Rewrite dynamic fusion

The new version introduces the following major changes:

* Change public interface to simplify and standardize the user experience
    - Use the term "Workload" uniformly
    - Simplify operator interface to be a set of static methods:
      validate_op(), create_op()

* Separate the kernel writing into its own component (template_writer).
  This is to allow the co-development of GpuKernelWriter, and to allow
  easy replacement once GpuKernelWriter is mature.

* Optimize the core fusion algorithm used by the component graph. The
  details can be found in GpuKernelComponentGraph::fuse()

* Use Gpu instead of Cl prefixes for most of the Workload interfaces
  (except for runtime and kernel components, which have to be language specific)
  This allows the potential extension to other Gpu langauges in the
  future.

* Refactor runtime memory interface so that auxiliary tensor handling
  is separate from the user tensor passing. This is because the former
  is less stable and may require extension in the future.

* Hide source code object from the user as it is not required at the
  moment

* Deprecate the old prototype entirely by disabling it in SCons build

Resolves COMPMID-5510, COMPMID-5512, COMPMID-5513

Change-Id: If69d2362856f2de4503546b7b6cf48a525cf3079
Signed-off-by: SiCong Li <sicong.li@arm.com>
Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/8406
Tested-by: Arm Jenkins <bsgcomp@arm.com>
Reviewed-by: Gian Marco Iodice <gianmarco.iodice@arm.com>
Reviewed-by: Jakub Sujak <jakub.sujak@arm.com>
Reviewed-by: Viet-Hoa Do <viet-hoa.do@arm.com>
Comments-Addressed: Arm Jenkins <bsgcomp@arm.com>
Benchmark: Arm Jenkins <bsgcomp@arm.com>

1 files changed, 104 insertions, 0 deletions

diff --git a/src/dynamic_fusion/sketch/gpu/GpuKernelComponentGraph.h b/src/dynamic_fusion/sketch/gpu/GpuKernelComponentGraph.h
new file mode 100644
index 0000000000..fbcb2c10ea
--- /dev/null
+++ b/src/dynamic_fusion/sketch/gpu/GpuKernelComponentGraph.h
@@ -0,0 +1,104 @@
+/*
+ * Copyright (c) 2022 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.
+ */
+#ifndef SRC_DYNAMIC_FUSION_SKETCH_GPU_GPUKERNELCOMPONENTGRAPH
+#define SRC_DYNAMIC_FUSION_SKETCH_GPU_GPUKERNELCOMPONENTGRAPH
+
+#include "src/dynamic_fusion/sketch/ArgumentPack.h"
+#include "src/dynamic_fusion/sketch/gpu/GpuComponentServices.h"
+#include "src/dynamic_fusion/sketch/gpu/GpuKernelComponentStream.h"
+#include "src/dynamic_fusion/sketch/utils/DependencyGraph.h"
+
+#include <vector>
+
+namespace arm_compute
+{
+namespace experimental
+{
+namespace dynamic_fusion
+{
+class IGpuKernelComponent;
+
+/** A multi-input (tensors), multi-output (tensors) acyclic directed graph of gpu kernel components
+ * Its main purposes are:
+ *  - Perform "graph-level" optimizations like fusion of kernel components (not the fusion of operators)
+ *  - Automatically assign memory descriptions @ref MemoryDescriptor of all tensors based on graph topology
+ */
+class GpuKernelComponentGraph
+{
+public:
+    /** Constructor
+     *
+     * @param[in] services @ref GpuComponentServices to be used by the graph
+     */
+    GpuKernelComponentGraph(GpuComponentServices *services);
+    /** Prevent instances of this class from being copy constructed */
+    GpuKernelComponentGraph(const GpuKernelComponentGraph &graph) = delete;
+    /** Prevent instances of this class from being copied */
+    GpuKernelComponentGraph &operator=(const GpuKernelComponentGraph &graph) = delete;
+    /** Allow instances of this class to be move constructed */
+    GpuKernelComponentGraph(GpuKernelComponentGraph &&graph) = default;
+    /** Allow instances of this class to be moved */
+    GpuKernelComponentGraph &operator=(GpuKernelComponentGraph &&graph) = default;
+    /** Create a new component and add it to the component graph
+     * Component id is automatically allocated
+     *
+     * @tparam T    Component type
+     * @tparam Args Component argument types
+     *
+     * @param[in] args Component arguments except for component id, which is auto-allocated
+     */
+    template <typename T, typename... Args>
+    void add_new_component(Args &&... args)
+    {
+        auto                      comp           = _services->component_factory().create<T>(std::forward<Args>(args)...);
+        ArgumentPack<ITensorInfo> tensors        = comp->tensors();
+        const auto                src_tensor_ids = get_tensor_ids(tensors.get_const_src_tensors());
+        const auto                dst_tensor_ids = get_tensor_ids(tensors.get_const_dst_tensors());
+        bool                      success        = _dependency_graph.add_operator(comp->id(), src_tensor_ids, dst_tensor_ids);
+        ARM_COMPUTE_ERROR_ON(!success);
+        _components[comp->id()] = std::move(comp);
+        for(auto t : tensors.get_const_src_tensors())
+        {
+            _tensors[t->id()] = t;
+        }
+        for(auto t : tensors.get_const_dst_tensors())
+        {
+            _tensors[t->id()] = t;
+        }
+    }
+    /** Perform component fusion and serialize the graph into a stream of component groups
+     */
+    GpuKernelComponentStream fuse() const;
+
+private:
+    static std::vector<DependencyGraph::TensorId> get_tensor_ids(const std::vector<const ITensorInfo *> tensors);
+    GpuComponentServices *_services;
+    std::map<ComponentId, std::unique_ptr<IGpuKernelComponent>> _components;
+    std::map<ITensorInfo::Id, const ITensorInfo *>              _tensors;
+    DependencyGraph _dependency_graph{};
+};
+} // namespace dynamic_fusion
+} // namespace experimental
+} // namespace arm_compute
+#endif /* SRC_DYNAMIC_FUSION_SKETCH_GPU_GPUKERNELCOMPONENTGRAPH */