version opencv

Author: Semyon1104

3rdparty/opencv

@@ -20,8 +20,7 @@ int main(int argc, char* argv[]) {
         std::cout << "Warning: oneDNN backend is not compatible with parallel "
                      "execution. Disabling parallelism."
                   << '\n';
-        options.setParallelBackend(
-            ParBackend::kSeq);  // Автоматически sets parallel=false
+        options.setParallelBackend(ParBackend::kSeq);
       }
     } else if (std::string(argv[i]) == "--parallel" && i + 1 < argc) {
       if (options.backend == Backend::kOneDnn) {

app/Graph/graph_build.cpp

@@ -37,21 +37,9 @@ void ConvolutionalLayer::run(const std::vector<Tensor>& input,
   }
   if (options.parallel) {
     switch (options.par_backend) {
-      /*case ParallelBackend::kTBB:
-        implType_ = kTBB;
-        break;*/
       case ParBackend::kThreads:
         implType_ = kSTL;
         break;
-      // case ParallelBackend::kOMP:
-      //   implType = kOMP;
-      //   break;
-      // case ParallelBackend::kKokkos:
-      //   implType = kKokkos;
-      //   break;
-      // case ParallelBackend::kSycl:
-      //   implType = kSycl;
-      //   break;
       case ParBackend::kSeq:
       default:
         implType_ = kDefault;
@@ -114,11 +102,6 @@ void ConvolutionalLayer::run(const std::vector<Tensor>& input,
                            group_, dilations_);
             break;
           }
-          /*case kTBB: {
-            Conv4DTBB<int>(input[0], kernel_, bias_, output[0], stride_, pads_,
-                        group_, dilations_);
-            break;
-          }*/
           default: {
             Conv4D<int>(input[0], kernel_, bias_, output[0], stride_, pads_,
                         group_, dilations_);
@@ -187,10 +170,6 @@ void ConvolutionalLayer::run(const std::vector<Tensor>& input,
                                pads_, group_, dilations_);
               break;
             }
-            /*case kTBB: {
-              Conv4DTBB<float>(input[0], kernel_, bias_, output[0], stride_,
-            pads_, group_, dilations_); break;
-            }*/
             default: {
               Conv4D<float>(input[0], kernel_, bias_, output[0], stride_, pads_,
                             group_, dilations_);

src/layers/ConvLayer.cpp

@@ -19,18 +19,6 @@ void PoolingLayer::run(const std::vector<Tensor>& input,
       case ParBackend::kTbb:
         implType_ = kTBB;
         break;
-      /*case ParallelBackend::kSTL:
-        implType_ = kSTL;
-        break;*/
-      // case ParallelBackend::kOMP:
-      //   implType = kOMP;
-      //   break;
-      // case ParallelBackend::kKokkos:
-      //   implType = kKokkos;
-      //   break;
-      // case ParallelBackend::kSycl:
-      //   implType = kSycl;
-      //   break;
       case ParBackend::kSeq:
       default:
         implType_ = kDefault;
@@ -48,14 +36,6 @@ void PoolingLayer::run(const std::vector<Tensor>& input,
                                   used_impl.get_output_shape());
           break;
         }
-        /*case kSTL: {
-          PoolingLayerImplSTL<int> used_impl(
-              input[0].get_shape(), poolingShape_, strides_, pads_, dilations_,
-              ceil_mode_, poolingType_);
-          output[0] = make_tensor(used_impl.run(*input[0].as<int>()),
-                                  used_impl.get_output_shape());
-          break;
-        }*/
         default: {
           PoolingLayerImpl<int> used_impl(input[0].get_shape(), poolingShape_,
                                           strides_, pads_, dilations_,
@@ -77,14 +57,6 @@ void PoolingLayer::run(const std::vector<Tensor>& input,
                                   used_impl.get_output_shape());
           break;
         }
-        /*case kSTL: {
-          PoolingLayerImplSTL<float> used_impl(
-              input[0].get_shape(), poolingShape_, strides_, pads_, dilations_,
-              ceil_mode_, poolingType_);
-          output[0] = make_tensor(used_impl.run(*input[0].as<float>()),
-                                  used_impl.get_output_shape());
-          break;
-        }*/
         default: {
           PoolingLayerImpl<float> used_impl(input[0].get_shape(), poolingShape_,
                                             strides_, pads_, dilations_,