Port DepthConvert to new Api
- Renames DepthConvert to Cast
- Ports both NEDepthConverLayer and CLDepthConvert variants
- Removes legacy shift capability from DepthConvert, allowing only
shifts of 0
Signed-off-by: Georgios Pinitas <georgios.pinitas@arm.com>
Change-Id: I806a0f8eb23d23502b632c529fda7edde19c8176
Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/5565
Tested-by: Arm Jenkins <bsgcomp@arm.com>
Reviewed-by: Michele Di Giorgio <michele.digiorgio@arm.com>
Comments-Addressed: Arm Jenkins <bsgcomp@arm.com>
diff --git a/src/runtime/NEON/functions/NECast.cpp b/src/runtime/NEON/functions/NECast.cpp
index a42f512..b519576 100644
--- a/src/runtime/NEON/functions/NECast.cpp
+++ b/src/runtime/NEON/functions/NECast.cpp
@@ -1,5 +1,5 @@
/*
- * Copyright (c) 2019-2020 Arm Limited.
+ * Copyright (c) 2019-2021 Arm Limited.
*
* SPDX-License-Identifier: MIT
*
@@ -23,23 +23,45 @@
*/
#include "arm_compute/runtime/NEON/functions/NECast.h"
-#include "arm_compute/core/ITensor.h"
-#include "arm_compute/core/TensorInfo.h"
-#include "src/core/NEON/kernels/NEDepthConvertLayerKernel.h"
-
-#include <utility>
+#include "arm_compute/core/Validate.h"
+#include "src/runtime/cpu/operators/CpuCast.h"
namespace arm_compute
{
+struct NECast::Impl
+{
+ const ITensor *src{ nullptr };
+ ITensor *dst{ nullptr };
+ std::unique_ptr<cpu::CpuCast> op{ nullptr };
+};
+
+NECast::NECast()
+ : _impl(std::make_unique<Impl>())
+{
+}
+NECast::NECast(NECast &&) = default;
+NECast &NECast::operator=(NECast &&) = default;
+NECast::~NECast() = default;
+
void NECast::configure(ITensor *input, ITensor *output, ConvertPolicy policy)
{
- auto k = std::make_unique<NEDepthConvertLayerKernel>();
- k->configure(input, output, policy, 0);
- _kernel = std::move(k);
+ _impl->src = input;
+ _impl->dst = output;
+
+ ARM_COMPUTE_ERROR_ON_NULLPTR(_impl->src, _impl->dst);
+
+ _impl->op = std::make_unique<cpu::CpuCast>();
+ _impl->op->configure(_impl->src->info(), _impl->dst->info(), policy);
}
Status NECast::validate(ITensorInfo *input, ITensorInfo *output, ConvertPolicy policy)
{
- return NEDepthConvertLayerKernel::validate(input, output, policy, 0);
+ return cpu::CpuCast::validate(input, output, policy);
+}
+
+void NECast::run()
+{
+ ITensorPack pack = { { ACL_SRC, _impl->src }, { ACL_DST, _impl->dst } };
+ _impl->op->run(pack);
}
} // namespace arm_compute
diff --git a/src/runtime/NEON/functions/NEDepthConvertLayer.cpp b/src/runtime/NEON/functions/NEDepthConvertLayer.cpp
index 761de8e..07e985c 100644
--- a/src/runtime/NEON/functions/NEDepthConvertLayer.cpp
+++ b/src/runtime/NEON/functions/NEDepthConvertLayer.cpp
@@ -1,5 +1,5 @@
/*
- * Copyright (c) 2016-2020 Arm Limited.
+ * Copyright (c) 2016-2021 Arm Limited.
*
* SPDX-License-Identifier: MIT
*
@@ -23,20 +23,51 @@
*/
#include "arm_compute/runtime/NEON/functions/NEDepthConvertLayer.h"
-#include "src/core/NEON/kernels/NEDepthConvertLayerKernel.h"
+#include "arm_compute/core/Validate.h"
+#include "src/runtime/cpu/operators/CpuCast.h"
#include <utility>
-using namespace arm_compute;
+namespace arm_compute
+{
+struct NEDepthConvertLayer::Impl
+{
+ const ITensor *src{ nullptr };
+ ITensor *dst{ nullptr };
+ std::unique_ptr<cpu::CpuCast> op{ nullptr };
+};
+
+NEDepthConvertLayer::NEDepthConvertLayer()
+ : _impl(std::make_unique<Impl>())
+{
+}
+NEDepthConvertLayer::NEDepthConvertLayer(NEDepthConvertLayer &&) = default;
+NEDepthConvertLayer &NEDepthConvertLayer::operator=(NEDepthConvertLayer &&) = default;
+NEDepthConvertLayer::~NEDepthConvertLayer() = default;
void NEDepthConvertLayer::configure(const ITensor *input, ITensor *output, ConvertPolicy policy, uint32_t shift)
{
- auto k = std::make_unique<NEDepthConvertLayerKernel>();
- k->configure(input, output, policy, shift);
- _kernel = std::move(k);
+ ARM_COMPUTE_UNUSED(shift);
+
+ _impl->src = input;
+ _impl->dst = output;
+
+ ARM_COMPUTE_ERROR_ON_NULLPTR(_impl->src, _impl->dst);
+ ARM_COMPUTE_ERROR_ON(shift != 0);
+
+ _impl->op = std::make_unique<cpu::CpuCast>();
+ _impl->op->configure(_impl->src->info(), _impl->dst->info(), policy);
}
Status NEDepthConvertLayer::validate(const ITensorInfo *input, const ITensorInfo *output, ConvertPolicy policy, uint32_t shift)
{
- return NEDepthConvertLayerKernel::validate(input, output, policy, shift);
+ ARM_COMPUTE_RETURN_ERROR_ON(shift != 0);
+ return cpu::CpuCast::validate(input, output, policy);
}
+
+void NEDepthConvertLayer::run()
+{
+ ITensorPack pack = { { ACL_SRC, _impl->src }, { ACL_DST, _impl->dst } };
+ _impl->op->run(pack);
+}
+} // namespace arm_compute