(model, "_P1_NAME_", "n");
 151         else
 152             REPLACE_WITH_STR(model, "_P1_NAME_", "alpha");
 153         REPLACE_WITH_NUM(model, "_P1_VAL_", p.alpha);
 154
 155         REPLACE_WITH_STR(model, "_P2_NAME_", "beta");
 156         REPLACE_WITH_NUM(model, "_P2_VAL_", p.beta);
 157
 158         REPLACE_WITH_NUM(model, "_IW_", p.in.w);
 159         REPLACE_WITH_NUM(model, "_IH_", p.in.h);
 160         REPLACE_WITH_NUM(model, "_IC_", p.in.c);
 161         REPLACE_WITH_NUM(model, "_IN_", p.in.n);
 162
 163         std::string impls;
 164         for (const auto& preferType : p.preferTypes) {
 165             if (!impls.empty())
 166                 impls += ",";
 167             impls += "cpu:" + MKLDNNGraphTestClass::getStrPrimitiveDescriptorType(preferType);
 168         }
 169         REPLACE_WITH_STR(model, "_IMPLS_", impls);
 170
 171         return model;
 172     }
 173
 174     virtual void SetUp() {
 175         try {
 176             TestsCommon::SetUp();
 177             activation_test_params p = ::testing::WithParamInterface<activation_test_params>::GetParam();
 178             std::string model = getModel(p);
 179
 180             InferenceEngine::CNNNetReader net_reader;
 181             ASSERT_NO_THROW(net_reader.ReadNetwork(model.data(), model.length()));
 182
 183             MKLDNNGraphTestClass graph;
 184             graph.CreateGraph(net_reader.getNetwork());
 185             auto& nodes = graph.getNodes();
 186             for (int i = 0; i < nodes.size(); i++) {
 187                 if (nodes[i]->getType() == MKLDNNPlugin::Activation) {
 188                     ASSERT_LE(p.num_prim_desc, nodes[i]->getSupportedPrimitiveDescriptors().size());
 189                     for (size_t j = 0; j < p.num_prim_desc && j < p.comp.size(); j++) {
 190                         p.comp.at(j)(nodes[i]->getSupportedPrimitiveDescriptors().at(j));
 191                     }
 192                     ASSERT_NE(nullptr, nodes[i]->getSelectedPrimitiveDescriptor());
 193                     ASSERT_EQ(p.selectedType,
 194                               nodes[i]->getSelectedPrimitiveDescriptor()->getImplementationType() & p.selectedType);
 195                 }
 196             }
 197
 198             InferenceEngine::SizeVector dims_src = {p.in.n, p.in.c, p.in.h, p.in.w};
 199
 200             InferenceEngine::Blob::Ptr src = InferenceEngine::make_shared_blob<float, const InferenceEngine::SizeVector>(InferenceEngine::Precision::FP32, InferenceEngine::NCHW, dims_src);
 201             src->allocate();
 202             fill_data(src->buffer(), src->size());
 203
 204             InferenceEngine::TBlob<float>* srcPtr = dynamic_cast<InferenceEngine::TBlob<float>*>(src.get());
 205
 206             if (srcPtr == nullptr)
 207                 FAIL() << "Cannot cast blob to TBlob<float>.";
 208
 209             InferenceEngine::BlobMap srcs;
 210             srcs.insert(std::pair<std::string, InferenceEngine::Blob::Ptr>("in1", src));
 211
 212             InferenceEngine::OutputsDataMap out;
 213             out = net_reader.getNetwork().getOutputsInfo();
 214             InferenceEngine::BlobMap outputBlobs;
 215
 216             std::pair<std::string, InferenceEngine::DataPtr> item = *out.begin();
 217
 218             InferenceEngine::TBlob<float>::Ptr output;
 219             output = InferenceEngine::make_shared_blob<float>(item.second->getTensorDesc());
 220             output->allocate();
 221             outputBlobs[item.first] = output;
 222
 223             graph.Infer(srcs, outputBlobs);
 224
 225             InferenceEngine::TBlob<float> dst_ref(item.second->getTensorDesc());
 226             dst_ref.allocate();
 227
 228             ref_activation(*srcPtr, dst_ref, p);
 229
 230             compare(*output, dst_ref);
 231         } catch (const InferenceEngine::details::InferenceEngineException &e) {
 232             FAIL() << e.what();
 233         }
 234     }
 235 };
 236
 237 TEST_P(MKLDNNGraphActivationTests, TestsActivation) {}
 238
 239 INSTANTIATE_TEST_CASE_P(
 240         TestsActivation, MKLDNNGraphActivationTests,
 241         ::testing::Values(
 242                 activation_test_params{eltwise_relu, 0.0f, 0.0f, {1, 32, 128, 256}, 3, MKLDNNPlugin::impl_desc_type::jit},
 243                 activation_test_params{eltwise_relu, 0.0f, 0.0f, {4, 3, 228, 228}, 3, MKLDNNPlugin::impl_desc_type::jit},
 244                 activation_test_params{eltwise_relu, 0.5f, 0.5f, {1, 32, 128, 256}, 3, MKLDNNPlugin::impl_desc_type::jit},
 245                 activation_test_params{eltwise_relu, 0.5f, 0.5f, {4, 3, 228, 228}, 3, MKLDNNPlugin::impl_desc_type::jit},
 246                 activation_test_params{eltwise_elu, 0.5f, 0.5f, {1, 32, 128, 256}, 3, MKLDNNPlugin::impl_desc_type::jit},
 247                 activation_test_params{eltwise_elu, 0.5f, 0.5f, {4, 3, 228, 228}, 3, MKLDNNPlugin::impl_desc_type::jit},
 248                 activation_test_params{eltwise_elu, 1.0f, 1.0f, {1, 32, 128, 256}, 3, MKLDNNPlugin::impl_desc_type::jit},
 249                 activation_test_params{eltwise_elu, 1.0f, 1.0f, {4, 3, 228, 228}, 3, MKLDNNPlugin::impl_desc_type::jit},
 250                 activation_test_params{eltwise_logistic, 0.0f, 0.0f, {1, 32, 128, 256}, 3, MKLDNNPlugin::impl_desc_type::jit},
 251                 activation_test_params{eltwise_logistic, 0.0f, 0.0f, {4, 3, 228, 228}, 3, MKLDNNPlugin::impl_desc_type::jit},
 252                 activation_test_params{eltwise_bounded_relu, 6.0f, 0.0f, {1, 32, 128, 256}, 3, MKLDNNPlugin::impl_desc_type::jit},
 253                 activation_test_params{eltwise_bounded_relu, 6.0f, 0.0f, {4, 3, 228, 228}, 3, MKLDNNPlugin::impl_desc_type::jit},
 254                 activation_test_params{eltwise_bounded_relu, 0.1f, 0.0f, {1, 32, 128, 256}, 3, MKLDNNPlugin::impl_desc_type::jit},
 255                 activation_test_params{eltwise_bounded_relu, 0.1f, 0.0f, {4, 3, 228, 228}, 3, MKLDNNPlugin::impl_desc_type::jit},
 256                 activation_test_params{eltwise_relu, 0.0f, 0.0f, {1, 32, 128, 256}, 3, MKLDNNPlugin::impl_desc_type::ref, {MKLDNNPlugin::impl_desc_type::ref_any}},
 257                 activation_test_params{eltwise_relu, 0.0f, 0.0f, {4, 3, 228, 228}, 3, MKLDNNPlugin::impl_desc_type::ref, {MKLDNNPlugin::impl_desc_type::ref_any}},
 258                 activation_test_params{eltwise_relu, 0.5f, 0.5f, {1, 32, 128, 256}, 3, MKLDNNPlugin::impl_desc_type::ref, {MKLDNNPlugin::impl_desc_type::ref_any}},
 259                 activation_test_params{eltwise_relu, 0.5f, 0.5f, {4, 3, 228, 228}, 3, MKLDNNPlugin::impl_desc_type::ref, {MKLDNNPlugin::impl_desc_type::ref_any}},
 260                 activation_test_params{eltwise_elu, 0.5f, 0.5f, {1, 32, 128, 256}, 3, MKLDNNPlugin::impl_desc_type::ref, {MKLDNNPlugin::impl_desc_type::ref_any}},
 261                 activation_test_params{eltwise_elu, 0.5f, 0.5f, {4, 3, 228, 228}, 3, MKLDNNPlugin::impl_desc_type::ref, {MKLDNNPlugin::impl_desc_type::ref_any}},
 262                 activation_test_params{eltwise_elu, 1.0f, 1.0f, {1, 32, 128, 256}, 3, MKLDNNPlugin::impl_desc_type::ref, {MKLDNNPlugin::impl_desc_type::ref_any}},
 263                 activation_test_params{eltwise_elu, 1.0f, 1.0f, {4, 3, 228, 228}, 3, MKLDNNPlugin::impl_desc_type::ref, {MKLDNNPlugin::impl_desc_type::ref_any}},
 264                 activation_test_params{eltwise_logistic, 0.0f, 0.0f, {1, 32, 128, 256}, 3, MKLDNNPlugin::impl_desc_type::ref, {MKLDNNPlugin::impl_desc_type::ref_any}},
 265                 activation_test_params{eltwise_logistic, 0.0f, 0.0f, {4, 3, 228, 228}, 3, MKLDNNPlugin::impl_desc_type::ref, {MKLDNNPlugin::impl_desc_type::ref_any}},
 266                 activation_test_params{eltwise_bounded_relu, 6.0f, 0.0f, {1, 32, 128, 256}, 3, MKLDNNPlugin::impl_desc_type::ref, {MKLDNNPlugin::impl_desc_type::ref_any}},
 267                 activation_test_params{eltwise_bounded_relu, 6.0f, 0.0f, {4, 3, 228, 228}, 3, MKLDNNPlugin::impl_desc_type::ref, {MKLDNNPlugin::impl_desc_type::ref_any}},
 268                 activation_test_params{eltwise_bounded_relu, 0.1f, 0.0f, {1, 32, 128, 256}, 3, MKLDNNPlugin::impl_desc_type::ref, {MKLDNNPlugin::impl_desc_type::ref_any}},
 269                 activation_test_params{eltwise_bounded_relu, 0.1f, 0.0f, {4, 3, 228, 228}, 3, MKLDNNPlugin::impl_desc_type::ref, {MKLDNNPlugin::impl_desc_type::ref_any}}
 270         ));
 271
 272 class MKLDNNGraphDynBatchActivationTests: public MKLDNNGraphActivationTests {
 273 protected:
 274     virtual void SetUp() {
 275         try {
 276             TestsCommon::SetUp();
 277             activation_test_params p = ::testing::WithParamInterface<activation_test_params>::GetParam();
 278             std::string model = getModel(p);
 279             size_t MB = p.in.n;
 280             if (MB < 2)
 281                 MB = 2;
 282
 283             InferenceEngine::CNNNetReader net_reader;
 284             ASSERT_NO_THROW(net_reader.ReadNetwork(model.data(), model.length()));
 285             InferenceEngine::CNNNetwork network = net_reader.getNetwork();
 286             auto implNet = dynamic_cast<InferenceEngine::details::CNNNetworkImpl *>(&((InferenceEngine::ICNNNetwork&)network));
 287             ASSERT_NE(nullptr, implNet) << "Failed to cast ICNNNetwork to CNNNetworkImpl";
 288             InferenceEngine::ResponseDesc resp;
 289             InferenceEngine::StatusCode sts  = implNet->setBatchSizeReshape(MB, &resp);
 290             ASSERT_EQ((int)InferenceEngine::StatusCode::OK, sts) << resp.msg;
 291
 292             MKLDNNGraphTestClass graph;
 293             graph.setProperty({{InferenceEngine::PluginConfigParams::KEY_DYN_BATCH_ENABLED, InferenceEngine::PluginConfigParams::YES}});
 294             graph.CreateGraph(net_reader.getNetwork());
 295
 296             InferenceEngine::SizeVector dims_src = {MB, p.in.c, p.in.h, p.in.w};
 297
 298             InferenceEngine::Blob::Ptr src = InferenceEngine::make_shared_blob<float, const InferenceEngine::SizeVector>(InferenceEngine::Precision::FP32, InferenceEngine::NCHW, dims_src);
 299             src->allocate();
 300             fill_data(src->buffer(), src->size());
 301
 302             auto * srcPtr = dynamic_cast<InferenceEngine::TBlob<float>*>(src.get());
 303
 304             if (srcPtr == nullptr)
 305                 FAIL() << "Cannot cast blob to TBlob<float>.";
 306
 307             InferenceEngine::BlobMap srcs;
 308             srcs.insert(std::pair<std::string, InferenceEngine::Blob::Ptr>("in1", src));
 309
 310             InferenceEngine::OutputsDataMap out;
 311             out = net_reader.getNetwork().getOutputsInfo();
 312             InferenceEngine::BlobMap outputBlobs;
 313
 314             std::pair<std::string, InferenceEngine::DataPtr> item = *out.begin();
 315
 316             InferenceEngine::TBlob<float>::Ptr output;
 317             output = InferenceEngine::make_shared_blob<float>(item.second->getTensorDesc());
 318             output->allocate();
 319             outputBlobs[item.first] = output;
 320
 321             auto checkActivation = [](const MKLDNNPlugin::MKLDNNNodePtr& node) {
 322                 return node->getType() == MKLDNNPlugin::Activation;
 323             };
 324
 325             graph.checkDynBatch(srcs, outputBlobs, MB, MB, checkActivation);
 326             graph.checkDynBatch(srcs, outputBlobs, 1, MB, checkActivation);
 327         } catch (const InferenceEngine::details::InferenceEngineException &e) {
 328             FAIL() << e.what();
 329         }
 330     }
 331 };
 332
 333 TEST_P(MKLDNNGraphDynBatchActivationTests, TestsDynBatchActivation) {}
 334
 335
 336 INSTANTIATE_TEST_CASE_P(
 337         TestsDynBatchActivation, MKLDNNGraphDynBatchActivationTests,
 338         ::testing::Values(
 339                 activation_test_params{eltwise_relu, 0.0f, 0.0f, {2, 32, 128, 256}, 3, MKLDNNPlugin::impl_desc_type::jit},
 340                 activation_test_params{eltwise_relu, 0.0f, 0.0f, {4, 3, 228, 228}, 3, MKLDNNPlugin::impl_desc_type::jit},
 341                 activation_test_params{eltwise_relu, 0.5f, 0.5f, {1, 32, 128, 256}, 3, MKLDNNPlugin::impl_desc_type::jit},
 342                 activation_test_params{eltwise_relu, 0.5f, 0.5f, {4, 3, 228, 228}, 3, MKLDNNPlugin::impl_desc_type::jit},
 343                 activation_test_params{eltwise_elu, 0.5f, 0.5f, {1, 32, 128, 256}, 3, MKLDNNPlugin::impl_desc_type::jit},
 344                 activation_test_params{eltwise_elu, 0.5f, 0.5f, {4, 3, 228, 228}, 3, MKLDNNPlugin::impl_desc_type::jit},
 345                 activation_test_params{eltwise_elu, 1.0f, 1.0f, {1, 32, 128, 256}, 3, MKLDNNPlugin::impl_desc_type::jit},
 346                 activation_test_params{eltwise_elu, 1.0f, 1.0f, {4, 3, 228, 228}, 3, MKLDNNPlugin::impl_desc_type::jit},
 347                 activation_test_params{eltwise_logistic, 0.0f, 0.0f, {1, 32, 128, 256}, 3, MKLDNNPlugin::impl_desc_type::jit},
 348                 activation_test_params{eltwise_logistic, 0.0f, 0.0f, {4, 3, 228, 228}, 3, MKLDNNPlugin::impl_desc_type::jit},
 349                 activation_test_params{eltwise_bounded_relu, 6.0f, 0.0f, {1, 32, 128, 256}, 3, MKLDNNPlugin::impl_desc_type::jit},
 350                 activation_test_params{eltwise_bounded_relu, 6.0f, 0.0f, {4, 3, 228, 228}, 3, MKLDNNPlugin::impl_desc_type::jit},
 351                 activation_test_params{eltwise_bounded_relu, 0.1f, 0.0f, {1, 32, 128, 256}, 3, MKLDNNPlugin::impl_desc_type::jit},
 352                 activation_test_params{eltwise_bounded_relu, 0.1f, 0.0f, {4, 3, 228, 228}, 3, MKLDNNPlugin::impl_desc_type::jit},
 353                 activation_test_params{eltwise_relu, 0.0f, 0.0f, {1, 32, 128, 256}, 3, MKLDNNPlugin::impl_desc_type::ref, {MKLDNNPlugin::impl_desc_type::ref_any}},
 354                 activation_test_params{eltwise_relu, 0.0f, 0.0f, {4, 3, 228, 228}, 3, MKLDNNPlugin::impl_desc_type::ref, {MKLDNNPlugin::impl_desc_type::ref_any}},
 355                 activation_test_params{eltwise_relu, 0.5f, 0.5f, {1, 32, 128, 256}, 3, MKLDNNPlugin::impl_desc_type::ref, {MKLDNNPlugin::impl_desc_type::ref_any}},
 356                 activation_test_params{eltwise_relu, 0.5f, 0.5f, {4, 3, 228, 228}, 3, MKLDNNPlugin::impl_desc_type::ref, {MKLDNNPlugin::impl_desc_type::ref_any}},
 357                 activation_test_params{eltwise_elu, 0.5f, 0.5f, {1, 32, 128, 256}, 3, MKLDNNPlugin::impl_desc_type::ref, {MKLDNNPlugin::impl_desc_type::ref_any}},
 358                 activation_test_params{eltwise_elu, 0.5f, 0.5f, {4, 3, 228, 228}, 3, MKLDNNPlugin::impl_desc_type::ref, {MKLDNNPlugin::impl_desc_type::ref_any}},
 359                 activation_test_params{eltwise_elu, 1.0f, 1.0f, {1, 32, 128, 256}, 3, MKLDNNPlugin::impl_desc_type::ref, {MKLDNNPlugin::impl_desc_type::ref_any}},
 360                 activation_test_params{eltwise_elu, 1.0f, 1.0f, {4, 3, 228, 228}, 3, MKLDNNPlugin::impl_desc_type::ref, {MKLDNNPlugin::impl_desc_type::ref_any}},
 361                 activation_test_params{eltwise_logistic, 0.0f, 0.0f, {1, 32, 128, 256}, 3, MKLDNNPlugin::impl_desc_type::ref, {MKLDNNPlugin::impl_desc_type::ref_any}},
 362                 activation_test_params{eltwise_logistic, 0.0f, 0.0f, {4, 3, 228, 228}, 3, MKLDNNPlugin::impl_desc_type::ref, {MKLDNNPlugin::impl_desc_type::ref_any}},
 363                 activation_test_params{eltwise_bounded_relu, 6.0f, 0.0f, {1, 32, 128, 256}, 3, MKLDNNPlugin::impl_desc_type::ref, {MKLDNNPlugin::impl_desc_type::ref_any}},
 364                 activation_test_params{eltwise_bounded_relu, 6.0f, 0.0f, {4, 3, 228, 228}, 3, MKLDNNPlugin::impl_desc_type::ref, {MKLDNNPlugin::impl_desc_type::ref_any}},
 365                 activation_test_params{eltwise_bounded_relu, 0.1f, 0.0f, {1, 32, 128, 256}, 3, MKLDNNPlugin::impl_desc_type::ref, {MKLDNNPlugin::impl_desc_type::ref_any}},
 366                 activation_test_params{eltwise_bounded_relu, 0.1f, 0.0f, {4, 3, 228, 228}, 3, MKLDNNPlugin::impl_desc_type::ref, {MKLDNNPlugin::impl_desc_type::ref_any}}
 367         ));