COMPMID-2991: Add support for QASYMM8_SIGNED in CL kernels/functions - part 2
Adding support for QASYMM8_SIGNED to the following CL kernels/functions:
- CLActivationLayerKernel/CLActivationLayer
- CLComparisonKernel/CLComparison
- CLConvertFullyConnectedWeightsKernel/CLConvertFullyConnectedWeights
- CLDeconvolutionLayerUpsampleKernel/CLDeconvolutionLayerUpsample
- CLDepthToSpaceLayerKernel/CLDepthToSpaceLayer
- CLDequantizationLayerKernel/CLDequantizationLayer
- CLGEMMMatrixVectorMultiplyKernel
- CLNormalizePlanarYUVLayerKernel
- CLPReluLayer
- CLPixelWiseMultiplicationKernel/CLPixelWiseMultiplication
- CLPoolingLayerKernel/CLPoolingLayer
Change-Id: I874bbb7c2b08baa9c5ff4c9e6bc8778b42a6bec5
Signed-off-by: Michele Di Giorgio <michele.digiorgio@arm.com>
Reviewed-on: https://review.mlplatform.org/c/2539
Reviewed-by: Michalis Spyrou <michalis.spyrou@arm.com>
Comments-Addressed: Arm Jenkins <bsgcomp@arm.com>
Tested-by: Arm Jenkins <bsgcomp@arm.com>
diff --git a/arm_compute/runtime/CL/functions/CLActivationLayer.h b/arm_compute/runtime/CL/functions/CLActivationLayer.h
index f778148..09f5d2b 100644
--- a/arm_compute/runtime/CL/functions/CLActivationLayer.h
+++ b/arm_compute/runtime/CL/functions/CLActivationLayer.h
@@ -1,5 +1,5 @@
/*
- * Copyright (c) 2016-2019 ARM Limited.
+ * Copyright (c) 2016-2020 ARM Limited.
*
* SPDX-License-Identifier: MIT
*
@@ -57,7 +57,7 @@
* @note If the output tensor is a nullptr or is equal to the input, the activation function will be performed in-place
*
* @param[in, out] input Source tensor. In case of @p output tensor = nullptr, this tensor will store the result
- * of the activation function. Data types supported: QASYMM8/QSYMM16/F16/F32.
+ * of the activation function. Data types supported: QASYMM8/QASYMM8_SIGNED/QSYMM16/F16/F32.
* @param[out] output Destination tensor. Data type supported: same as @p input
* @param[in] act_info Activation layer parameters.
*/
@@ -65,7 +65,7 @@
/** Static function to check if given info will lead to a valid configuration of @ref CLActivationLayer
*
* @param[in] input Source tensor info. In case of @p output tensor info = nullptr, this tensor will store the result
- * of the activation function. Data types supported: QASYMM8/QSYMM16/F16/F32.
+ * of the activation function. Data types supported: QASYMM8/QASYMM8_SIGNED/QSYMM16/F16/F32.
* @param[in] output Destination tensor info. Data type supported: same as @p input
* @param[in] act_info Activation layer information.
*
diff --git a/arm_compute/runtime/CL/functions/CLComparison.h b/arm_compute/runtime/CL/functions/CLComparison.h
index 7f0b223..85dbe71 100644
--- a/arm_compute/runtime/CL/functions/CLComparison.h
+++ b/arm_compute/runtime/CL/functions/CLComparison.h
@@ -1,5 +1,5 @@
/*
- * Copyright (c) 2018-2019 ARM Limited.
+ * Copyright (c) 2018-2020 ARM Limited.
*
* SPDX-License-Identifier: MIT
*
@@ -38,7 +38,7 @@
public:
/** Initialise the kernel's inputs and outputs.
*
- * @param[in] input1 Source tensor. Data types supported: U8/S8/QASYMM8/U16/S16/U32/S32/F16/F32.
+ * @param[in] input1 Source tensor. Data types supported: All.
* The input1 tensor is [in, out] because its TensorInfo might be modified inside the kernel in case of broadcasting of dimension 0.
* @param[in] input2 Source tensor. Data types supported: Same as @p input1.
* The input2 tensor is [in, out] because its TensorInfo might be modified inside the kernel in case of broadcasting of dimension 0.
@@ -48,7 +48,7 @@
void configure(ICLTensor *input1, ICLTensor *input2, ICLTensor *output, ComparisonOperation operation);
/** Static function to check if given info will lead to a valid configuration of @ref CLComparison
*
- * @param[in] input1 Source tensor. Data types supported: U8/S8/QASYMM8/U16/S16/U32/S32/F16/F32.
+ * @param[in] input1 Source tensor. Data types supported: All.
* @param[in] input2 Source tensor. Data types supported: Same as @p input1.
* @param[in] output Destination tensor. Data types supported: U8.
* @param[out] operation Comparison operation to be used.
diff --git a/arm_compute/runtime/CL/functions/CLConvertFullyConnectedWeights.h b/arm_compute/runtime/CL/functions/CLConvertFullyConnectedWeights.h
index f0359ec..76a28ed 100644
--- a/arm_compute/runtime/CL/functions/CLConvertFullyConnectedWeights.h
+++ b/arm_compute/runtime/CL/functions/CLConvertFullyConnectedWeights.h
@@ -1,5 +1,5 @@
/*
- * Copyright (c) 2018-2019 ARM Limited.
+ * Copyright (c) 2018-2020 ARM Limited.
*
* SPDX-License-Identifier: MIT
*
@@ -39,7 +39,7 @@
public:
/** Initialize the function.
*
- * @param[in] input Source weights tensor to convert. Must be 2 dimensional. Data types supported: U8/S8/QASYMM8/U16/S16/U32/S32/F16/F32.
+ * @param[in] input Source weights tensor to convert. Must be 2 dimensional. Data types supported: All.
* @param[out] output The converted weights tensor. Shape and Data Type: Same as @p input.
* @param[in] original_input_shape Shape of the original input tensor (the one entering fully connected layer).
* @param[in] data_layout The data layout the weights have been trained in.
@@ -49,7 +49,7 @@
void configure(const ICLTensor *input, ICLTensor *output, const TensorShape &original_input_shape, DataLayout data_layout);
/** Static function to check if given info will lead to a valid configuration of @ref CLConvertFullyConnectedWeights
*
- * @param[in] input Source weights tensor info to convert. Must be 2 dimensional. Data types supported: U8/S8/QASYMM8/U16/S16/U32/S32/F16/F32.
+ * @param[in] input Source weights tensor info to convert. Must be 2 dimensional. Data types supported: All.
* @param[in] output The converted weights tensor info. Shape and Data Type: Same as @p input.
* @param[in] original_input_shape Shape of the original input tensor (the one entering fully connected layer).
* @param[in] data_layout The data layout the weights have been trained in.
@@ -90,7 +90,7 @@
}
/** Configures the @ref CLConvertFullyConnectedWeights function
*
- * @param[in] input Source weights tensor info to convert. Data type supported: U8/S8/QASYMM8/U16/S16/U32/S32/F16/F32.
+ * @param[in] input Source weights tensor info to convert. Data type supported: All.
* @param[in] original_input_shape Shape of the original input tensor (the one entering fully connected layer).
* @param[in] data_layout The data layout the weights have been trained in.
*/
diff --git a/arm_compute/runtime/CL/functions/CLDeconvolutionLayerUpsample.h b/arm_compute/runtime/CL/functions/CLDeconvolutionLayerUpsample.h
index 6f01574..5a1009c 100644
--- a/arm_compute/runtime/CL/functions/CLDeconvolutionLayerUpsample.h
+++ b/arm_compute/runtime/CL/functions/CLDeconvolutionLayerUpsample.h
@@ -1,5 +1,5 @@
/*
- * Copyright (c) 2017-2019 ARM Limited.
+ * Copyright (c) 2017-2020 ARM Limited.
*
* SPDX-License-Identifier: MIT
*
@@ -59,14 +59,14 @@
/** Initialize the function's source, destination, interpolation type and border_mode.
*
- * @param[in, out] input Source tensor. Data type supported: QASYMM8/F16/F32.
+ * @param[in, out] input Source tensor. Data type supported: All.
* @param[out] output Destination tensor. Data type supported: same as @p input.
* @param[in] info Contains padding and policies to be used in the deconvolution.
*/
void configure(ICLTensor *input, ICLTensor *output, const PadStrideInfo &info);
/** Static function to check if given info will lead to a valid configuration of @ref CLDeconvolutionLayerUpsample
*
- * @param[in] input Source tensor info. Data type supported: QASYMM8/F16/F32.
+ * @param[in] input Source tensor info. Data type supported: All.
* @param[in] output Destination tensor info. Data type supported: same as @p input.
* @param[in] info Contains padding and policies to be used in the deconvolution.
*
diff --git a/arm_compute/runtime/CL/functions/CLDepthToSpaceLayer.h b/arm_compute/runtime/CL/functions/CLDepthToSpaceLayer.h
index ddee04a..0c33ed3 100644
--- a/arm_compute/runtime/CL/functions/CLDepthToSpaceLayer.h
+++ b/arm_compute/runtime/CL/functions/CLDepthToSpaceLayer.h
@@ -1,5 +1,5 @@
/*
- * Copyright (c) 2019 ARM Limited.
+ * Copyright (c) 2019-2020 ARM Limited.
*
* SPDX-License-Identifier: MIT
*
@@ -24,43 +24,33 @@
#ifndef ARM_COMPUTE_CLDEPTHTOSPACELAYER_H
#define ARM_COMPUTE_CLDEPTHTOSPACELAYER_H
-#include "arm_compute/runtime/IFunction.h"
-
-#include "arm_compute/core/CL/kernels/CLDepthToSpaceLayerKernel.h"
#include "arm_compute/core/Types.h"
+#include "arm_compute/runtime/CL/ICLSimpleFunction.h"
namespace arm_compute
{
class ICLTensor;
/** Basic function to run @ref CLDepthToSpaceLayerKernel. */
-class CLDepthToSpaceLayer : public IFunction
+class CLDepthToSpaceLayer : public ICLSimpleFunction
{
public:
- /** Default constructor */
- CLDepthToSpaceLayer();
/** Set the input and output tensors.
*
- * @param[in] input Tensor input. Supported tensor rank: 4. Data types supported: U8/S8/QASYMM8/U16/S16/F16/U32/S32/F32.
+ * @param[in] input Tensor input. Supported tensor rank: 4. Data types supported: All.
* @param[out] output Tensor output. Data types supported: same as @p input
* @param[in] block_shape Block shape value.
*/
void configure(const ICLTensor *input, ICLTensor *output, int32_t block_shape);
/** Static function to check if given info will lead to a valid configuration of @ref CLDepthToSpaceLayer.
*
- * @param[in] input Tensor input info. Supported tensor rank: 4. Data types supported: U8/S8/QASYMM8/U16/S16/F16/U32/S32/F32.
+ * @param[in] input Tensor input info. Supported tensor rank: 4. Data types supported: All.
* @param[in] output Tensor output info. Data types supported: same as @p input
* @param[in] block_shape Block shape value.
*
* @return a status
*/
static Status validate(const ITensorInfo *input, const ITensorInfo *output, int32_t block_shape);
-
- // Inherited methods overridden:
- void run() override;
-
-private:
- CLDepthToSpaceLayerKernel _depth_to_space_kernel; /**< CLDepthToSpaceLayerKernel to run */
};
}
#endif /* ARM_COMPUTE_CLDEPTHTOSPACELAYER_H */
diff --git a/arm_compute/runtime/CL/functions/CLDequantizationLayer.h b/arm_compute/runtime/CL/functions/CLDequantizationLayer.h
index 308349a..48d6ba8 100644
--- a/arm_compute/runtime/CL/functions/CLDequantizationLayer.h
+++ b/arm_compute/runtime/CL/functions/CLDequantizationLayer.h
@@ -1,5 +1,5 @@
/*
- * Copyright (c) 2017-2019 ARM Limited.
+ * Copyright (c) 2017-2020 ARM Limited.
*
* SPDX-License-Identifier: MIT
*
@@ -40,13 +40,13 @@
/** Set the input and output tensors.
*
* @param[in] input Source tensor with at least 3 dimensions. The dimensions over the third will be interpreted as batches.
- * Data types supported: QASYMM8/QSYMM8_PER_CHANNEL/QSYMM8/QSYMM16.
+ * Data types supported: QASYMM8/QASYMM8_SIGNED/QSYMM8_PER_CHANNEL/QSYMM8/QSYMM16.
* @param[out] output Destination tensor with the same dimensions of input. Data type supported: F16/F32.
*/
void configure(const ICLTensor *input, ICLTensor *output);
/** Static function to check if given info will lead to a valid configuration of @ref CLDequantizationLayer
*
- * @param[in] input Input tensor info. Data types supported: QASYMM8/QSYMM8_PER_CHANNEL/QSYMM8/QSYMM16.
+ * @param[in] input Input tensor info. Data types supported: QASYMM8/QASYMM8_SIGNED/QSYMM8_PER_CHANNEL/QSYMM8/QSYMM16.
* @param[in] output Output tensor info. Data type supported: F16/F32.
*
* @return a status
diff --git a/arm_compute/runtime/CL/functions/CLNormalizePlanarYUVLayer.h b/arm_compute/runtime/CL/functions/CLNormalizePlanarYUVLayer.h
index 4fe5a11..5fbfdd1 100644
--- a/arm_compute/runtime/CL/functions/CLNormalizePlanarYUVLayer.h
+++ b/arm_compute/runtime/CL/functions/CLNormalizePlanarYUVLayer.h
@@ -1,5 +1,5 @@
/*
- * Copyright (c) 2018-2019 ARM Limited.
+ * Copyright (c) 2018-2020 ARM Limited.
*
* SPDX-License-Identifier: MIT
*
@@ -24,10 +24,10 @@
#ifndef ARM_COMPUTE_CLNORMALIZEPLANARYUVLAYER_H
#define ARM_COMPUTE_CLNORMALIZEPLANARYUVLAYER_H
-#include "arm_compute/runtime/IFunction.h"
-
-#include "arm_compute/core/CL/kernels/CLNormalizePlanarYUVLayerKernel.h"
#include "arm_compute/core/Types.h"
+#include "arm_compute/runtime/CL/ICLSimpleFunction.h"
+
+#include <cstdint>
namespace arm_compute
{
@@ -37,15 +37,13 @@
*
* @note The function simulates a NormalizePlanarYUV layer.
*/
-class CLNormalizePlanarYUVLayer : public IFunction
+class CLNormalizePlanarYUVLayer : public ICLSimpleFunction
{
public:
- /** Default constructor */
- CLNormalizePlanarYUVLayer();
/** Set the input and output tensors.
*
* @param[in] input Source tensor. 3 lower dimensions represent a single input with dimensions [width, height, channels].
- * Data types supported: QASYMM8/F16/F32.
+ * Data types supported: QASYMM8/QASYMM8_SIGNED/F16/F32.
* @param[out] output Destinationfeature tensor. Data type supported: same as @p input
* @param[in] mean Mean values tensor. 1 dimension with size equal to the number of input channels. Data types supported: Same as @p input
* @param[in] std Standard deviation values tensor. 1 dimension with size equal to the number of input channels.
@@ -55,7 +53,7 @@
/** Static function to check if given info will lead to a valid configuration of @ref CLNormalizePlanarYUVLayer
*
* @param[in] input Source tensor info. 3 lower dimensions represent a single input with dimensions [width, height, channels].
- * Data types supported: QASYMM8/F16/F32.
+ * Data types supported: QASYMM8/QASYMM8_SIGNED/F16/F32.
* @param[out] output Destination tensor info. Data type supported: same as @p input
* @param[in] mean Mean values tensor info. 1 dimension with size equal to the number of input channels. Data types supported: Same as @p input
* @param[in] std Standard deviation values tensor info. 1 dimension with size equal to the number of input channels.
@@ -64,12 +62,6 @@
* @return a status
*/
static Status validate(const ITensorInfo *input, const ITensorInfo *output, const ITensorInfo *mean, const ITensorInfo *std);
-
- // Inherited methods overridden:
- void run() override;
-
-private:
- CLNormalizePlanarYUVLayerKernel _norm_kernel; /**< NormalizePlanarYUV layer kernel to run */
};
} // namespace arm_compute
#endif /* ARM_COMPUTE_CLNORMALIZEPLANARYUVLAYER_H */
diff --git a/arm_compute/runtime/CL/functions/CLPReluLayer.h b/arm_compute/runtime/CL/functions/CLPReluLayer.h
index 42876cd..7f8a412 100644
--- a/arm_compute/runtime/CL/functions/CLPReluLayer.h
+++ b/arm_compute/runtime/CL/functions/CLPReluLayer.h
@@ -1,5 +1,5 @@
/*
- * Copyright (c) 2019 ARM Limited.
+ * Copyright (c) 2019-2020 ARM Limited.
*
* SPDX-License-Identifier: MIT
*
@@ -42,14 +42,14 @@
*
* @note If the output tensor is a nullptr or is equal to the input, the activation function will be performed in-place
*
- * @param[in] input Source tensor. Data types supported: QASYMM8/F16/F32.
+ * @param[in] input Source tensor. Data types supported: QASYMM8/QASYMM8_SIGNED/F16/F32.
* @param[in] alpha PRelu layer parameters. Data types supported: same of @p input.
* @param[out] output Destination tensor. Data type supported: same as @p input
*/
void configure(ICLTensor *input, ICLTensor *alpha, ICLTensor *output);
/** Static function to check if given info will lead to a valid configuration of @ref CLPReluLayer
*
- * @param[in] input Source tensor info. Data types supported: QASYMM8/F16/F32.
+ * @param[in] input Source tensor info. Data types supported: QASYMM8/QASYMM8_SIGNED/F16/F32.
* @param[in] alpha PRelu layer parameters. Data types supported: same of @p input.
* @param[in] output Destination tensor info. Data type supported: same as @p input
*
diff --git a/arm_compute/runtime/CL/functions/CLPixelWiseMultiplication.h b/arm_compute/runtime/CL/functions/CLPixelWiseMultiplication.h
index fd64d7b..72b1587 100644
--- a/arm_compute/runtime/CL/functions/CLPixelWiseMultiplication.h
+++ b/arm_compute/runtime/CL/functions/CLPixelWiseMultiplication.h
@@ -1,5 +1,5 @@
/*
- * Copyright (c) 2016-2019 ARM Limited.
+ * Copyright (c) 2016-2020 ARM Limited.
*
* SPDX-License-Identifier: MIT
*
@@ -38,7 +38,7 @@
public:
/** Initialise the kernel's inputs, output and convertion policy.
*
- * @param[in, out] input1 An input tensor. Data types supported: U8/QASYMM8/S16/QSYMM16/F16/F32.
+ * @param[in, out] input1 An input tensor. Data types supported: U8/QASYMM8/QASYMM8_SIGNED/S16/QSYMM16/F16/F32.
* The input tensor is [in, out] because its TensorInfo might be modified inside the kernel in case of broadcasting of dimension 0.
* @param[in, out] input2 An input tensor. Data types supported: same as @p input1.
* The input tensor is [in, out] because its TensorInfo might be modified inside the kernel in case of broadcasting of dimension 0.
@@ -52,7 +52,7 @@
ConvertPolicy overflow_policy, RoundingPolicy rounding_policy);
/** Static function to check if given info will lead to a valid configuration of @ref CLPixelWiseMultiplication
*
- * @param[in] input1 An input tensor info. Data types supported: U8/QASYMM8/S16/QSYMM16/F16/F32.
+ * @param[in] input1 An input tensor info. Data types supported: U8/QASYMM8/QASYMM8_SIGNED/S16/QSYMM16/F16/F32.
* @param[in] input2 An input tensor info. Data types supported: same as @p input1.
* @param[in] output The output tensor info, Data types supported: same as @p input1. Note: U8 requires both inputs to be U8.
* @param[in] scale Scale to apply after multiplication.
diff --git a/arm_compute/runtime/CL/functions/CLPoolingLayer.h b/arm_compute/runtime/CL/functions/CLPoolingLayer.h
index 19acb7f..c78b558 100644
--- a/arm_compute/runtime/CL/functions/CLPoolingLayer.h
+++ b/arm_compute/runtime/CL/functions/CLPoolingLayer.h
@@ -1,5 +1,5 @@
/*
- * Copyright (c) 2017-2019 ARM Limited.
+ * Copyright (c) 2017-2020 ARM Limited.
*
* SPDX-License-Identifier: MIT
*
@@ -43,14 +43,14 @@
public:
/** Set the input and output tensors.
*
- * @param[in,out] input Source tensor. (Written to only when padding != 0) Data types supported: QASYMM8/F16/F32.
+ * @param[in,out] input Source tensor. (Written to only when padding != 0) Data types supported: QASYMM8/QASYMM8_SIGNED/F16/F32.
* @param[out] output Destination tensor. Data types supported: Same as @p input.
* @param[in] pool_info Contains pooling operation information described in @ref PoolingLayerInfo.
*/
void configure(ICLTensor *input, ICLTensor *output, const PoolingLayerInfo &pool_info);
/** Static function to check if given info will lead to a valid configuration of @ref CLPoolingLayer
*
- * @param[in] input Source tensor info. Data types supported: QASYMM8/F16/F32.
+ * @param[in] input Source tensor info. Data types supported: QASYMM8/QASYMM8_SIGNED/F16/F32.
* @param[in] output Destination tensor info. Data types supported: Same as @p input.
* @param[in] pool_info Contains pooling operation information described in @ref PoolingLayerInfo.
*