diff options
Diffstat (limited to 'arm_compute/runtime/NEON')
9 files changed, 170 insertions, 4 deletions
diff --git a/arm_compute/runtime/NEON/functions/NEArithmeticAddition.h b/arm_compute/runtime/NEON/functions/NEArithmeticAddition.h index 734e3502dd..b8e46ff36e 100644 --- a/arm_compute/runtime/NEON/functions/NEArithmeticAddition.h +++ b/arm_compute/runtime/NEON/functions/NEArithmeticAddition.h @@ -52,8 +52,7 @@ public: /** Initialise the kernel's inputs, output and conversion policy. * * Valid data layouts: - * - NHWC - * - NCHW + * - All * * Valid data type configurations: * |src0 |src1 |dst | diff --git a/arm_compute/runtime/NEON/functions/NEArithmeticSubtraction.h b/arm_compute/runtime/NEON/functions/NEArithmeticSubtraction.h index c741db3223..0c72e946f6 100644 --- a/arm_compute/runtime/NEON/functions/NEArithmeticSubtraction.h +++ b/arm_compute/runtime/NEON/functions/NEArithmeticSubtraction.h @@ -57,6 +57,25 @@ public: NEArithmeticSubtraction &operator=(NEArithmeticSubtraction &&); /** Initialise the kernel's inputs, output and conversion policy. * + * Valid data layouts: + * - All + * + * Valid data type configurations: + * |src0 |src1 |dst | + * |:--------------|:--------------|:--------------| + * |QASYMM8 |QASYMM8 |QASYMM8 | + * |QASYMM8_SIGNED |QASYMM8_SIGNED |QASYMM8_SIGNED | + * |QSYMM16 |QSYMM16 |QASYMM16 | + * |QSYMM16 |QSYMM16 |S32 | + * |U8 |U8 |U8 | + * |U8 |U8 |S16 | + * |U8 |S16 |S16 | + * |S16 |U8 |S16 | + * |S16 |S16 |S16 | + * |S32 |S32 |S32 | + * |F16 |F16 |F16 | + * |F32 |F32 |F32 | + * * @param[in] input1 First tensor input. Data types supported: U8/QASYMM8/QASYMM8_SIGNED/QSYMM16/S16/S32/F16/F32 * @param[in] input2 Second tensor input. Data types supported: U8/QASYMM8/QASYMM8_SIGNED/QSYMM16/S16/S32/F16/F32 * @param[out] output Output tensor. Data types supported: U8/QASYMM8/QASYMM8_SIGNED/QSYMM16/S16/S32/F16/F32 diff --git a/arm_compute/runtime/NEON/functions/NEDetectionPostProcessLayer.h b/arm_compute/runtime/NEON/functions/NEDetectionPostProcessLayer.h index d5c1f0ab6f..2affa8d49e 100644 --- a/arm_compute/runtime/NEON/functions/NEDetectionPostProcessLayer.h +++ b/arm_compute/runtime/NEON/functions/NEDetectionPostProcessLayer.h @@ -1,5 +1,5 @@ /* - * Copyright (c) 2019-2020 Arm Limited. + * Copyright (c) 2019-2021 Arm Limited. * * SPDX-License-Identifier: MIT * @@ -57,6 +57,16 @@ public: ~NEDetectionPostProcessLayer() = default; /** Configure the detection output layer NE function * + * Valid data layouts: + * - All + * + * Valid data type configurations: + * |src0 - src2 |dst0 - dst3 | + * |:--------------|:--------------| + * |QASYMM8 |F32 | + * |QASYMM8_SIGNED |F32 | + * |F32 |F32 | + * * @param[in] input_box_encoding The bounding box input tensor. Data types supported: QASYMM8/QASYMM8_SIGNED/F32. * @param[in] input_score The class prediction input tensor. Data types supported: same as @p input_box_encoding. * @param[in] input_anchors The anchors input tensor. Data types supported: same as @p input_box_encoding. diff --git a/arm_compute/runtime/NEON/functions/NEElementwiseOperations.h b/arm_compute/runtime/NEON/functions/NEElementwiseOperations.h index 44b70bbe85..95274bdb0c 100644 --- a/arm_compute/runtime/NEON/functions/NEElementwiseOperations.h +++ b/arm_compute/runtime/NEON/functions/NEElementwiseOperations.h @@ -54,6 +54,19 @@ public: NEElementwiseMax &operator=(NEElementwiseMax &&); /** Initialise the kernel's inputs, output and conversion policy. * + * Valid data layouts: + * - All + * + * Valid data type configurations: + * |src0 |src1 |dst | + * |:--------------|:--------------|:--------------| + * |QASYMM8 |QASYMM8 |QASYMM8 | + * |QASYMM8_SIGNED |QASYMM8_SIGNED |QASYMM8_SIGNED | + * |S32 |S32 |S32 | + * |S16 |S16 |S16 | + * |F16 |F16 |F16 | + * |F32 |F32 |F32 | + * * @param[in, out] input1 First tensor input. Data types supported: QASYMM8/QASYMM8_SIGNED/S16/F16/S32/F32. * @param[in, out] input2 Second tensor input. Data types supported: Same as @p input1. * @param[out] output Output tensor. Data types supported: Same as @p input1. @@ -101,6 +114,19 @@ public: NEElementwiseMin &operator=(NEElementwiseMin &&); /** Initialise the kernel's inputs, output and conversion policy. * + * Valid data layouts: + * - All + * + * Valid data type configurations: + * |src0 |src1 |dst | + * |:--------------|:--------------|:--------------| + * |QASYMM8 |QASYMM8 |QASYMM8 | + * |QASYMM8_SIGNED |QASYMM8_SIGNED |QASYMM8_SIGNED | + * |S32 |S32 |S32 | + * |S16 |S16 |S16 | + * |F16 |F16 |F16 | + * |F32 |F32 |F32 | + * * @param[in, out] input1 First tensor input. Data types supported: QASYMM8/QASYMM8_SIGNED/S16/F16/S32/F32. * @param[in, out] input2 Second tensor input. Data types supported: Same as @p input1. * @param[out] output Output tensor. Data types supported: Same as @p input1. @@ -148,6 +174,19 @@ public: NEElementwiseSquaredDiff &operator=(NEElementwiseSquaredDiff &&); /** Initialise the kernel's inputs, output and conversion policy. * + * Valid data layouts: + * - All + * + * Valid data type configurations: + * |src0 |src1 |dst | + * |:--------------|:--------------|:--------------| + * |QASYMM8 |QASYMM8 |QASYMM8 | + * |QASYMM8_SIGNED |QASYMM8_SIGNED |QASYMM8_SIGNED | + * |S32 |S32 |S32 | + * |S16 |S16 |S16 | + * |F16 |F16 |F16 | + * |F32 |F32 |F32 | + * * @param[in, out] input1 First tensor input. Data types supported: QASYMM8/QASYMM8_SIGNED/S16/F16/S32/F32. * @param[in, out] input2 Second tensor input. Data types supported: Same as @p input1. * @param[out] output Output tensor. Data types supported: Same as @p input1. @@ -195,6 +234,15 @@ public: NEElementwiseDivision &operator=(NEElementwiseDivision &&); /** Initialise the kernel's inputs, output and conversion policy. * + * Valid data layouts: + * - All + * + * Valid data type configurations: + * |src0 |src1 |dst | + * |:--------------|:--------------|:--------------| + * |F16 |F16 |F16 | + * |F32 |F32 |F32 | + * * @param[in, out] input1 First tensor input. Data types supported: F16/F32. * @param[in, out] input2 Second tensor input. Data types supported: Same as @p input1. * @param[out] output Output tensor. Data types supported: Same as @p input1. @@ -243,6 +291,15 @@ public: NEElementwisePower &operator=(NEElementwisePower &&); /** Initialise the kernel's inputs, output and conversion policy. * + * Valid data layouts: + * - All + * + * Valid data type configurations: + * |src0 |src1 |dst | + * |:--------------|:--------------|:--------------| + * |F16 |F16 |F16 | + * |F32 |F32 |F32 | + * * @param[in, out] input1 First tensor input. Data types supported: F16/F32. * @param[in, out] input2 Second tensor input. Data types supported: Same as @p input1. * @param[out] output Output tensor. Data types supported: Same as @p input1. @@ -290,6 +347,20 @@ public: NEElementwiseComparison &operator=(NEElementwiseComparison &&); /** Initialise the kernel's inputs, output and conversion policy. * + * Valid data layouts: + * - All + * + * Valid data type configurations: + * |src0 |src1 |dst | + * |:--------------|:--------------|:-----| + * |QASYMM8 |QASYMM8 |U8 | + * |QASYMM8_SIGNED |QASYMM8_SIGNED |U8 | + * |S32 |S32 |U8 | + * |U8 |U8 |U8 | + * |S16 |S16 |U8 | + * |F16 |F16 |U8 | + * |F32 |F32 |U8 | + * * @param[in, out] input1 First tensor input. Data types supported: U8/QASYMM8/QASYMM8_SIGNED/S16/F16/S32/F32. * @param[in, out] input2 Second tensor input. Data types supported: Same as @p input1. * @param[out] output Output tensor. Data types supported: U8. diff --git a/arm_compute/runtime/NEON/functions/NEElementwiseUnaryLayer.h b/arm_compute/runtime/NEON/functions/NEElementwiseUnaryLayer.h index 4786f71cf8..63e47b8377 100644 --- a/arm_compute/runtime/NEON/functions/NEElementwiseUnaryLayer.h +++ b/arm_compute/runtime/NEON/functions/NEElementwiseUnaryLayer.h @@ -54,6 +54,16 @@ public: /** Initialize the function * + * Valid data layouts: + * - All + * + * Valid data type configurations: + * |src |dst | + * |:--------------|:--------------| + * |F16 |F16 | + * |F32 |F32 | + * |S32 |S32 | + * * @param[in] input Input tensor. Data types supported: F16/F32, F16/F32/S32 for NEG/ABS operations. * @param[out] output Output tensor. Data types supported: Same as @p input. */ diff --git a/arm_compute/runtime/NEON/functions/NEGEMMConv2d.h b/arm_compute/runtime/NEON/functions/NEGEMMConv2d.h index 8c3ba4f0c8..b2ffd038de 100644 --- a/arm_compute/runtime/NEON/functions/NEGEMMConv2d.h +++ b/arm_compute/runtime/NEON/functions/NEGEMMConv2d.h @@ -65,6 +65,18 @@ public: ~NEGEMMConv2d(); /** Set the input and output tensors. * + * Valid data layouts: + * - All + * + * Valid data type configurations: + * |src0 |src1 |src2 |dst | + * |:--------------|:--------------|:--------------|:--------------| + * |QASYMM8 |QASYMM8 |S32 |QASYMM8 | + * |QASYMM8_SIGNED |QASYMM8_SIGNED |S32 |QASYMM8_SIGNED | + * |F16 |F16 |F16 |F16 | + * |F32 |F32 |F32 |F32 | + * |BFLOAT16 |BFLOAT16 |BFLOAT16 |BFLOAT16 | + * * @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/BFLOAT16/F16/F32. diff --git a/arm_compute/runtime/NEON/functions/NEGEMMLowpOutputStage.h b/arm_compute/runtime/NEON/functions/NEGEMMLowpOutputStage.h index c22ed1b5c4..fa5f5e3826 100644 --- a/arm_compute/runtime/NEON/functions/NEGEMMLowpOutputStage.h +++ b/arm_compute/runtime/NEON/functions/NEGEMMLowpOutputStage.h @@ -282,6 +282,16 @@ public: ~NEGEMMLowpOutputStage(); /** Initialise the kernel's inputs, output * + * Valid data layouts: + * - All + * + * Valid data type configurations: + * |src0 |src1 |dst | + * |:--------------|:-------------|:-------------| + * |S32 |S32 |QASYMM8 | + * |S32 |S32 |QASYMM8_SIGNED| + * |S32 |S32 |QSYMM16 | + * * @param[in] input Input tensor. Data type supported: S32 * @param[in] bias Biases tensor. Only shared biases supported and it can be a nullptr if the biases addition is not required. * Biases are 1D tensor with dimensions [OFM]. Data type supported: Same as @p input. diff --git a/arm_compute/runtime/NEON/functions/NELogical.h b/arm_compute/runtime/NEON/functions/NELogical.h index 04ffce6221..5cf5336f4f 100644 --- a/arm_compute/runtime/NEON/functions/NELogical.h +++ b/arm_compute/runtime/NEON/functions/NELogical.h @@ -1,5 +1,5 @@ /* - * Copyright (c) 2020 Arm Limited. + * Copyright (c) 2020-2021 Arm Limited. * * SPDX-License-Identifier: MIT * @@ -48,6 +48,14 @@ public: /** Initialise the kernel's inputs and output * + * Valid data layouts: + * - All + * + * Valid data type configurations: + * |src0 |src1 |dst | + * |:--------------|:-------------|:------------| + * |U8 |U8 |U8 | + * * @param[in] input1 First tensor input. Data type supported: U8. * @param[in] input2 Second tensor input. Data type supported: U8. * @param[out] output Output tensor. Data type supported: U8. @@ -83,6 +91,14 @@ public: /** Initialise the kernel's inputs and output * + * Valid data layouts: + * - All + * + * Valid data type configurations: + * |src0 |src1 |dst | + * |:--------------|:-------------|:------------| + * |U8 |U8 |U8 | + * * @param[in] input1 First tensor input. Data type supported: U8. * @param[in] input2 Second tensor input. Data type supported: U8. * @param[out] output Output tensor. Data type supported: U8. @@ -118,6 +134,14 @@ public: /** Initialise the kernel's inputs and output * + * Valid data layouts: + * - All + * + * Valid data type configurations: + * |src |dst | + * |:--------------|:-------------| + * |U8 |U8 | + * * @param[in] input Input tensor. Data type supported: U8. * @param[out] output Output tensor. Data type supported: U8. */ diff --git a/arm_compute/runtime/NEON/functions/NESoftmaxLayer.h b/arm_compute/runtime/NEON/functions/NESoftmaxLayer.h index 8a2ae10129..efe959f14e 100644 --- a/arm_compute/runtime/NEON/functions/NESoftmaxLayer.h +++ b/arm_compute/runtime/NEON/functions/NESoftmaxLayer.h @@ -52,6 +52,17 @@ public: ~NESoftmaxLayerGeneric(); /** 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 | + * * @param[in,out] input Source tensor. Data types supported: QASYMM8/QASYMM8_SIGNED/F16/F32. If the width is not a * multiple of the internal processing block size, @ref NEFillBorder replicates the * last value of each row to the nearest multiple. |