IVGCVSW-4625 Add CL Rank Workload
* Added CL implementation of Rank Workload
* Removed references to memcpy_s as it's a windows only function
Signed-off-by: David Monahan <david.monahan@arm.com>
Change-Id: Ia63666b9640d76a775f2ab98b3cd7e9f77b5a507
diff --git a/src/backends/cl/ClWorkloadFactory.cpp b/src/backends/cl/ClWorkloadFactory.cpp
index 35186f2..5a5cb89 100644
--- a/src/backends/cl/ClWorkloadFactory.cpp
+++ b/src/backends/cl/ClWorkloadFactory.cpp
@@ -560,6 +560,12 @@
return MakeWorkload<ClQuantizedLstmWorkload>(descriptor, info, m_CLCompileContext);
}
+std::unique_ptr<IWorkload> ClWorkloadFactory::CreateRank(const RankQueueDescriptor& descriptor,
+ const WorkloadInfo& info) const
+{
+ return std::make_unique<ClRankWorkload>(descriptor, info);
+}
+
std::unique_ptr<IWorkload> ClWorkloadFactory::CreateReshape(const ReshapeQueueDescriptor& descriptor,
const WorkloadInfo& info) const
{
diff --git a/src/backends/cl/ClWorkloadFactory.hpp b/src/backends/cl/ClWorkloadFactory.hpp
index c8812cf..66aea84 100644
--- a/src/backends/cl/ClWorkloadFactory.hpp
+++ b/src/backends/cl/ClWorkloadFactory.hpp
@@ -203,6 +203,9 @@
std::unique_ptr<IWorkload> CreateQuantizedLstm(const QuantizedLstmQueueDescriptor& descriptor,
const WorkloadInfo& info) const override;
+ std::unique_ptr<IWorkload> CreateRank(const RankQueueDescriptor& descriptor,
+ const WorkloadInfo& info) const override;
+
std::unique_ptr<IWorkload> CreateReshape(const ReshapeQueueDescriptor& descriptor,
const WorkloadInfo& info) const override;
diff --git a/src/backends/cl/test/ClLayerTests.cpp b/src/backends/cl/test/ClLayerTests.cpp
index 7d40a69..018a62d 100644
--- a/src/backends/cl/test/ClLayerTests.cpp
+++ b/src/backends/cl/test/ClLayerTests.cpp
@@ -340,6 +340,35 @@
ARMNN_AUTO_TEST_CASE_WITH_THF(BatchNormFloat32, BatchNormFloat32Test)
ARMNN_AUTO_TEST_CASE_WITH_THF(BatchNormFloat32Nhwc, BatchNormFloat32NhwcTest)
+// Rank
+ARMNN_AUTO_TEST_CASE_WITH_THF(RankDimSize1Float16, RankDimSize1Test<DataType::Float16>)
+ARMNN_AUTO_TEST_CASE_WITH_THF(RankDimSize1Float32, RankDimSize1Test<DataType::Float32>)
+ARMNN_AUTO_TEST_CASE_WITH_THF(RankDimSize1QAsymmU8, RankDimSize1Test<DataType::QAsymmU8>)
+ARMNN_AUTO_TEST_CASE_WITH_THF(RankDimSize1Signed32, RankDimSize1Test<DataType::Signed32>)
+ARMNN_AUTO_TEST_CASE_WITH_THF(RankDimSize1QSymmS16, RankDimSize1Test<DataType::QSymmS16>)
+ARMNN_AUTO_TEST_CASE_WITH_THF(RankDimSize1QAsymmS8, RankDimSize1Test<DataType::QAsymmS8>)
+
+ARMNN_AUTO_TEST_CASE_WITH_THF(RankDimSize2Float16, RankDimSize2Test<DataType::Float16>)
+ARMNN_AUTO_TEST_CASE_WITH_THF(RankDimSize2Float32, RankDimSize2Test<DataType::Float32>)
+ARMNN_AUTO_TEST_CASE_WITH_THF(RankDimSize2QAsymmU8, RankDimSize2Test<DataType::QAsymmU8>)
+ARMNN_AUTO_TEST_CASE_WITH_THF(RankDimSize2Signed32, RankDimSize2Test<DataType::Signed32>)
+ARMNN_AUTO_TEST_CASE_WITH_THF(RankDimSize2QSymmS16, RankDimSize2Test<DataType::QSymmS16>)
+ARMNN_AUTO_TEST_CASE_WITH_THF(RankDimSize2QAsymmS8, RankDimSize2Test<DataType::QAsymmS8>)
+
+ARMNN_AUTO_TEST_CASE_WITH_THF(RankDimSize3Float16, RankDimSize3Test<DataType::Float16>)
+ARMNN_AUTO_TEST_CASE_WITH_THF(RankDimSize3Float32, RankDimSize3Test<DataType::Float32>)
+ARMNN_AUTO_TEST_CASE_WITH_THF(RankDimSize3QAsymmU8, RankDimSize3Test<DataType::QAsymmU8>)
+ARMNN_AUTO_TEST_CASE_WITH_THF(RankDimSize3Signed32, RankDimSize3Test<DataType::Signed32>)
+ARMNN_AUTO_TEST_CASE_WITH_THF(RankDimSize3QSymmS16, RankDimSize3Test<DataType::QSymmS16>)
+ARMNN_AUTO_TEST_CASE_WITH_THF(RankDimSize3QAsymmS8, RankDimSize3Test<DataType::QAsymmS8>)
+
+ARMNN_AUTO_TEST_CASE_WITH_THF(RankDimSize4Float16, RankDimSize4Test<DataType::Float16>)
+ARMNN_AUTO_TEST_CASE_WITH_THF(RankDimSize4Float32, RankDimSize4Test<DataType::Float32>)
+ARMNN_AUTO_TEST_CASE_WITH_THF(RankDimSize4QAsymmU8, RankDimSize4Test<DataType::QAsymmU8>)
+ARMNN_AUTO_TEST_CASE_WITH_THF(RankDimSize4Signed32, RankDimSize4Test<DataType::Signed32>)
+ARMNN_AUTO_TEST_CASE_WITH_THF(RankDimSize4QSymmS16, RankDimSize4Test<DataType::QSymmS16>)
+ARMNN_AUTO_TEST_CASE_WITH_THF(RankDimSize4QAsymmS8, RankDimSize4Test<DataType::QAsymmS8>)
+
// InstanceNormalization
ARMNN_AUTO_TEST_CASE_WITH_THF(InstanceNormFloat32Nchw, InstanceNormFloat32Test, DataLayout::NCHW);
ARMNN_AUTO_TEST_CASE_WITH_THF(InstanceNormFloat16Nchw, InstanceNormFloat16Test, DataLayout::NCHW);
diff --git a/src/backends/cl/workloads/CMakeLists.txt b/src/backends/cl/workloads/CMakeLists.txt
index 6118d9b..7427ea0 100644
--- a/src/backends/cl/workloads/CMakeLists.txt
+++ b/src/backends/cl/workloads/CMakeLists.txt
@@ -86,6 +86,7 @@
ClQuantizedLstmWorkload.hpp
ClQuantizeWorkload.cpp
ClQuantizeWorkload.hpp
+ ClRankWorkload.hpp
ClReshapeWorkload.cpp
ClReshapeWorkload.hpp
ClResizeWorkload.cpp
diff --git a/src/backends/cl/workloads/ClRankWorkload.hpp b/src/backends/cl/workloads/ClRankWorkload.hpp
new file mode 100644
index 0000000..0a7bccf
--- /dev/null
+++ b/src/backends/cl/workloads/ClRankWorkload.hpp
@@ -0,0 +1,30 @@
+//
+// Copyright © 2020 Arm Ltd and Contributors. All rights reserved.
+// SPDX-License-Identifier: MIT
+//
+
+#pragma once
+
+#include <backendsCommon/Workload.hpp>
+#include <backendsCommon/WorkloadData.hpp>
+
+#include "ClWorkloadUtils.hpp"
+
+namespace armnn
+{
+
+struct ClRankWorkload : public BaseWorkload<RankQueueDescriptor>
+{
+public:
+ using BaseWorkload<RankQueueDescriptor>::BaseWorkload;
+ virtual void Execute() const override
+ {
+ const ClTensorHandle* clTensorHandle = PolymorphicDowncast<const ClTensorHandle*>(m_Data.m_Inputs[0]);
+ const int32_t rank = static_cast<int32_t>(clTensorHandle->GetShape().GetNumDimensions());
+
+ std::memcpy(GetOutputTensorData<void>(0, m_Data), &rank, sizeof(int32_t));
+ m_Data.m_Outputs[0]->Unmap();
+ }
+};
+
+} //namespace armnn
diff --git a/src/backends/cl/workloads/ClWorkloadUtils.hpp b/src/backends/cl/workloads/ClWorkloadUtils.hpp
index 89f13a5..b0cc8b4 100644
--- a/src/backends/cl/workloads/ClWorkloadUtils.hpp
+++ b/src/backends/cl/workloads/ClWorkloadUtils.hpp
@@ -143,4 +143,11 @@
}
}
+template <typename DataType, typename PayloadType>
+DataType* GetOutputTensorData(unsigned int idx, const PayloadType& data)
+{
+ ITensorHandle* tensorHandle = data.m_Outputs[idx];
+ return reinterpret_cast<DataType*>(tensorHandle->Map());
+}
+
} //namespace armnn
diff --git a/src/backends/cl/workloads/ClWorkloads.hpp b/src/backends/cl/workloads/ClWorkloads.hpp
index efcccb3..0045e7a 100644
--- a/src/backends/cl/workloads/ClWorkloads.hpp
+++ b/src/backends/cl/workloads/ClWorkloads.hpp
@@ -43,6 +43,7 @@
#include "ClQLstmWorkload.hpp"
#include "ClQuantizeWorkload.hpp"
#include "ClQuantizedLstmWorkload.hpp"
+#include "ClRankWorkload.hpp"
#include "ClReshapeWorkload.hpp"
#include "ClResizeWorkload.hpp"
#include "ClRsqrtWorkload.hpp"
diff --git a/src/backends/neon/workloads/NeonRankWorkload.hpp b/src/backends/neon/workloads/NeonRankWorkload.hpp
index f4e6959..e9c7dab 100644
--- a/src/backends/neon/workloads/NeonRankWorkload.hpp
+++ b/src/backends/neon/workloads/NeonRankWorkload.hpp
@@ -23,6 +23,7 @@
const int32_t rank = static_cast<int32_t>(neonTensorHandle->GetShape().GetNumDimensions());
std::memcpy(GetOutputTensorData<void>(0, m_Data), &rank, sizeof(int32_t));
+ m_Data.m_Outputs[0]->Unmap();
}
};
diff --git a/src/backends/reference/workloads/RefRankWorkload.hpp b/src/backends/reference/workloads/RefRankWorkload.hpp
index 780d3be..660db6b 100644
--- a/src/backends/reference/workloads/RefRankWorkload.hpp
+++ b/src/backends/reference/workloads/RefRankWorkload.hpp
@@ -22,6 +22,7 @@
const int32_t rank = static_cast<int32_t>(GetTensorInfo(m_Data.m_Inputs[0]).GetNumDimensions());
std::memcpy(GetOutputTensorData<void>(0, m_Data), &rank, sizeof(int32_t));
+ m_Data.m_Outputs[0]->Unmap();
}
};