add back missing part (#1785)

src/ngraph/runtime/gpu/cudnn_emitter.cpp

@@ -1179,7 +1179,31 @@ size_t runtime::gpu::CUDNNEmitter::build_convolution(const std::string& dtype,
     auto& filter_desc = get_cudnn_filter_descriptor(input_filter_shape, data_type, tensor_format);
     auto& conv_desc = get_cudnn_convolution_descriptor(
         padding_below, window_movement_strides, window_dilation_strides, mode, data_type);
-    const cudnnConvolutionFwdAlgo_t conv_fwd_algo = CUDNN_CONVOLUTION_FWD_ALGO_IMPLICIT_GEMM;
+    cudnnConvolutionFwdAlgo_t conv_fwd_algo = CUDNN_CONVOLUTION_FWD_ALGO_IMPLICIT_GEMM;
+
+    if (find_algo != algo_search::NONE)
+    {
+        int num_algos;
+        int max_algos = 0;
+        CUDNN_SAFE_CALL(
+            cudnnGetConvolutionForwardAlgorithmMaxCount(*m_ctx->cudnn_handle, &max_algos));
+        std::vector<cudnnConvolutionFwdAlgoPerf_t> results(max_algos);
+        auto cudnn_algo_search = (find_algo == algo_search::EXPLICIT)
+                                     ? cudnnFindConvolutionForwardAlgorithm
+                                     : cudnnGetConvolutionForwardAlgorithm_v7;
+        CUDNN_SAFE_CALL((*cudnn_algo_search)(*m_ctx->cudnn_handle,
+                                             tensor_desc_0,
+                                             filter_desc,
+                                             conv_desc,
+                                             tensor_desc_1,
+                                             static_cast<int>(results.size()),
+                                             &num_algos,
+                                             results.data()));
+        results.resize(num_algos);
+        conv_fwd_algo =
+            select_cudnn_algo<cudnnConvolutionFwdAlgoPerf_t, cudnnConvolutionFwdAlgo_t>(results);
+    }
+
     void* alpha = m_host_parameters.allocate_by_datatype(data_type, 1.0);
     void* beta = m_host_parameters.allocate_by_datatype(data_type, 0);