Update operator list documentation. Part 2.
All data type and data layout information for the operators are store in the function header files
Signed-off-by: Teresa Charlin <teresa.charlinreyes@arm.com>
Change-Id: I30b564f7eda6bbd99bf3ad36ddb6639ac118eb8b
Reviewed-on: https://eu-gerrit-1.euhpc.arm.com/c/VisualCompute/ComputeLibrary/+/319829
Tested-by: bsgcomp <bsgcomp@arm.com>
Reviewed-by: Michele DiGiorgio <michele.digiorgio@arm.com>
Comments-Addressed: bsgcomp <bsgcomp@arm.com>
Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/5531
Tested-by: Arm Jenkins <bsgcomp@arm.com>
Reviewed-by: SiCong Li <sicong.li@arm.com>
Comments-Addressed: Arm Jenkins <bsgcomp@arm.com>
diff --git a/arm_compute/runtime/CL/functions/CLArgMinMaxLayer.h b/arm_compute/runtime/CL/functions/CLArgMinMaxLayer.h
index c254284..a971163 100644
--- a/arm_compute/runtime/CL/functions/CLArgMinMaxLayer.h
+++ b/arm_compute/runtime/CL/functions/CLArgMinMaxLayer.h
@@ -1,5 +1,5 @@
/*
- * Copyright (c) 2018-2020 Arm Limited.
+ * Copyright (c) 2018-2021 Arm Limited.
*
* SPDX-License-Identifier: MIT
*
@@ -65,6 +65,18 @@
~CLArgMinMaxLayer();
/** Set the input and output tensors.
*
+ * Valid data layouts:
+ * - All
+ *
+ * Valid data type configurations:
+ * |src |dst |
+ * |:--------------|:---------|
+ * |QASYMM8 |U32, S32 |
+ * |QASYMM8_SIGNED |U32, S32 |
+ * |S32 |U32, S32 |
+ * |F16 |U32, S32 |
+ * |F32 |U32, S32 |
+ *
* @param[in] input Input source tensor. Data types supported: QASYMM8/QASYMM8_SIGNED/S32/F16/F32.
* @param[in] axis Axis to find max/min index.
* @param[out] output Output source tensor. Data types supported: U32/S32.
diff --git a/arm_compute/runtime/CL/functions/CLBatchNormalizationLayer.h b/arm_compute/runtime/CL/functions/CLBatchNormalizationLayer.h
index c8acf9f..fcfeb5e 100644
--- a/arm_compute/runtime/CL/functions/CLBatchNormalizationLayer.h
+++ b/arm_compute/runtime/CL/functions/CLBatchNormalizationLayer.h
@@ -1,5 +1,5 @@
/*
- * Copyright (c) 2017-2020 Arm Limited.
+ * Copyright (c) 2017-2021 Arm Limited.
*
* SPDX-License-Identifier: MIT
*
@@ -60,6 +60,16 @@
~CLBatchNormalizationLayer();
/** Set the input and output tensors.
*
+ * Valid data layouts:
+ * - NHWC
+ * - NCHW
+ *
+ * Valid data type configurations:
+ * |src |dst |
+ * |:--------------|:--------------|
+ * |F32 |F32 |
+ * |F16 |F16 |
+ *
* @note If the output tensor is a nullptr or is equal to the input, the batch normalization 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.
diff --git a/arm_compute/runtime/CL/functions/CLBatchToSpaceLayer.h b/arm_compute/runtime/CL/functions/CLBatchToSpaceLayer.h
index bdb5853..f6ba2b0 100644
--- a/arm_compute/runtime/CL/functions/CLBatchToSpaceLayer.h
+++ b/arm_compute/runtime/CL/functions/CLBatchToSpaceLayer.h
@@ -1,5 +1,5 @@
/*
- * Copyright (c) 2018-2020 Arm Limited.
+ * Copyright (c) 2018-2021 Arm Limited.
*
* SPDX-License-Identifier: MIT
*
@@ -55,6 +55,15 @@
~CLBatchToSpaceLayer();
/** Set the input and output tensors.
*
+ * Valid data layouts:
+ * - NHWC
+ * - NCHW
+ *
+ * Valid data type configurations:
+ * |src0 |src1 |dst |
+ * |:---------|:---------|:----------|
+ * |All |s32 |All |
+ *
* @param[in] input Tensor input. Supported tensor rank: 4. Data types supported: All.
* @param[in] block_shape 1-D tensor with shape [M]. Data types supported: S32
* @param[out] output Tensor output. Data types supported: same as @p input
diff --git a/arm_compute/runtime/CL/functions/CLBitwiseAnd.h b/arm_compute/runtime/CL/functions/CLBitwiseAnd.h
index a703242..b30be9b 100644
--- a/arm_compute/runtime/CL/functions/CLBitwiseAnd.h
+++ b/arm_compute/runtime/CL/functions/CLBitwiseAnd.h
@@ -41,6 +41,14 @@
public:
/** Initialize the function
*
+ * Valid data layouts:
+ * - All
+ *
+ * Valid data type configurations:
+ * |src |dst |
+ * |:--------------|:--------------|
+ * |U8 |U8 |
+ *
* @param[in] input1 Input tensor. Data types supported: U8.
* @param[in] input2 Input tensor. Data types supported: U8.
* @param[out] output Output tensor. Data types supported: U8.
diff --git a/arm_compute/runtime/CL/functions/CLBitwiseNot.h b/arm_compute/runtime/CL/functions/CLBitwiseNot.h
index 6f65749..1456ebe 100644
--- a/arm_compute/runtime/CL/functions/CLBitwiseNot.h
+++ b/arm_compute/runtime/CL/functions/CLBitwiseNot.h
@@ -40,6 +40,13 @@
{
public:
/** Initialize the function
+ * Valid data layouts:
+ * - All
+ *
+ * Valid data type configurations:
+ * |src |dst |
+ * |:--------------|:--------------|
+ * |U8 |U8 |
*
* @param[in] input Input tensor. Data types supported: U8.
* @param[out] output Output tensor. Data types supported: U8.
diff --git a/arm_compute/runtime/CL/functions/CLBitwiseOr.h b/arm_compute/runtime/CL/functions/CLBitwiseOr.h
index 3c904fb..ff0a1f0 100644
--- a/arm_compute/runtime/CL/functions/CLBitwiseOr.h
+++ b/arm_compute/runtime/CL/functions/CLBitwiseOr.h
@@ -41,6 +41,14 @@
public:
/** Initialize the function
*
+ * Valid data layouts:
+ * - All
+ *
+ * Valid data type configurations:
+ * |src |dst |
+ * |:--------------|:--------------|
+ * |U8 |U8 |
+ *
* @param[in] input1 Input tensor. Data types supported: U8.
* @param[in] input2 Input tensor. Data types supported: U8.
* @param[out] output Output tensor. Data types supported: U8.
diff --git a/arm_compute/runtime/CL/functions/CLBitwiseXor.h b/arm_compute/runtime/CL/functions/CLBitwiseXor.h
index a33a64a..0cd9d07 100644
--- a/arm_compute/runtime/CL/functions/CLBitwiseXor.h
+++ b/arm_compute/runtime/CL/functions/CLBitwiseXor.h
@@ -41,6 +41,14 @@
public:
/** Initialize the function
*
+ * Valid data layouts:
+ * - All
+ *
+ * Valid data type configurations:
+ * |src |dst |
+ * |:--------------|:--------------|
+ * |U8 |U8 |
+ *
* @param[in] input1 Input tensor. Data types supported: U8.
* @param[in] input2 Input tensor. Data types supported: U8.
* @param[out] output Output tensor. Data types supported: U8.
diff --git a/arm_compute/runtime/CL/functions/CLBoundingBoxTransform.h b/arm_compute/runtime/CL/functions/CLBoundingBoxTransform.h
index d640910..d3499c3 100644
--- a/arm_compute/runtime/CL/functions/CLBoundingBoxTransform.h
+++ b/arm_compute/runtime/CL/functions/CLBoundingBoxTransform.h
@@ -1,5 +1,5 @@
/*
- * Copyright (c) 2018-2020 Arm Limited.
+ * Copyright (c) 2018-2021 Arm Limited.
*
* SPDX-License-Identifier: MIT
*
@@ -45,6 +45,17 @@
public:
/** Set the input and output tensors.
*
+ * Valid data layouts:
+ * - NHWC
+ * - NCHW
+ *
+ * Valid data type configurations:
+ * |src0 |src1 |dst |
+ * |:--------------|:--------------|:--------------|
+ * |QASYMM16 |QASYMM8 |QASYMM16 |
+ * |F16 |F16 |F16 |
+ * |F32 |F32 |F32 |
+ *
* @param[in] boxes Source tensor. Bounding box proposals in pixel coordinates. Size(M, 4), format [x1, y1, x2, y2]. Data types supported: QASYMM16/F16/F32.
* @param[out] pred_boxes Destination tensor. Pixel coordinates of the transformed bounding boxes. Size (M, 4*K), format [x1, y1, x2, y2]. Data types supported: Same as @p input
* @param[in] deltas Bounding box translations and scales. Size (M, 4*K), format [dx, dy, dw, dh], K is the number of classes.
diff --git a/arm_compute/runtime/CL/functions/CLCast.h b/arm_compute/runtime/CL/functions/CLCast.h
index bd333d4..6e4cf62 100644
--- a/arm_compute/runtime/CL/functions/CLCast.h
+++ b/arm_compute/runtime/CL/functions/CLCast.h
@@ -1,5 +1,5 @@
/*
- * Copyright (c) 2018-2020 Arm Limited.
+ * Copyright (c) 2018-2021 Arm Limited.
*
* SPDX-License-Identifier: MIT
*
@@ -41,18 +41,22 @@
public:
/** Initialize the function's source, destination
*
+ * Valid data layouts:
+ * - All
+ *
+ * Valid data type configurations:
+ * |src |dst |
+ * |:--------------|:--------------------------------------|
+ * |U8 | S8, U16, S16, U32, S32, F16, F32 |
+ * |U16 | U8, S8, S16, U32, S32, F16, F32 |
+ * |S16 | U8, S8, U16, U32, S32, F16, F32 |
+ * |U32 | U8, S8, U16, S16, S32, F16, F32 |
+ * |S32 | U8, S8, U16, S16, U32, F16, F32 |
+ * |F16 | U8, S8, U16, S16, U32, F32 |
+ * |F32 | U8, S8, U16, S16, U32, F16 |
+ *
* Input data type must be different than output data type.
*
- * Valid conversions Input -> Output :
- *
- * - U8 -> S8, U16, S16, U32, S32, F16, F32
- * - U16 -> U8, S8, S16, U32, S32, F16, F32
- * - S16 -> U8, S8, U16, U32, S32, F16, F32
- * - U32 -> U8, S8, U16, S16, S32, F16, F32
- * - S32 -> U8, S8, U16, S16, U32, F16, F32
- * - F16 -> U8, S8, U16, S16, U32, F32
- * - F32 -> U8, S8, U16, S16, U32, F16
- *
* @param[in] input The input tensor to convert. Data types supported: U8/S8/U16/S16/U32/S32/F16/F32.
* @param[out] output The output tensor. Data types supported: U8/S8/U16/S16/U32/S32/F16/F32.
* @param[in] policy Conversion policy.
diff --git a/arm_compute/runtime/CL/functions/CLChannelShuffleLayer.h b/arm_compute/runtime/CL/functions/CLChannelShuffleLayer.h
index 54cf59f..d60548d 100644
--- a/arm_compute/runtime/CL/functions/CLChannelShuffleLayer.h
+++ b/arm_compute/runtime/CL/functions/CLChannelShuffleLayer.h
@@ -1,5 +1,5 @@
/*
- * Copyright (c) 2018-2020 Arm Limited.
+ * Copyright (c) 2018-2021 Arm Limited.
*
* SPDX-License-Identifier: MIT
*
@@ -44,6 +44,14 @@
public:
/** Initialize the function
*
+ * Valid data layouts:
+ * - NCHW
+ *
+ * Valid data type configurations:
+ * |src |dst |
+ * |:--------------|:--------------|
+ * |All |All |
+ *
* @param[in] input Input tensor. Data types supported: All.
* @param[out] output Output tensor. Data type supported: Same as @p input
* @param[in] num_groups Number of groups. Must be greater than 1 and the number of channels of the tensors must be a multiple of the number of groups.
diff --git a/arm_compute/runtime/CL/functions/CLConvolutionLayer.h b/arm_compute/runtime/CL/functions/CLConvolutionLayer.h
index d1de721..6884754 100644
--- a/arm_compute/runtime/CL/functions/CLConvolutionLayer.h
+++ b/arm_compute/runtime/CL/functions/CLConvolutionLayer.h
@@ -1,5 +1,5 @@
/*
- * Copyright (c) 2017-2020 Arm Limited.
+ * Copyright (c) 2017-2021 Arm Limited.
*
* SPDX-License-Identifier: MIT
*
@@ -85,13 +85,27 @@
CLConvolutionLayer &operator=(CLConvolutionLayer &&) = default;
/** Set the input and output tensors.
*
+ * Valid data layouts:
+ * - NHWC
+ * - NCHW
+ *
+ * Valid data type configurations:
+ * |src0 |src1 |src2 |dst |
+ * |:--------------|:------------------|:------|:--------------|
+ * |F16 |F16 |F16 |F16 |
+ * |F32 |F32 |F32 |F32 |
+ * |QASYMM8 |QASYMM8 |S32 |QASYMM8 |
+ * |QASYMM8 |QSYMM8_PER_CHANNEL |S32 |QASYMM8 |
+ * |QASYMM8_SIGNED |QASYMM8_SIGNED |S32 |QASYMM8_SIGNED |
+ * |QASYMM8_SIGNED |QSYMM8_PER_CHANNEL |S32 |QASYMM8_SIGNED |
+ *
* @param[in] input Source tensor. 3 lower dimensions represent a single input [width, height, IFM],
* while every optional dimension from 4 and above represent a batch of inputs.
* Data types supported: QASYMM8/QASYMM8_SIGNED/F16/F32.
* @param[in] weights Weights tensor. Weights are 4D tensor with dimensions [kernel_x, kernel_y, IFM, OFM].
- * Data type supported: Same as @p input or QASYMM8/QSYMM8_PER_CHANNEL when @p input is QASYMM8.
+ * Data type supported: Same as @p input, also could be QSYMM8_PER_CHANNEL if input is QASYMM8/QASYMM8_SIGNED.
* @param[in] biases Biases tensor. Shared biases supported. Biases are 1D tensor with dimensions [OFM].
- * Data type supported: Should match @p input data type, except for input of QASYMM8 type where biases should be of S32 type.
+ * Data type supported: Same as @p input, except for input of QASYMM8/QASYMM8_SIGNED type where biases should be of S32 type.
* @param[out] output Destination tensor. 3 lower dimensions represent a single output [width, height, OFM], while the rest represent batch of outputs.
* Data types supported: Same as @p input.
* @param[in] conv_info Contains padding and stride information described in @ref PadStrideInfo.
@@ -111,9 +125,9 @@
* while every optional dimension from 4 and above represent a batch of inputs.
* Data types supported: QASYMM8/QASYMM8_SIGNED/F16/F32.
* @param[in] weights Weights tensor. Weights are 4D tensor with dimensions [kernel_x, kernel_y, IFM, OFM].
- * Data type supported: Same as @p input or QASYMM8/QSYMM8_PER_CHANNEL when @p input is QASYMM8.
+ * Data type supported: Same as @p input, also could be QSYMM8_PER_CHANNEL if input is QASYMM8/QASYMM8_SIGNED.
* @param[in] biases Biases tensor. Shared biases supported. Biases are 1D tensor with dimensions [OFM].
- * Data type supported: Should match @p input data type, except for input of QASYMM8 type where biases should be of S32 type.
+ * Data type supported: Same as @p input, except for input of QASYMM8/QASYMM8_SIGNED type where biases should be of S32 type.
* @param[out] output Destination tensor. 3 lower dimensions represent a single output [width, height, OFM], while the rest represent batch of outputs.
* Data types supported: Same as @p input.
* @param[in] conv_info Contains padding and stride information described in @ref PadStrideInfo.
@@ -133,8 +147,9 @@
* while every optional dimension from 4 and above represent a batch of inputs.
* Data types supported: QASYMM8/QASYMM8_SIGNED/F16/F32.
* @param[in] weights Weights tensor. Weights are 4D tensor with dimensions [kernel_x, kernel_y, IFM, OFM].
- * Data type supported: Same as @p input or QASYMM8/QSYMM8_PER_CHANNEL when @p input is QASYMM8.
- * @param[in] biases Biases tensor. Shared biases supported. Biases are 1D tensor with dimensions [OFM]. Data type supported:Same as @p input.
+ * Data type supported: Same as @p input, also could be QSYMM8_PER_CHANNEL if input is QASYMM8/QASYMM8_SIGNED.
+ * @param[in] biases Biases tensor. Shared biases supported. Biases are 1D tensor with dimensions [OFM].
+ * Data type supported: Same as @p input, except for input of QASYMM8/QASYMM8_SIGNED type where biases should be of S32 type.
* @param[in] output Destination tensor. 3 lower dimensions represent a single output [width, height, OFM], while the rest represent batch of outputs.
* Data types supported: Same as @p input.
* @param[in] conv_info Contains padding and stride information described in @ref PadStrideInfo.
@@ -156,7 +171,7 @@
* while every optional dimension from 4 and above represent a batch of inputs.
* Data types supported: QASYMM8/QASYMM8_SIGNED/F16/F32.
* @param[in] weights Weights tensor. Weights are 4D tensor with dimensions [kernel_x, kernel_y, IFM, OFM].
- * Data type supported: Same as @p input or QASYMM8/QSYMM8_PER_CHANNEL when @p input is QASYMM8.
+ * Data type supported: Same as @p input, also could be QSYMM8_PER_CHANNEL if input is QASYMM8/QASYMM8_SIGNED.
* @param[in] output Destination tensor. 3 lower dimensions represent a single output [width, height, OFM], while the rest represent batch of outputs.
* Data types supported: Same as @p input.
* @param[in] conv_info Contains padding and stride information described in @ref PadStrideInfo.
diff --git a/arm_compute/runtime/CL/functions/CLCropResize.h b/arm_compute/runtime/CL/functions/CLCropResize.h
index 0dc3c48..5c60c28 100644
--- a/arm_compute/runtime/CL/functions/CLCropResize.h
+++ b/arm_compute/runtime/CL/functions/CLCropResize.h
@@ -61,6 +61,14 @@
/** Configure kernel
*
+ * Valid data layouts:
+ * - NHWC
+ *
+ * Valid data type configurations:
+ * |src0 |src1 |src2 |dst |
+ * |:--------|:--------|:------|:--------|
+ * |All |F32 |F32 |F32 |
+ *
* @note Supported tensor rank: up to 4
* @note Box indices may be outside of the bounds, in which case @p extrapolation_value is used.
* @note Start and end indices of boxes are inclusive.
diff --git a/arm_compute/runtime/CL/functions/CLDeconvolutionLayer.h b/arm_compute/runtime/CL/functions/CLDeconvolutionLayer.h
index 4be8c17..2dd4cd4 100644
--- a/arm_compute/runtime/CL/functions/CLDeconvolutionLayer.h
+++ b/arm_compute/runtime/CL/functions/CLDeconvolutionLayer.h
@@ -46,6 +46,20 @@
/** Set the input, weights, biases and output tensors.
*
+ * Valid data layouts:
+ * - NHWC
+ * - NCHW
+ *
+ * Valid data type configurations:
+ * |src0 |src1 |src2 |dst |
+ * |:--------------|:------------------|:------|:--------------|
+ * |F16 |F16 |F16 |F16 |
+ * |F32 |F32 |F32 |F32 |
+ * |QASYMM8 |QASYMM8 |S32 |QASYMM8 |
+ * |QASYMM8 |QSYMM8_PER_CHANNEL |S32 |QASYMM8 |
+ * |QASYMM8_SIGNED |QASYMM8_SIGNED |S32 |QASYMM8_SIGNED |
+ * |QASYMM8_SIGNED |QSYMM8_PER_CHANNEL |S32 |QASYMM8_SIGNED |
+ *
* @param[in,out] input Input tensor. 3 lower dimensions represent a single input, and an optional 4th dimension for batch of inputs. Data types supported: QASYMM8_SIGNED/QASYMM8/F16/F32.
* @param[in] weights The 4d weights with dimensions [width, height, IFM, OFM]. Data type supported: Same as @p input or QSYMM8_PER_CHANNEL if @p input is QASYMM8/QASYMM8_SIGNED.
* @param[in] bias (Optional) The biases have one dimension. Data type supported: Should match @p input data type, except for input of QASYMM8 and QASYMM8_SIGNED type where biases should be of S32 type
diff --git a/arm_compute/runtime/CL/functions/CLDepthConvertLayer.h b/arm_compute/runtime/CL/functions/CLDepthConvertLayer.h
index b0f297a..34dfdd7 100644
--- a/arm_compute/runtime/CL/functions/CLDepthConvertLayer.h
+++ b/arm_compute/runtime/CL/functions/CLDepthConvertLayer.h
@@ -1,5 +1,5 @@
/*
- * Copyright (c) 2016-2020 Arm Limited.
+ * Copyright (c) 2016-2021 Arm Limited.
*
* SPDX-License-Identifier: MIT
*
@@ -41,18 +41,22 @@
public:
/** Initialize the function's source, destination
*
+ * Valid data layouts:
+ * - All
+ *
+ * Valid data type configurations:
+ * |src |dst |
+ * |:--------------|:-------------------------------------|
+ * |U8 | S8, U16, S16, U32, S32, F16, F32 |
+ * |U16 | U8, S8, S16, U32, S32, F16, F32 |
+ * |S16 | U8, S8, U16, U32, S32, F16, F32 |
+ * |U32 | U8, S8, U16, S16, S32, F16, F32 |
+ * |S32 | U8, S8, U16, S16, U32, F16, F32 |
+ * |F16 | U8, S8, U16, S16, U32, F32 |
+ * |F32 | U8, S8, U16, S16, U32, F16 |
+ *
* Input data type must be different than output data type.
*
- * Valid conversions Input -> Output :
- *
- * - U8 -> S8, U16, S16, U32, S32, F16, F32
- * - U16 -> U8, S8, S16, U32, S32, F16, F32
- * - S16 -> U8, S8, U16, U32, S32, F16, F32
- * - U32 -> U8, S8, U16, S16, S32, F16, F32
- * - S32 -> U8, S8, U16, S16, U32, F16, F32
- * - F16 -> U8, S8, U16, S16, U32, F32
- * - F32 -> U8, S8, U16, S16, U32, F16
- *
* @param[in] input The input tensor to convert. Data types supported: U8/S8/U16/S16/U32/S32/F16/F32.
* @param[out] output The output tensor. Data types supported: U8/S8/U16/S16/U32/S32/F16/F32.
* @param[in] policy Conversion policy.
diff --git a/arm_compute/runtime/CL/functions/CLDepthToSpaceLayer.h b/arm_compute/runtime/CL/functions/CLDepthToSpaceLayer.h
index a0aa288..0026cc2 100644
--- a/arm_compute/runtime/CL/functions/CLDepthToSpaceLayer.h
+++ b/arm_compute/runtime/CL/functions/CLDepthToSpaceLayer.h
@@ -1,5 +1,5 @@
/*
- * Copyright (c) 2019-2020 Arm Limited.
+ * Copyright (c) 2019-2021 Arm Limited.
*
* SPDX-License-Identifier: MIT
*
@@ -39,6 +39,15 @@
public:
/** Set the input and output tensors.
*
+ * Valid data layouts:
+ * - NHWC
+ * - NCHW
+ *
+ * Valid data type configurations:
+ * |src |dst |
+ * |:--------------|:--------------|
+ * |All |All |
+ *
* @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.
diff --git a/arm_compute/runtime/CL/functions/CLDepthwiseConvolutionLayer.h b/arm_compute/runtime/CL/functions/CLDepthwiseConvolutionLayer.h
index 1af9e1d..f31a17d 100644
--- a/arm_compute/runtime/CL/functions/CLDepthwiseConvolutionLayer.h
+++ b/arm_compute/runtime/CL/functions/CLDepthwiseConvolutionLayer.h
@@ -58,6 +58,20 @@
~CLDepthwiseConvolutionLayer();
/** Initialize the function's source, destination, weights and convolution information.
*
+ * Valid data layouts:
+ * - NHWC
+ * - NCHW
+ *
+ * Valid data type configurations:
+ * |src0 |src1 |src2 |dst |
+ * |:--------------|:------------------|:------|:--------------|
+ * |F16 |F16 |F16 |F16 |
+ * |F32 |F32 |F32 |F32 |
+ * |QASYMM8 |QASYMM8 |S32 |QASYMM8 |
+ * |QASYMM8 |QSYMM8_PER_CHANNEL |S32 |QASYMM8 |
+ * |QASYMM8_SIGNED |QASYMM8_SIGNED |S32 |QASYMM8_SIGNED |
+ * |QASYMM8_SIGNED |QSYMM8_PER_CHANNEL |S32 |QASYMM8_SIGNED |
+ *
* @param[in, out] input Source tensor. Data type supported: QASYMM8/QASYMM8_SIGNED/FP16/FP32. Data layout supported: NHWC, NCHW
* @param[in] weights Weights tensor. These are 3D tensors with shape [kernel_x, kernel_y, IFM].
* Data type supported: Same as @p input or QASYMM8/QASYMM8_SIGNED/QSYMM8_PER_CHANNEL when @p input is QASYMM8.
diff --git a/arm_compute/runtime/CL/functions/CLDequantizationLayer.h b/arm_compute/runtime/CL/functions/CLDequantizationLayer.h
index 4a5c3a3..601c13d 100644
--- a/arm_compute/runtime/CL/functions/CLDequantizationLayer.h
+++ b/arm_compute/runtime/CL/functions/CLDequantizationLayer.h
@@ -58,18 +58,13 @@
* - All
*
* Valid data type configurations:
- * |src |dst |
- * |:------------------|:--------------|
- * |QASYMM8 |F16 |
- * |QASYMM8 |F32 |
- * |QASYMM8_SIGNED |F16 |
- * |QASYMM8_SIGNED |F32 |
- * |QSYMM8_PER_CHANNEL |F16 |
- * |QSYMM8_PER_CHANNEL |F32 |
- * |QSYMM8 |F16 |
- * |QSYMM8 |F32 |
- * |QSYMM16 |F16 |
- * |QSYMM16 |F32 |
+ * |src |dst |
+ * |:------------------|:---------|
+ * |QASYMM8 |F16, F32 |
+ * |QASYMM8_SIGNED |F16, F32 |
+ * |QSYMM8_PER_CHANNEL |F16, F32 |
+ * |QSYMM8 |F16, F32 |
+ * |QSYMM16 |F16, F32 |
*
* @param[in] input Source tensor with at least 3 dimensions. The dimensions over the third will be interpreted as batches.
* Data types supported: QASYMM8/QASYMM8_SIGNED/QSYMM8_PER_CHANNEL/QSYMM8/QSYMM16.
diff --git a/arm_compute/runtime/CL/functions/CLFillBorder.h b/arm_compute/runtime/CL/functions/CLFillBorder.h
index a4ad82d..20f2e15 100644
--- a/arm_compute/runtime/CL/functions/CLFillBorder.h
+++ b/arm_compute/runtime/CL/functions/CLFillBorder.h
@@ -1,5 +1,5 @@
/*
- * Copyright (c) 2016-2020 Arm Limited.
+ * Copyright (c) 2016-2021 Arm Limited.
*
* SPDX-License-Identifier: MIT
*
@@ -39,6 +39,14 @@
public:
/** Initialize the function
*
+ * Valid data layouts:
+ * - All
+ *
+ * Valid data type configurations:
+ * |src |dst |
+ * |:--------------|:--------------|
+ * |All |All |
+ *
* @param[in,out] tensor Source tensor. Data types supported: U8/QASYMM8/S8/QASYMM8_SIGNED/U16/S16/U32/S32/F16/F32.
* @param[in] border_width The border width
* @param[in] border_mode Strategy to use for borders.
diff --git a/arm_compute/runtime/CL/functions/CLFlattenLayer.h b/arm_compute/runtime/CL/functions/CLFlattenLayer.h
index 8dedd74..d2c37b1 100644
--- a/arm_compute/runtime/CL/functions/CLFlattenLayer.h
+++ b/arm_compute/runtime/CL/functions/CLFlattenLayer.h
@@ -44,6 +44,14 @@
public:
/** Initialise the kernel's input and output.
*
+ * Valid data layouts:
+ * - All
+ *
+ * Valid data type configurations:
+ * |src |dst |
+ * |:--------------|:--------------|
+ * |All |All |
+ *
* @param[in] input First input tensor to flatten with at least 3 dimensions.
* The dimensions above the third will be interpreted as batches. Data types supported: All.
* @param[out] output Output tensor with shape [w*h*d, input_batches] where:
diff --git a/arm_compute/runtime/CL/functions/CLFullyConnectedLayer.h b/arm_compute/runtime/CL/functions/CLFullyConnectedLayer.h
index 75cb2dc..eec01bc 100644
--- a/arm_compute/runtime/CL/functions/CLFullyConnectedLayer.h
+++ b/arm_compute/runtime/CL/functions/CLFullyConnectedLayer.h
@@ -117,6 +117,18 @@
CLFullyConnectedLayer &operator=(CLFullyConnectedLayer &&) = default;
/** Set the input and output tensors.
*
+ * Valid data layouts:
+ * - NHWC
+ * - NCHW
+ *
+ * Valid data type configurations:
+ * |src0 |src1 |src2 |dst |
+ * |:--------------|:------------------|:------|:--------------|
+ * |F16 |F16 |F16 |F16 |
+ * |F32 |F32 |F32 |F32 |
+ * |QASYMM8 |QASYMM8 |S32 |QASYMM8 |
+ * |QASYMM8_SIGNED |QASYMM8_SIGNED |S32 |QASYMM8_SIGNED |
+ *
* @param[in] input Source tensor. Data type supported: QASYMM8/QASYMM8_SIGNED/F16/F32.
* @param[in] weights Weights tensor. The weights must be 2 dimensional.
* If this function is called after a Convolution Layer, the (transposed) weights will have as many rows as the product of the first 3 input's dimensions.
diff --git a/arm_compute/runtime/CL/functions/CLFuseBatchNormalization.h b/arm_compute/runtime/CL/functions/CLFuseBatchNormalization.h
index e35905f..cd75270 100644
--- a/arm_compute/runtime/CL/functions/CLFuseBatchNormalization.h
+++ b/arm_compute/runtime/CL/functions/CLFuseBatchNormalization.h
@@ -1,5 +1,5 @@
/*
- * Copyright (c) 2018-2020 Arm Limited.
+ * Copyright (c) 2018-2021 Arm Limited.
*
* SPDX-License-Identifier: MIT
*
@@ -55,6 +55,16 @@
~CLFuseBatchNormalization();
/** Set the input and output tensors.
*
+ * Valid data layouts:
+ * - NHWC
+ * - NCHW
+ *
+ * Valid data type configurations:
+ * |src |dst |
+ * |:--------------|:--------------|
+ * |F32 |F32 |
+ * |F16 |F16 |
+ *
* @param[in] input_weights Input weights tensor for convolution or depthwise convolution layer. Data type supported: F16/F32. Data layout supported: NCHW, NHWC
* @param[in] bn_mean Batch normalization layer mean tensor. Same as @p input_weights
* @param[in] bn_var Batch normalization layer variance tensor. Same as @p input_weights
diff --git a/arm_compute/runtime/CL/functions/CLGEMM.h b/arm_compute/runtime/CL/functions/CLGEMM.h
index 8a210a2..1e2ae7b 100644
--- a/arm_compute/runtime/CL/functions/CLGEMM.h
+++ b/arm_compute/runtime/CL/functions/CLGEMM.h
@@ -126,6 +126,15 @@
~CLGEMM();
/** Initialise the kernel's inputs and output
*
+ * Valid data layouts:
+ * - All
+ *
+ * Valid data type configurations:
+ * |src0 |src1 |src2 |dst |
+ * |:------------|:-----------|:---------|:--------------|
+ * |F32 |F32 |F32 |F32 |
+ * |F16 |F16 |F16 |F16 |
+ *
* @note GEMM: General Matrix Multiply - [alpha * A * B + beta * C].
*
* @note All tensors must have the same data type.
diff --git a/arm_compute/runtime/CL/functions/CLGEMMConvolutionLayer.h b/arm_compute/runtime/CL/functions/CLGEMMConvolutionLayer.h
index 4dbd0f8..082b481 100644
--- a/arm_compute/runtime/CL/functions/CLGEMMConvolutionLayer.h
+++ b/arm_compute/runtime/CL/functions/CLGEMMConvolutionLayer.h
@@ -1,5 +1,5 @@
/*
- * Copyright (c) 2017-2020 Arm Limited.
+ * Copyright (c) 2017-2021 Arm Limited.
*
* SPDX-License-Identifier: MIT
*
@@ -194,6 +194,20 @@
~CLGEMMConvolutionLayer();
/** Set the input and output tensors.
*
+ * Valid data layouts:
+ * - NHWC
+ * - NCHW
+ *
+ * Valid data type configurations:
+ * |src0 |src1 |src2 |dst |
+ * |:--------------|:------------------|:--------|:--------------|
+ * |F16 |F16 |F16 |F16 |
+ * |F32 |F32 |F32 |F32 |
+ * |QASYMM8 |QASYMM8 |S32 |QASYMM8 |
+ * |QASYMM8 |QSYMM8_PER_CHANNEL |S32 |QASYMM8 |
+ * |QASYMM8_SIGNED |QASYMM8_SIGNED |S32 |QASYMM8_SIGNED |
+ * |QASYMM8_SIGNED |QSYMM8_PER_CHANNEL |S32 |QASYMM8_SIGNED |
+ *
* @param[in] input Source tensor. 3 lower dimensions represent a single input [width, height, IFM],
* while every optional dimension from 4 and above represent a batch of inputs.
* Data types supported: QASYMM8/QASYMM8_SIGNED/F16/F32.
diff --git a/arm_compute/runtime/CL/functions/CLGEMMLowpMatrixMultiplyCore.h b/arm_compute/runtime/CL/functions/CLGEMMLowpMatrixMultiplyCore.h
index 4cc8899..e7f4cb9 100644
--- a/arm_compute/runtime/CL/functions/CLGEMMLowpMatrixMultiplyCore.h
+++ b/arm_compute/runtime/CL/functions/CLGEMMLowpMatrixMultiplyCore.h
@@ -1,5 +1,5 @@
/*
- * Copyright (c) 2017-2020 Arm Limited.
+ * Copyright (c) 2017-2021 Arm Limited.
*
* SPDX-License-Identifier: MIT
*
@@ -61,6 +61,26 @@
~CLGEMMLowpMatrixMultiplyCore();
/** Initialise the kernel's inputs, output
*
+ * Valid data layouts:
+ * - NHWC
+ * - NCHW
+ *
+ * Valid data type configurations:
+ * |src0 |src1 |src2 |dst |
+ * |:--------------|:------------------|:--------|:--------------|
+ * |QASYMM8 |QASYMM8 |S32 |QASYMM8 |
+ * |QASYMM8 |QSYMM8_PER_CHANNEL |S32 |QASYMM8 |
+ * |QASYMM8 |QSYMM8 |S32 |QASYMM8 |
+ * |QASYMM8 |QASYMM8 |S32 |S32 |
+ * |QASYMM8 |QSYMM8_PER_CHANNEL |S32 |S32 |
+ * |QASYMM8 |QSYMM8 |S32 |S32 |
+ * |QASYMM8_SIGNED |QASYMM8_SIGNED |S32 |QASYMM8_SIGNED |
+ * |QASYMM8_SIGNED |QSYMM8_PER_CHANNEL |S32 |QASYMM8_SIGNED |
+ * |QASYMM8_SIGNED |QSYMM8 |S32 |QASYMM8_SIGNED |
+ * |QASYMM8_SIGNED |QASYMM8_SIGNED |S32 |S32 |
+ * |QASYMM8_SIGNED |QSYMM8_PER_CHANNEL |S32 |S32 |
+ * |QASYMM8_SIGNED |QSYMM8 |S32 |S32 |
+ *
* @note GEMMLowp: low precision GEMM kernel. [A * B + C]
* This kernel performs the following computations:
*
diff --git a/arm_compute/runtime/CL/functions/CLGather.h b/arm_compute/runtime/CL/functions/CLGather.h
index 9c659be..7a57c73 100644
--- a/arm_compute/runtime/CL/functions/CLGather.h
+++ b/arm_compute/runtime/CL/functions/CLGather.h
@@ -1,5 +1,5 @@
/*
- * Copyright (c) 2018-2020 Arm Limited.
+ * Copyright (c) 2018-2021 Arm Limited.
*
* SPDX-License-Identifier: MIT
*
@@ -40,6 +40,14 @@
public:
/** Initialise the kernel's inputs and outputs
*
+ * Valid data layouts:
+ * - All
+ *
+ * Valid data type configurations:
+ * |src |dst |
+ * |:--------------|:--------------|
+ * |All |All |
+ *
* @param[in] input Source tensor. Supported tensor rank: up to 4. Data type supported: All.
* @param[in] indices Indices tensor. Supported tensor rank: up to 1. Must be one of the following types: U32/S32. Each value must be in range [0, input.shape[@p axis])
* @param[out] output Destination tensor. Data type supported: Same as @p input
diff --git a/arm_compute/runtime/CL/functions/CLGenerateProposalsLayer.h b/arm_compute/runtime/CL/functions/CLGenerateProposalsLayer.h
index bea4707..aec5cdf 100644
--- a/arm_compute/runtime/CL/functions/CLGenerateProposalsLayer.h
+++ b/arm_compute/runtime/CL/functions/CLGenerateProposalsLayer.h
@@ -77,6 +77,16 @@
/** Set the input and output tensors.
*
+ * Valid data layouts:
+ * - All
+ *
+ * Valid data type configurations:
+ * |src0 |src1 |src2 |dst |
+ * |:--------------|:------------------|:--------|:--------------|
+ * |F16 |F16 |F16 |F16 |
+ * |F32 |F32 |F32 |F32 |
+ * |QASYMM8 |QSYMM8 |QSYMM16 |QASYMM8 |
+ *
* @param[in] scores Scores from convolution layer of size (W, H, A), where H and W are the height and width of the feature map, and A is the number of anchors.
* Data types supported: QASYMM8/F16/F32
* @param[in] deltas Bounding box deltas from convolution layer of size (W, H, 4*A). Data types supported: Same as @p scores
diff --git a/arm_compute/runtime/CL/functions/CLInstanceNormalizationLayer.h b/arm_compute/runtime/CL/functions/CLInstanceNormalizationLayer.h
index a6e5b16..985a6a7 100644
--- a/arm_compute/runtime/CL/functions/CLInstanceNormalizationLayer.h
+++ b/arm_compute/runtime/CL/functions/CLInstanceNormalizationLayer.h
@@ -65,6 +65,16 @@
/** Set the input and output tensors.
*
+ * Valid data layouts:
+ * - NHWC
+ * - NCHW
+ *
+ * Valid data type configurations:
+ * |src |dst |
+ * |:--------|:---------|
+ * |F16 |F16 |
+ * |F32 |F32 |
+ *
* @param[in, out] input Source tensor. In case of @p output tensor = nullptr this tensor will store the result of the normalization.
* Data types supported: F16/F32. Data layout supported: NHWC, NCHW
* @param[out] output Destination tensor. Data types and data layouts supported: same as @p input.
diff --git a/arm_compute/runtime/CL/functions/CLL2NormalizeLayer.h b/arm_compute/runtime/CL/functions/CLL2NormalizeLayer.h
index 401d249..4dc5c77 100644
--- a/arm_compute/runtime/CL/functions/CLL2NormalizeLayer.h
+++ b/arm_compute/runtime/CL/functions/CLL2NormalizeLayer.h
@@ -1,5 +1,5 @@
/*
- * Copyright (c) 2017-2020 Arm Limited.
+ * Copyright (c) 2017-2021 Arm Limited.
*
* SPDX-License-Identifier: MIT
*
@@ -65,6 +65,16 @@
/** Set the input and output tensors.
*
+ * Valid data layouts:
+ * - NHWC
+ * - NCHW
+ *
+ * Valid data type configurations:
+ * |src |dst |
+ * |:--------|:---------|
+ * |F16 |F16 |
+ * |F32 |F32 |
+ *
* @param[in] input Source tensor. Data types supported: F16/F32. Data layouts supported: NCHW/NHWC.
* @param[out] output Destination tensor. Data types and data layouts supported: Same as @p input.
* @param[in] axis Axis along which to reduce. Negative values wrap around. Maximum supported actual reduction axis : 2
diff --git a/arm_compute/runtime/CL/functions/CLLSTMLayer.h b/arm_compute/runtime/CL/functions/CLLSTMLayer.h
index 38a24d0..d26b4c5 100644
--- a/arm_compute/runtime/CL/functions/CLLSTMLayer.h
+++ b/arm_compute/runtime/CL/functions/CLLSTMLayer.h
@@ -75,6 +75,15 @@
~CLLSTMLayer();
/** Initialize function's tensors.
*
+ * Valid data layouts:
+ * - All
+ *
+ * Valid data type configurations:
+ * |src0 - src13 | dst0 - dst3 |
+ * |:------------|:------------|
+ * |F16 |F16 |
+ * |F32 |F32 |
+ *
* @param[in] input Source tensor. Input is a 2D tensor with dimensions [input_size, batch_size]. Data types supported: F16/F32.
* @param[in] input_to_forget_weights 2D weights tensor with dimensions [input_size, num_units]. Data type supported: Same as @p input.
* @param[in] input_to_cell_weights 2D weights tensor with dimensions [input_size, num_units]. Data type supported: Same as @p input.
diff --git a/arm_compute/runtime/CL/functions/CLLSTMLayerQuantized.h b/arm_compute/runtime/CL/functions/CLLSTMLayerQuantized.h
index 0829052..2ef7427 100644
--- a/arm_compute/runtime/CL/functions/CLLSTMLayerQuantized.h
+++ b/arm_compute/runtime/CL/functions/CLLSTMLayerQuantized.h
@@ -1,5 +1,5 @@
/*
- * Copyright (c) 2019-2020 Arm Limited.
+ * Copyright (c) 2019-2021 Arm Limited.
*
* SPDX-License-Identifier: MIT
*
@@ -73,6 +73,14 @@
CLLSTMLayerQuantized &operator=(CLLSTMLayerQuantized &&) = default;
/** Initialize function's tensors.
*
+ * Valid data layouts:
+ * - All
+ *
+ * Valid data type configurations:
+ * |src0 - src8 |src9 - src12 |src13 |src14 |dst0 |dst1 |
+ * |:-----------|:------------|:-------|:------|:------|:------|
+ * |QASYMM8 |S32 |QSYMM16 |QASYMM8|QSYMM16|QASYMM8|
+ *
* @param[in] input Source tensor. Input is a 2D tensor with dimensions [input_size, batch_size]. Data types supported: QASYMM8.
* @param[in] input_to_input_weights 2D weights tensor with dimensions [input_size, output_size]. Data type supported: Same as @p input.
* @param[in] input_to_forget_weights 2D weights tensor with dimensions [input_size, output_size]. Data type supported: Same as @p input.
diff --git a/arm_compute/runtime/CL/functions/CLMaxUnpoolingLayer.h b/arm_compute/runtime/CL/functions/CLMaxUnpoolingLayer.h
index 24d620d..f7ff123 100644
--- a/arm_compute/runtime/CL/functions/CLMaxUnpoolingLayer.h
+++ b/arm_compute/runtime/CL/functions/CLMaxUnpoolingLayer.h
@@ -56,6 +56,18 @@
~CLMaxUnpoolingLayer();
/** Set the input and output tensors.
*
+ * Valid data layouts:
+ * - NHWC
+ * - NCHW
+ *
+ * Valid data type configurations:
+ * |src |dst |
+ * |:--------------|:--------------|
+ * |QASYMM8 |QASYMM8 |
+ * |QASYMM8_SIGNED |QASYMM8_SIGNED |
+ * |F16 |F16 |
+ * |F32 |F32 |
+ *
* @note Output shape must be equal to the shape of the original input to pool.
*
* @param[in] input Source tensor. Data types supported: QASYMM8/QASYMM8_SIGNED/F16/F32.
diff --git a/arm_compute/runtime/CL/functions/CLMeanStdDevNormalizationLayer.h b/arm_compute/runtime/CL/functions/CLMeanStdDevNormalizationLayer.h
index cfe59ea..68a7df2 100644
--- a/arm_compute/runtime/CL/functions/CLMeanStdDevNormalizationLayer.h
+++ b/arm_compute/runtime/CL/functions/CLMeanStdDevNormalizationLayer.h
@@ -1,5 +1,5 @@
/*
- * Copyright (c) 2019-2020 Arm Limited.
+ * Copyright (c) 2019-2021 Arm Limited.
*
* SPDX-License-Identifier: MIT
*
@@ -39,6 +39,16 @@
public:
/** Initialise the function's input and outputs.
*
+ * Valid data layouts:
+ * - NHWC
+ * - NCHW
+ *
+ * Valid data type configurations:
+ * |src |dst |
+ * |:--------|:---------|
+ * |F32 |F32 |
+ * |F16 |F16 |
+ *
* @note If the output tensor is a nullptr, the normalization will be performed in-place.
*
* @param[in, out] input Input tensor with 2 dimensions. Data types supported: F16/F32.
diff --git a/arm_compute/runtime/CL/functions/CLNormalizationLayer.h b/arm_compute/runtime/CL/functions/CLNormalizationLayer.h
index 706cb6f..15406f7 100644
--- a/arm_compute/runtime/CL/functions/CLNormalizationLayer.h
+++ b/arm_compute/runtime/CL/functions/CLNormalizationLayer.h
@@ -61,6 +61,16 @@
~CLNormalizationLayer();
/** Set the input and output tensors.
*
+ * Valid data layouts:
+ * - NHWC
+ * - NCHW
+ *
+ * Valid data type configurations:
+ * |src |dst |
+ * |:--------|:---------|
+ * |F32 |F32 |
+ * |F16 |F16 |
+ *
* @param[in, out] input Source tensor. 3 lower dims represent a single input with dimensions [width, height, IFM],
* and an optional 4th dimension for batch of inputs. Data types supported: F16/F32 (Written to by the border handler).
* Data layouts supported: NCHW/NHWC.
diff --git a/arm_compute/runtime/CL/functions/CLPadLayer.h b/arm_compute/runtime/CL/functions/CLPadLayer.h
index dae95f6..7f950bc 100644
--- a/arm_compute/runtime/CL/functions/CLPadLayer.h
+++ b/arm_compute/runtime/CL/functions/CLPadLayer.h
@@ -59,6 +59,15 @@
/** Initialize the function
*
+ * Valid data layouts:
+ * - NHWC
+ * - NCHW
+ *
+ * Valid data type configurations:
+ * |src |dst |
+ * |:--------|:---------|
+ * |All |All |
+ *
* @param[in] input Source tensor. Data types supported: All.
* @param[out] output Output tensor. Data type supported: same as @p input
* @param[in] padding The padding for each spatial dimension of the input tensor. The pair padding[i]
diff --git a/arm_compute/runtime/CL/functions/CLPriorBoxLayer.h b/arm_compute/runtime/CL/functions/CLPriorBoxLayer.h
index 9129bfd..9b36c9e 100644
--- a/arm_compute/runtime/CL/functions/CLPriorBoxLayer.h
+++ b/arm_compute/runtime/CL/functions/CLPriorBoxLayer.h
@@ -1,5 +1,5 @@
/*
- * Copyright (c) 2018-2020 Arm Limited.
+ * Copyright (c) 2018-2021 Arm Limited.
*
* SPDX-License-Identifier: MIT
*
@@ -43,6 +43,15 @@
CLPriorBoxLayer();
/** Set the input and output tensors.
*
+ * Valid data layouts:
+ * - NHWC
+ * - NCHW
+ *
+ * Valid data type configurations:
+ * |src0 |src1 |dst |
+ * |:--------|:--------|:--------|
+ * |F32 |F32 |F32 |
+ *
* @param[in] input1 First source tensor. Data types supported: F32. Data layouts supported: NCHW/NHWC.
* @param[in] input2 Second source tensor. Data types and layouts supported: same as @p input1
* @param[out] output Destination tensor. Output dimensions are [W * H * num_priors * 4, 2]. Data types and layouts supported: same as @p input1
diff --git a/arm_compute/runtime/CL/functions/CLQLSTMLayer.h b/arm_compute/runtime/CL/functions/CLQLSTMLayer.h
index 954f224..bd00d56 100644
--- a/arm_compute/runtime/CL/functions/CLQLSTMLayer.h
+++ b/arm_compute/runtime/CL/functions/CLQLSTMLayer.h
@@ -74,6 +74,14 @@
~CLQLSTMLayer();
/** Initialize function's tensors.
*
+ * Valid data layouts:
+ * - All
+ *
+ * Valid data type configurations:
+ * |src0 |src1 - src6 |src7 -src9 |src10 |src11 |dst0 |dst1 - dst2 |
+ * |:-------------|:------------|:------------|:------|:-------------|:------|:-----------------|
+ * |QASYMM8_SIGNED|QASYMM8 |S32 |QSYMM16|QASYMM8_SIGNED|QSYMM16|QASYMM8_SIGNED |
+ *
* @param[in] input Source tensor. Input is a 2D tensor with dimensions [input_size, batch_size]. Data types supported: QASYMM8_SIGNED.
* @param[in] input_to_forget_weights 2D weights tensor with dimensions [input_size, num_units]. Data type supported: QSYMM8.
* @param[in] input_to_cell_weights 2D weights tensor with dimensions [input_size, num_units]. Data type supported: QSYMM8.
diff --git a/arm_compute/runtime/CL/functions/CLQuantizationLayer.h b/arm_compute/runtime/CL/functions/CLQuantizationLayer.h
index 6a44a22..a61735c 100644
--- a/arm_compute/runtime/CL/functions/CLQuantizationLayer.h
+++ b/arm_compute/runtime/CL/functions/CLQuantizationLayer.h
@@ -63,20 +63,12 @@
* - All
*
* Valid data type configurations:
- * |src |dst |
- * |:------------------|:--------------|
- * |QASYMM8 |QASYMM8 |
- * |QASYMM8 |QASYMM8_SIGNED |
- * |QASYMM8 |QASYMM16 |
- * |QASYMM8_SIGNED |QASYMM8 |
- * |QASYMM8_SIGNED |QASYMM8_SIGNED |
- * |QASYMM8_SIGNED |QASYMM16 |
- * |F16 |QASYMM8 |
- * |F16 |QASYMM8_SIGNED |
- * |F16 |QASYMM16 |
- * |F32 |QASYMM8 |
- * |F32 |QASYMM8_SIGNED |
- * |F32 |QASYMM16 |
+ * |src |dst |
+ * |:------------------|:----------------------------------|
+ * |QASYMM8 |QASYMM8, QASYMM8_SIGNED, QASYMM16 |
+ * |QASYMM8_SIGNED |QASYMM8, QASYMM8_SIGNED, QASYMM16 |
+ * |F16 |QASYMM8, QASYMM8_SIGNED, QASYMM16 |
+ * |F32 |QASYMM8, QASYMM8_SIGNED, QASYMM16 |
*
* @param[in] input Source tensor. The dimensions over the third will be interpreted as batches. Data types supported: QASYMM8/QASYMM8_SIGNED/F16/32.
* @param[out] output Destination tensor with the same dimensions of input. Data types supported: QASYMM8/QASYMM8_SIGNED/QASYMM16.
diff --git a/arm_compute/runtime/CL/functions/CLRNNLayer.h b/arm_compute/runtime/CL/functions/CLRNNLayer.h
index 50575da..2b3b35e 100644
--- a/arm_compute/runtime/CL/functions/CLRNNLayer.h
+++ b/arm_compute/runtime/CL/functions/CLRNNLayer.h
@@ -51,6 +51,16 @@
~CLRNNLayer();
/** Initialize the function
*
+ * Valid data layouts:
+ * - NHWC
+ * - NCHW
+ *
+ * Valid data type configurations:
+ * |src0 |src1 |src2 |src3 |dst0 |dst1 |
+ * |:------|:------|:------|:------|:------|:------|
+ * |F16 |F16 |F16 |F16 |F16 |F16 |
+ * |F32 |F32 |F32 |F32 |F32 |F32 |
+ *
* @param[in] input Input is a 2-D tensor of shape [input_size, batch_size]. Data types supported: F16/F32
* @param[in] weights Weights tensor of shape [input_size, num_units] that multiplies the input. Data types supported: Same as @p input
* @param[in] recurrent_weights Weights tensor of shape [num_units, num_units] that multiplies the current 'state'. Data types supported: Same as @p input
diff --git a/arm_compute/runtime/CL/functions/CLROIAlignLayer.h b/arm_compute/runtime/CL/functions/CLROIAlignLayer.h
index b4cd556..1eaea1b 100644
--- a/arm_compute/runtime/CL/functions/CLROIAlignLayer.h
+++ b/arm_compute/runtime/CL/functions/CLROIAlignLayer.h
@@ -1,5 +1,5 @@
/*
- * Copyright (c) 2018-2020 Arm Limited.
+ * Copyright (c) 2018-2021 Arm Limited.
*
* SPDX-License-Identifier: MIT
*
@@ -45,6 +45,17 @@
public:
/** Set the input and output tensors.
*
+ * Valid data layouts:
+ * - All
+ *
+ * Valid data type configurations:
+ * |src0 |src1 |dst |
+ * |:--------------|:--------------|:--------------|
+ * |F16 |F16 |F16 |
+ * |F32 |F32 |F32 |
+ * |QASYMM8 |QASYMM16 |QASYMM8 |
+ * |QASYMM8_SIGNED |QASYMM16 |QASYMM8_SIGNED |
+ *
* @param[in] input Source tensor. Data types supported: QASYMM8/QASYMM8_SIGNED/F16/F32.
* @param[in] rois ROIs tensor, it is a 2D tensor of size [5, N] (where N is the number of ROIs) containing top left and bottom right corner
* as coordinate of an image and batch_id of ROI [ batch_id, x1, y1, x2, y2 ].
diff --git a/arm_compute/runtime/CL/functions/CLROIPoolingLayer.h b/arm_compute/runtime/CL/functions/CLROIPoolingLayer.h
index a4c5c76..151586a 100644
--- a/arm_compute/runtime/CL/functions/CLROIPoolingLayer.h
+++ b/arm_compute/runtime/CL/functions/CLROIPoolingLayer.h
@@ -45,6 +45,16 @@
public:
/** Set the input and output tensors.
*
+ * Valid data layouts:
+ * - All
+ *
+ * Valid data type configurations:
+ * |src0 |src1 |dst |
+ * |:--------------|:--------------|:--------------|
+ * |F16 |U16 |F16 |
+ * |F32 |U16 |F32 |
+ * |QASYMM8 |U16 |QASYMM8 |
+ *
* @param[in] input Source tensor. Data types supported: F16/F32/QASYMM8
* @param[in] rois ROIs tensor, it is a 2D tensor of size [5, N] (where N is the number of ROIs) containing top left and bottom right corner
* as coordinate of an image and batch_id of ROI [ batch_id, x1, y1, x2, y2 ]. Data types supported: U16
diff --git a/arm_compute/runtime/CL/functions/CLRange.h b/arm_compute/runtime/CL/functions/CLRange.h
index e11e740..fbce051 100644
--- a/arm_compute/runtime/CL/functions/CLRange.h
+++ b/arm_compute/runtime/CL/functions/CLRange.h
@@ -1,5 +1,5 @@
/*
- * Copyright (c) 2018-2020 Arm Limited.
+ * Copyright (c) 2018-2021 Arm Limited.
*
* SPDX-License-Identifier: MIT
*
@@ -43,6 +43,22 @@
public:
/** Initialize the kernel's start, end, step and output tensor.
*
+ * Valid data layouts:
+ * - All
+ *
+ * Valid data type configurations:
+ * |dst |
+ * |:---------|
+ * |U8 |
+ * |S8 |
+ * |QASYMM8 |
+ * |U16 |
+ * |S16 |
+ * |U32 |
+ * |S32 |
+ * |F16 |
+ * |F32 |
+ *
* @param[out] output Output tensor. Data types supported: U8/S8/QASYMM8/U16/S16/U32/S32/F16/F32.
* @param[in] start The starting value of the sequence.
* @param[in] end The ending (not including) value of the sequence.
diff --git a/arm_compute/runtime/CL/functions/CLReduceMean.h b/arm_compute/runtime/CL/functions/CLReduceMean.h
index c37ee8c..1ce088b 100644
--- a/arm_compute/runtime/CL/functions/CLReduceMean.h
+++ b/arm_compute/runtime/CL/functions/CLReduceMean.h
@@ -1,5 +1,5 @@
/*
- * Copyright (c) 2018-2020 Arm Limited.
+ * Copyright (c) 2018-2021 Arm Limited.
*
* SPDX-License-Identifier: MIT
*
@@ -45,6 +45,17 @@
CLReduceMean(std::shared_ptr<IMemoryManager> memory_manager = nullptr);
/** Configure kernel
*
+ * Valid data layouts:
+ * - All
+ *
+ * Valid data type configurations:
+ * |src |dst |
+ * |:--------------|:--------------|
+ * |QASYMM8 |QASYMM8 |
+ * |QASYMM8_SIGNED |QASYMM8_SIGNED |
+ * |F16 |F16 |
+ * |F32 |F32 |
+ *
* @note Supported tensor rank: up to 4
*
* @param[in] input Source tensor. Data type supported: QASYMM8/QASYMM8_SIGNED/F16/F32
diff --git a/arm_compute/runtime/CL/functions/CLReductionOperation.h b/arm_compute/runtime/CL/functions/CLReductionOperation.h
index 58164fd..2245735 100644
--- a/arm_compute/runtime/CL/functions/CLReductionOperation.h
+++ b/arm_compute/runtime/CL/functions/CLReductionOperation.h
@@ -62,6 +62,18 @@
/** Set the input and output tensors.
*
+ * Valid data layouts:
+ * - All
+ *
+ * Valid data type configurations:
+ * |src |dst |
+ * |:--------------|:--------------|
+ * |QASYMM8 |QASYMM8 |
+ * |QASYMM8_SIGNED |QASYMM8_SIGNED |
+ * |F16 |F16 |
+ * |F32 |F32 |
+ * |S32 |S32 |
+ *
* @param[in] input Source tensor. Data types supported: QASYMM8/QASYMM8_SIGNED/F16/F32/S32.
* @param[out] output Destination tensor. Data types and data layouts supported: Same as @p input.
* @param[in] axis Axis along which to reduce. Supported reduction axis : 0, 1, 2, 3
diff --git a/arm_compute/runtime/CL/functions/CLRemap.h b/arm_compute/runtime/CL/functions/CLRemap.h
index 87d5f9f..39ee13b 100644
--- a/arm_compute/runtime/CL/functions/CLRemap.h
+++ b/arm_compute/runtime/CL/functions/CLRemap.h
@@ -1,5 +1,5 @@
/*
- * Copyright (c) 2017-2020 Arm Limited.
+ * Copyright (c) 2017-2021 Arm Limited.
*
* SPDX-License-Identifier: MIT
*
@@ -47,6 +47,14 @@
public:
/** Initialise the function's sources, destination, interpolation policy and border mode.
*
+ * Valid data layouts:
+ * - All
+ *
+ * Valid data type configurations:
+ * |src0 |src1 |src2 |dst |
+ * |:------|:------|:------|:------|
+ * |U8 |F32 |F32 |U 8 |
+ *
* @param[in,out] input Source tensor. Data types supported: U8. (Written to only for @p border_mode != UNDEFINED)
* @param[in] map_x Map for X coords. Data types supported: F32.
* @param[in] map_y Map for Y coords. Data types supported: F32.
diff --git a/arm_compute/runtime/CL/functions/CLReorgLayer.h b/arm_compute/runtime/CL/functions/CLReorgLayer.h
index 0840fd1..976b8f6 100644
--- a/arm_compute/runtime/CL/functions/CLReorgLayer.h
+++ b/arm_compute/runtime/CL/functions/CLReorgLayer.h
@@ -1,5 +1,5 @@
/*
- * Copyright (c) 2018-2020 Arm Limited.
+ * Copyright (c) 2018-2021 Arm Limited.
*
* SPDX-License-Identifier: MIT
*
@@ -38,6 +38,15 @@
public:
/** Initialise the function's source and destination.
*
+ * Valid data layouts:
+ * - NHWC
+ * - NCHW
+ *
+ * Valid data type configurations:
+ * |src |dst |
+ * |:--------------|:--------------|
+ * |All |All |
+ *
* @param[in] input Source tensor. Data types supported: All.
* @param[out] output Destination tensor with tensor shape:
* [width_input / stride, height_input / stride, channels_input * stride * stride, batch_size]. This means the output has
diff --git a/arm_compute/runtime/CL/functions/CLReverse.h b/arm_compute/runtime/CL/functions/CLReverse.h
index 81fa04b..94c63ca 100644
--- a/arm_compute/runtime/CL/functions/CLReverse.h
+++ b/arm_compute/runtime/CL/functions/CLReverse.h
@@ -1,5 +1,5 @@
/*
- * Copyright (c) 2018-2020 Arm Limited.
+ * Copyright (c) 2018-2021 Arm Limited.
*
* SPDX-License-Identifier: MIT
*
@@ -39,6 +39,14 @@
public:
/** Initialize the function
*
+ * Valid data layouts:
+ * - All
+ *
+ * Valid data type configurations:
+ * |src0 |src1 |dst |
+ * |:--------------|:--------------|:--------------|
+ * |All |U32 |All |
+ *
* @param[in] input Input tensor. Data types supported: All.
* @param[out] output Output tensor. Data type supported: Same as @p input
* @param[in] axis Axis tensor. Contains the indices of the dimensions to reverse. Data type supported: U32
diff --git a/arm_compute/runtime/CL/functions/CLSelect.h b/arm_compute/runtime/CL/functions/CLSelect.h
index 7fd5231..8b1e6b2 100644
--- a/arm_compute/runtime/CL/functions/CLSelect.h
+++ b/arm_compute/runtime/CL/functions/CLSelect.h
@@ -1,5 +1,5 @@
/*
- * Copyright (c) 2018-2020 Arm Limited.
+ * Copyright (c) 2018-2021 Arm Limited.
*
* SPDX-License-Identifier: MIT
*
@@ -40,6 +40,14 @@
public:
/** Initialise the kernel's inputs and output.
*
+ * Valid data layouts:
+ * - All
+ *
+ * Valid data type configurations:
+ * |src0 |src1 |src2 |dst |
+ * |:--------------|:--------------|:------|:--------------|
+ * |U8 |All |All |All |
+ *
* @param[in] c Condition input tensor. Data types supported: U8.
* @param[in] x First input tensor. Data types supported: All.
* @param[in] y Second input tensor. Data types supported: Same as @p x
diff --git a/arm_compute/runtime/CL/functions/CLSpaceToBatchLayer.h b/arm_compute/runtime/CL/functions/CLSpaceToBatchLayer.h
index dc02fa1..304a741 100644
--- a/arm_compute/runtime/CL/functions/CLSpaceToBatchLayer.h
+++ b/arm_compute/runtime/CL/functions/CLSpaceToBatchLayer.h
@@ -60,6 +60,15 @@
~CLSpaceToBatchLayer();
/** Set the input and output tensors.
*
+ * Valid data layouts:
+ * - NHWC
+ * - NCHW
+ *
+ * Valid data type configurations:
+ * |src0 |src1 |src2 |dst |
+ * |:---------|:---------|:---------|:---------|
+ * |All |S32 |S32 |All |
+ *
* @param[in] input Tensor input. Supported tensor rank: 4. Data types supported: All.
* @param[in] block_shape 1-D tensor with shape [M]. Supported M: 2. Data types supported: S32
* @param[in] paddings 2-D tensor with shape [2, M] (First dimension is the fastest-changing dimension). Supported M: 2. Data types supported: S32
diff --git a/arm_compute/runtime/CL/functions/CLSpaceToDepthLayer.h b/arm_compute/runtime/CL/functions/CLSpaceToDepthLayer.h
index 9e476fe..8a47e95 100644
--- a/arm_compute/runtime/CL/functions/CLSpaceToDepthLayer.h
+++ b/arm_compute/runtime/CL/functions/CLSpaceToDepthLayer.h
@@ -1,5 +1,5 @@
/*
- * Copyright (c) 2019-2020 Arm Limited.
+ * Copyright (c) 2019-2021 Arm Limited.
*
* SPDX-License-Identifier: MIT
*
@@ -54,6 +54,15 @@
~CLSpaceToDepthLayer();
/** Set the input and output tensors.
*
+ * Valid data layouts:
+ * - NHWC
+ * - NCHW
+ *
+ * Valid data type configurations:
+ * |src |dst |
+ * |:--------------|:--------------|
+ * |All |All |
+ *
* @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.
diff --git a/arm_compute/runtime/CL/functions/CLSplit.h b/arm_compute/runtime/CL/functions/CLSplit.h
index 2931203..86c7bdd 100644
--- a/arm_compute/runtime/CL/functions/CLSplit.h
+++ b/arm_compute/runtime/CL/functions/CLSplit.h
@@ -1,5 +1,5 @@
/*
- * Copyright (c) 2018-2020 Arm Limited.
+ * Copyright (c) 2018-2021 Arm Limited.
*
* SPDX-License-Identifier: MIT
*
@@ -40,6 +40,18 @@
class CLSplit : public CPPSplit<CLSlice, ICLTensor>
{
public:
+ /** CLSplit
+ *
+ * Valid data layouts:
+ * - All
+ *
+ * Valid data type configurations:
+ * |src |dst |
+ * |:--------------|:--------------|
+ * |All |All |
+ *
+ */
+
// Inherited methods overridden:
void run() override;
};
diff --git a/arm_compute/runtime/CL/functions/CLStackLayer.h b/arm_compute/runtime/CL/functions/CLStackLayer.h
index 3861fd2..54c903a 100644
--- a/arm_compute/runtime/CL/functions/CLStackLayer.h
+++ b/arm_compute/runtime/CL/functions/CLStackLayer.h
@@ -1,5 +1,5 @@
/*
- * Copyright (c) 2018-2020 Arm Limited.
+ * Copyright (c) 2018-2021 Arm Limited.
*
* SPDX-License-Identifier: MIT
*
@@ -59,6 +59,14 @@
~CLStackLayer();
/** Initialise the kernel's inputs vector and output.
*
+ * Valid data layouts:
+ * - All
+ *
+ * Valid data type configurations:
+ * |src |dst |
+ * |:--------------|:--------------|
+ * |All |All |
+ *
* @note Supported input tensor rank: up to 4
*
* @param[in] input The vectors containing all the tensors with the same shape to stack. Data types supported: All.
diff --git a/arm_compute/runtime/CL/functions/CLTile.h b/arm_compute/runtime/CL/functions/CLTile.h
index 6974369..c266adb 100644
--- a/arm_compute/runtime/CL/functions/CLTile.h
+++ b/arm_compute/runtime/CL/functions/CLTile.h
@@ -1,5 +1,5 @@
/*
- * Copyright (c) 2018-2020 Arm Limited.
+ * Copyright (c) 2018-2021 Arm Limited.
*
* SPDX-License-Identifier: MIT
*
@@ -39,6 +39,14 @@
public:
/** Set the source, destination of the kernel
*
+ * Valid data layouts:
+ * - All
+ *
+ * Valid data type configurations:
+ * |src |dst |
+ * |:--------------|:--------------|
+ * |All |All |
+ *
* @param[in] input Source tensor. Data type supported: All.
* @param[in] multiples Contains the number of times the input tensor should be replicated on the given dimension.
* @param[out] output Destination tensor. Same as @p input
diff --git a/arm_compute/runtime/CL/functions/CLUnstack.h b/arm_compute/runtime/CL/functions/CLUnstack.h
index 5d4d571..32ad439 100644
--- a/arm_compute/runtime/CL/functions/CLUnstack.h
+++ b/arm_compute/runtime/CL/functions/CLUnstack.h
@@ -1,5 +1,5 @@
/*
- * Copyright (c) 2018-2020 Arm Limited.
+ * Copyright (c) 2018-2021 Arm Limited.
*
* SPDX-License-Identifier: MIT
*
@@ -48,6 +48,14 @@
CLUnstack();
/** Set the input, output and unstacking axis.
*
+ * Valid data layouts:
+ * - All
+ *
+ * Valid data type configurations:
+ * |src |dst |
+ * |:--------------|:--------------|
+ * |All |All |
+ *
* @param[in] input A tensor to be unstacked. Data type supported: All.
* @param[in,out] output_vector A vector of tensors. Data types supported: same as @p input.
* Note: The number of elements of the vector will be used as the number of slices to be taken from the axis.
diff --git a/arm_compute/runtime/CL/functions/CLWinogradConvolutionLayer.h b/arm_compute/runtime/CL/functions/CLWinogradConvolutionLayer.h
index 9ced69c..7b42932 100644
--- a/arm_compute/runtime/CL/functions/CLWinogradConvolutionLayer.h
+++ b/arm_compute/runtime/CL/functions/CLWinogradConvolutionLayer.h
@@ -1,5 +1,5 @@
/*
- * Copyright (c) 2018-2020 Arm Limited.
+ * Copyright (c) 2018-2021 Arm Limited.
*
* SPDX-License-Identifier: MIT
*
@@ -62,6 +62,16 @@
~CLWinogradConvolutionLayer();
/** Set the input and output tensors.
*
+ * Valid data layouts:
+ * - NHWC
+ * - NCHW
+ *
+ * Valid data type configurations:
+ * |src0 |src1 |src2 |dst |
+ * |:--------------|:--------------|:------|:--------------|
+ * |F16 |F16 |F16 |F16 |
+ * |F32 |F32 |F32 |F32 |
+ *
* @note: This function only works with 3x3,3x1,1x3,5x5,5x1,1x5,7x1 and 1x7 kernels along with unit strides for both NCHW and NHWC data layout
* @note Some Winograd configurations (i.e. F(4x4, 5x5)) are supported only with enable_fast_math = true
*