1 files changed, 86 insertions, 0 deletions
diff --git a/src/graph2/mutators/DepthConcatSubTensorMutator.cpp b/src/graph2/mutators/DepthConcatSubTensorMutator.cpp
new file mode 100644
index 0000000000..cc8de6bb1b
--- /dev/null
+++ b/src/graph2/mutators/DepthConcatSubTensorMutator.cpp
@@ -0,0 +1,86 @@
+/*
+ * Copyright (c) 2018 ARM Limited.
+ *
+ * SPDX-License-Identifier: MIT
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a copy
+ * of this software and associated documentation files (the "Software"), to
+ * deal in the Software without restriction, including without limitation the
+ * rights to use, copy, modify, merge, publish, distribute, sublicense, and/or
+ * sell copies of the Software, and to permit persons to whom the Software is
+ * furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in all
+ * copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
+ * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+ * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+ * OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ */
+#include "arm_compute/graph2/mutators/DepthConcatSubTensorMutator.h"
+
+#include "arm_compute/graph2/Graph.h"
+#include "arm_compute/graph2/Logger.h"
+#include "arm_compute/graph2/backends/BackendRegistry.h"
+#include "arm_compute/graph2/nodes/DepthConcatenateLayerNode.h"
+
+#include "arm_compute/core/utils/misc/Cast.h"
+#include "arm_compute/core/utils/misc/Iterable.h"
+
+namespace arm_compute
+{
+namespace graph2
+{
+const char *DepthConcatSubTensorMutator::name()
+{
+    return "DepthConcatSubTensorMutator";
+}
+
+void DepthConcatSubTensorMutator::mutate(Graph &g)
+{
+    // Should be in reverse order of execution
+    for(auto &node : arm_compute::utils::iterable::reverse_iterate(g.nodes()))
+    {
+        if(node && node->type() == NodeType::DepthConcatenateLayer && node->output(0) != nullptr)
+        {
+            // Get output tensor
+            auto output_tensor = node->output(0);
+
+            // Check that all tensor have the same target and valid inputs
+            bool is_valid = std::all_of(node->input_edges().cbegin(), node->input_edges().cend(),
+                                        [&](const EdgeID & eid)
+            {
+                return (g.edge(eid) != nullptr) && (g.edge(eid)->tensor() != nullptr) && (g.edge(eid)->tensor()->desc().target == output_tensor->desc().target);
+            });
+
+            // Create subtensors
+            if(is_valid && backends::BackendRegistry::get().find_backend(output_tensor->desc().target) != nullptr)
+            {
+                ARM_COMPUTE_LOG_GRAPH_VERBOSE("Using sub-tensors for the node with ID : "
+                                              << node->id() << " and name : " << node->name() << std::endl);
+                // Create sub-tensor handles
+                unsigned depth = 0;
+                for(unsigned int i = 0; i < node->input_edges().size(); ++i)
+                {
+                    auto       input_tensor = node->input(i);
+                    const auto input_shape  = input_tensor->desc().shape;
+
+                    auto backend = backends::BackendRegistry::get().find_backend(input_tensor->desc().target);
+                    auto handle  = backend->create_subtensor(output_tensor->handle(), input_shape, Coordinates(0, 0, depth));
+                    input_tensor->set_handle(std::move(handle));
+
+                    depth += input_shape.z();
+                }
+
+                auto *dc_node = arm_compute::utils::cast::polymorphic_downcast<DepthConcatenateLayerNode *>(node.get());
+                dc_node->set_enabled(false);
+            }
+        }
+    }
+}
+} // namespace graph2
+} // namespace arm_compute