Implement the OpenCL kernel to compute the indirect convolution

- Implement indirect convolution kernel
- Add operator support
- Add test

Resolves COMPMID-5709

Change-Id: I9272304163471a5a40da7fdec204599f3c1d8e32
Signed-off-by: Gian Marco Iodice <gianmarco.iodice@arm.com>
Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/8701
Comments-Addressed: Arm Jenkins <bsgcomp@arm.com>
Reviewed-by: Gunes Bayir <gunes.bayir@arm.com>
Tested-by: Arm Jenkins <bsgcomp@arm.com>
Benchmark: Arm Jenkins <bsgcomp@arm.com>

1 files changed, 86 insertions, 0 deletions

diff --git a/src/runtime/CL/functions/CLIndirectConvolutionLayer.cpp b/src/runtime/CL/functions/CLIndirectConvolutionLayer.cpp
new file mode 100644
index 0000000000..90af36aa77
--- /dev/null
+++ b/src/runtime/CL/functions/CLIndirectConvolutionLayer.cpp
@@ -0,0 +1,86 @@
+/*
+ * 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.
+ */
+#include "arm_compute/runtime/CL/functions/CLIndirectConvolutionLayer.h"
+
+#include "arm_compute/core/CL/ICLTensor.h"
+#include "arm_compute/core/Utils.h"
+#include "arm_compute/core/Validate.h"
+#include "src/gpu/cl/operators/ClIndirectConv2d.h"
+
+#include "src/common/utils/Log.h"
+
+namespace arm_compute
+{
+struct CLIndirectConvolutionLayer::Impl
+{
+    const ICLTensor                          *src{ nullptr };
+    const ICLTensor                          *weights{ nullptr };
+    const ICLTensor                          *biases{ nullptr };
+    ICLTensor                                *dst{ nullptr };
+    std::unique_ptr<opencl::ClIndirectConv2d> op{ nullptr };
+};
+
+CLIndirectConvolutionLayer::CLIndirectConvolutionLayer()
+    : _impl(std::make_unique<Impl>())
+{
+}
+CLIndirectConvolutionLayer::CLIndirectConvolutionLayer(CLIndirectConvolutionLayer &&) = default;
+CLIndirectConvolutionLayer &CLIndirectConvolutionLayer::operator=(CLIndirectConvolutionLayer &&) = default;
+CLIndirectConvolutionLayer::~CLIndirectConvolutionLayer()                                        = default;
+
+void CLIndirectConvolutionLayer::configure(ICLTensor *input, const ICLTensor *weights, const ICLTensor *biases, ICLTensor *output, const PadStrideInfo &conv_info, const ActivationLayerInfo &act_info)
+{
+    configure(CLKernelLibrary::get().get_compile_context(), input, weights, biases, output, conv_info, act_info);
+}
+
+void CLIndirectConvolutionLayer::configure(const CLCompileContext &compile_context, ICLTensor *input, const ICLTensor *weights, const ICLTensor *biases, ICLTensor *output,
+                                           const PadStrideInfo &conv_info, const ActivationLayerInfo &act_info)
+{
+    ARM_COMPUTE_ERROR_ON_NULLPTR(input, weights, output);
+    ARM_COMPUTE_LOG_PARAMS(input, weights, biases, output, conv_info, act_info);
+
+    _impl->src     = input;
+    _impl->weights = weights;
+    _impl->biases  = biases;
+    _impl->dst     = output;
+    _impl->op      = std::make_unique<opencl::ClIndirectConv2d>();
+    _impl->op->configure(compile_context, input->info(), weights->info(), (biases != nullptr) ? biases->info() : nullptr, output->info(), conv_info, act_info);
+}
+
+Status CLIndirectConvolutionLayer::validate(const ITensorInfo *input, const ITensorInfo *weights, const ITensorInfo *biases, const ITensorInfo *output, const PadStrideInfo &conv_info,
+                                            const ActivationLayerInfo &act_info)
+{
+    return opencl::ClIndirectConv2d::validate(input, weights, biases, output, conv_info, act_info);
+}
+
+void CLIndirectConvolutionLayer::run()
+{
+    ITensorPack pack;
+    pack.add_tensor(TensorType::ACL_SRC, _impl->src);
+    pack.add_tensor(TensorType::ACL_SRC_1, _impl->weights);
+    pack.add_tensor(TensorType::ACL_SRC_2, _impl->biases);
+    pack.add_tensor(TensorType::ACL_DST, _impl->dst);
+    _impl->op->run(pack);
+}
+}