From c9309f22a026dfce92365e2f0802c40e8e1c449e Mon Sep 17 00:00:00 2001 From: Sang-Hoon Park Date: Wed, 5 May 2021 10:34:47 +0100 Subject: Restructure Documentation (Part 2) The followings are done: - Move operator list documnetation - Introduction page is moved to the top - The sections for experimental API and programming model are merged into library architecture page. Resolves: COMPMID-4198 Change-Id: Iad824d6c8ba8d31e0bf76afd3fb67abbe32a1667 Signed-off-by: Sang-Hoon Park Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/5570 Tested-by: Arm Jenkins Reviewed-by: Michele Di Giorgio Reviewed-by: Sheri Zhang Comments-Addressed: Arm Jenkins --- docs/user_guide/operator_list.dox | 3123 +++++++++++++++++++++++++++++++++++++ 1 file changed, 3123 insertions(+) create mode 100644 docs/user_guide/operator_list.dox (limited to 'docs/user_guide/operator_list.dox') diff --git a/docs/user_guide/operator_list.dox b/docs/user_guide/operator_list.dox new file mode 100644 index 0000000000..fc41265738 --- /dev/null +++ b/docs/user_guide/operator_list.dox @@ -0,0 +1,3123 @@ +/// +/// Copyright (c) 2021 Arm Limited. +/// +/// SPDX-License-Identifier: MIT +/// +/// Permission is hereby granted, free of charge, to any person obtaining a copy +/// of this software and associated documentation files (the "Software"), to +/// deal in the Software without restriction, including without limitation the +/// rights to use, copy, modify, merge, publish, distribute, sublicense, and/or +/// sell copies of the Software, and to permit persons to whom the Software is +/// furnished to do so, subject to the following conditions: +/// +/// The above copyright notice and this permission notice shall be included in all +/// copies or substantial portions of the Software. +/// +/// THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR +/// IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, +/// FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE +/// AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER +/// LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, +/// OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE +/// SOFTWARE. +/// +namespace arm_compute +{ +/** +@page operators_list Supported Operators + +@tableofcontents + +@section S9_1_operators_list Supported Operators + +Compute Library supports operators that are listed in below table. + +Compute Library supports a wide list of data-types, information can been directly found in the documentation of each kernel/function. +The main data-types that the Machine Learning functions support are the following: +
    +
  • BFLOAT16: 16-bit non-standard brain floating point +
  • QASYMM8: 8-bit unsigned asymmetric quantized +
  • QASYMM8_SIGNED: 8-bit signed asymmetric quantized +
  • QSYMM8_PER_CHANNEL: 8-bit signed symmetric quantized (Used for the weights) +
  • QSYMM8: 8-bit unsigned symmetric quantized +
  • QSYMM16: 16-bit unsigned symmetric quantized +
  • F32: 32-bit single precision floating point +
  • F16: 16-bit half precision floating point +
  • S32: 32-bit signed integer +
  • U8: 8-bit unsigned char +
  • All: include all above data types +
+ +Compute Library supports the following data layouts (fast changing dimension from right to left): +
    +
  • NHWC: The native layout of Compute Library that delivers the best performance where channels are in the fastest changing dimension +
  • NCHW: Legacy layout where width is in the fastest changing dimension +
  • All: include all above data layouts +
+where N = batches, C = channels, H = height, W = width + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + +
Function + Description + Equivalent Android NNAPI Op + Backends + Data Layouts + Data Types +
ActivationLayer + Function to simulate an activation layer with the specified activation function. + +
    +
  • ANEURALNETWORKS_ELU +
  • ANEURALNETWORKS_HARD_SWISH +
  • ANEURALNETWORKS_LOGISTIC +
  • ANEURALNETWORKS_RELU +
  • ANEURALNETWORKS_RELU1 +
  • ANEURALNETWORKS_RELU6 +
  • ANEURALNETWORKS_TANH +
+ 		NEActivationLayer + +
    +
  • All +
+ 		+ +
srcdst +
QASYMM8QASYMM8 +
QASYMM8_SIGNEDQASYMM8_SIGNED +
QSYMM16QSYMM16 +
F16F16 +
F32F32 +
+
CLActivationLayer + +
    +
  • All +
+ 		+ +
srcdst +
QASYMM8QASYMM8 +
QASYMM8_SIGNEDQASYMM8_SIGNED +
QSYMM16QSYMM16 +
F16F16 +
F32F32 +
+
ArgMinMaxLayer + Function to calculate the index of the minimum or maximum values in a tensor based on an axis. + +
    +
  • ANEURALNETWORKS_ARGMAX +
  • ANEURALNETWORKS_ARGMIN +
+ 		NEArgMinMaxLayer + +
    +
  • All +
+ 		+ +
srcdst +
QASYMM8U32, S32 +
QASYMM8_SIGNEDU32, S32 +
S32U32, S32 +
F16U32, S32 +
F32U32, S32 +
+
CLArgMinMaxLayer + +
    +
  • All +
+ 		+ +
srcdst +
QASYMM8U32, S32 +
QASYMM8_SIGNEDU32, S32 +
S32U32, S32 +
F16U32, S32 +
F32U32, S32 +
+
ArithmeticAddition + Function to add 2 tensors. + +
    +
  • ANEURALNETWORKS_ADD +
+ 		NEArithmeticAddition + +
    +
  • All +
+ 		+ +
src0src1dst +
QASYMM8QASYMM8QASYMM8 +
QASYMM8_SIGNEDQASYMM8_SIGNEDQASYMM8_SIGNED +
QSYMM16QSYMM16QASYMM16 +
QSYMM16QSYMM16S32 +
U8U8U8 +
U8U8S16 +
U8S16S16 +
S16U8S16 +
S16S16S16 +
S32S32S32 +
F16F16F16 +
F32F32F32 +
+
ArithmeticSubtraction + Function to substract 2 tensors. + +
    +
  • ANEURALNETWORKS_SUB +
+ 		NEArithmeticSubtraction + +
    +
  • All +
+ 		+ +
src0src1dst +
QASYMM8QASYMM8QASYMM8 +
QASYMM8_SIGNEDQASYMM8_SIGNEDQASYMM8_SIGNED +
QSYMM16QSYMM16QASYMM16 +
QSYMM16QSYMM16S32 +
U8U8U8 +
U8U8S16 +
U8S16S16 +
S16U8S16 +
S16S16S16 +
S32S32S32 +
F16F16F16 +
F32F32F32 +
+
BatchNormalizationLayer + Function to perform batch normalization. + +
    +
  • n/a +
+ 		NEBatchNormalizationLayer + +
    +
  • NHWC +
  • NCHW +
+ 		+ +
srcdst +
F32F32 +
F16F16 +
+
CLBatchNormalizationLayer + +
    +
  • NHWC +
  • NCHW +
+ 		+ +
srcdst +
F32F32 +
F16F16 +
+
BatchToSpaceLayer + Batch to space transformation. + +
    +
  • ANEURALNETWORKS_BATCH_TO_SPACE_ND +
+ 		NEBatchToSpaceLayer + +
    +
  • NHWC +
  • NCHW +
+ 		+ +
src0src1dst +
Alls32All +
+
CLBatchToSpaceLayer + +
    +
  • NHWC +
  • NCHW +
+ 		+ +
src0src1dst +
Alls32All +
+
BitwiseAnd + Function to performe bitwise AND between 2 tensors. + +
    +
  • ANEURALNETWORKS_LOGICAL_AND +
+ 		NEBitwiseAnd + +
    +
  • All +
+ 		+ +
srcdst +
U8U8 +
+
CLBitwiseAnd + +
    +
  • All +
+ 		+ +
srcdst +
U8U8 +
+
BitwiseNot + Function to performe bitwise NOT. + +
    +
  • ANEURALNETWORKS_LOGICAL_NOT +
+ 		NEBitwiseNot + +
    +
  • All +
+ 		+ +
srcdst +
U8U8 +
+
CLBitwiseNot + +
    +
  • All +
+ 		+ +
srcdst +
U8U8 +
+
BitwiseOr + Function to performe bitwise OR between 2 tensors. + +
    +
  • ANEURALNETWORKS_LOGICAL_OR +
+ 		NEBitwiseOr + +
    +
  • All +
+ 		+ +
srcdst +
U8U8 +
+
CLBitwiseOr + +
    +
  • All +
+ 		+ +
srcdst +
U8U8 +
+
BitwiseXor + Function to performe bitwise XOR between 2 tensors. + +
    +
  • n/a +
+ 		NEBitwiseXor + +
    +
  • All +
+ 		+ +
srcdst +
U8U8 +
+
CLBitwiseXor + +
    +
  • All +
+ 		+ +
srcdst +
U8U8 +
+
BoundingBoxTransform + Transform proposal bounding boxes to target bounding box using bounding box deltas. + +
    +
  • n/a +
+ 		NEBoundingBoxTransform + +
    +
  • NHWC +
  • NCHW +
+ 		+ +
src0src1dst +
QASYMM16QASYMM8QASYMM16 +
F16F16F16 +
F32F32F32 +
+
CLBoundingBoxTransform + +
    +
  • NHWC +
  • NCHW +
+ 		+ +
src0src1dst +
QASYMM16QASYMM8QASYMM16 +
F16F16F16 +
F32F32F32 +
+
Cast + Function to cast a tensor. + +
    +
  • ANEURALNETWORKS_CAST +
+ 		NECast + +
    +
  • All +
+ 		+ +
srcdst +
QASYMM8_SIGNEDS16, S32, F32, F16 +
QASYMM8U16, S16, S32, F32, F16 +
U8U16, S16, S32, F32, F16 +
U16U8, U32 +
S16QASYMM8_SIGNED, U8, S32 +
F16QASYMM8_SIGNED, QASYMM8, F32, S32, U8 +
S32QASYMM8_SIGNED, QASYMM8, F16, F32, U8 +
F32QASYMM8_SIGNED, QASYMM8, BFLOAT16, F16, S32, U8 +
+
CLCast + +
    +
  • All +
+ 		+ +
srcdst +
U8S8, U16, S16, U32, S32, F16, F32 +
U16U8, S8, S16, U32, S32, F16, F32 +
S16U8, S8, U16, U32, S32, F16, F32 +
U32U8, S8, U16, S16, S32, F16, F32 +
S32U8, S8, U16, S16, U32, F16, F32 +
F16U8, S8, U16, S16, U32, F32 +
F32U8, S8, U16, S16, U32, F16 +
+
ChannelShuffleLayer + Function to shuffle the channels of the input tensor. + +
    +
  • ANEURALNETWORKS_CHANNEL_SHUFFLE +
+ 		NEChannelShuffleLayer + +
    +
  • NCHW +
+ 		+ +
srcdst +
AllAll +
+
CLChannelShuffleLayer + +
    +
  • NCHW +
+ 		+ +
srcdst +
AllAll +
+
Comparison + Function to compare 2 tensors. + +
    +
  • ANEURALNETWORKS_EQUAL +
  • ANEURALNETWORKS_GREATER +
  • ANEURALNETWORKS_GREATER_EQUAL +
  • ANEURALNETWORKS_LESS +
  • ANEURALNETWORKS_LESS_EQUAL +
  • ANEURALNETWORKS_NOT_EQUAL +
+ 		CLComparison + +
    +
  • All +
+ 		+ +
src0src1dst +
AllAllU8 +
+
ConcatenateLayer + Function to concatenate tensors along a given axis. + +
    +
  • ANEURALNETWORKS_CONCATENATION +
+ 		NEConcatenateLayer + +
    +
  • All +
+ 		+ +
srcdst +
QASYMM8QASYMM8 +
QASYMM8_SIGNEDQASYMM8_SIGNED +
F16F16 +
F32F32 +
+
CLConcatenateLayer + +
    +
  • All +
+ 		+ +
srcdst +
QASYMM8QASYMM8 +
QASYMM8_SIGNEDQASYMM8_SIGNED +
F16F16 +
F32F32 +
+
ConvertFullyConnectedWeights + Function to tranpose the wieghts for the fully connected layer. + +
    +
  • n/a +
+ 		NEConvertFullyConnectedWeights + +
    +
  • NHWC +
  • NCHW +
+ 		+ +
srcdst +
AllAll +
+
CLConvertFullyConnectedWeights + +
    +
  • NHWC +
  • NCHW +
+ 		+ +
srcdst +
AllAll +
+
ConvolutionLayer + Function to compute a convolution layer. + +
    +
  • ANEURALNETWORKS_CONV_2D +
+ 		NEConvolutionLayer + +
    +
  • NHWC +
  • NCHW +
+ 		+ +
src0src1src2dst +
F16F16F16F16 +
F32F32F32F32 +
QASYMM8QASYMM8S32QASYMM8 +
QASYMM8QSYMM8_PER_CHANNELS32QASYMM8 +
QASYMM8_SIGNEDQASYMM8_SIGNEDS32QASYMM8_SIGNED +
QASYMM8_SIGNEDQSYMM8_PER_CHANNELS32QASYMM8_SIGNED +
+
CLConvolutionLayer + +
    +
  • NHWC +
  • NCHW +
+ 		+ +
src0src1src2dst +
F16F16F16F16 +
F32F32F32F32 +
QASYMM8QASYMM8S32QASYMM8 +
QASYMM8QSYMM8_PER_CHANNELS32QASYMM8 +
QASYMM8_SIGNEDQASYMM8_SIGNEDS32QASYMM8_SIGNED +
QASYMM8_SIGNEDQSYMM8_PER_CHANNELS32QASYMM8_SIGNED +
+
Copy + Function to copy a tensor. + +
    +
  • n/a +
+ 		NECopy + +
    +
  • All +
+ 		+ +
srcdst +
AllAll +
+
CLCopy + +
    +
  • All +
+ 		+ +
srcdst +
AllAll +
+
Crop + Performs a copy of input tensor to the output tensor. + +
    +
  • n/a +
+ 		CLCrop + +
    +
  • NHWC +
+ 		+ +
srcdst +
AllF32 +
+
CropResize + Function to perform cropping and resizing. + +
    +
  • n/a +
+ 		NECropResize + +
    +
  • NHWC +
+ 		+ +
src0src1src2dst +
AllF32F32F32 +
+
CLCropResize + +
    +
  • NHWC +
+ 		+ +
src0src1src2dst +
AllF32F32F32 +
+
DeconvolutionLayer + Function to compute a deconvolution or tranpose convolution. + +
    +
  • ANEURALNETWORKS_TRANSPOSE_CONV_2D +
+ 		NEDeconvolutionLayer + +
    +
  • NHWC +
  • NCHW +
+ 		+ +
src0src1src2dst +
F16F16F16F16 +
F32F32F32F32 +
QASYMM8QASYMM8S32QASYMM8 +
QASYMM8QSYMM8_PER_CHANNELS32QASYMM8 +
QASYMM8_SIGNEDQASYMM8_SIGNEDS32QASYMM8_SIGNED +
QASYMM8_SIGNEDQSYMM8_PER_CHANNELS32QASYMM8_SIGNED +
+
CLDeconvolutionLayer + +
    +
  • NHWC +
  • NCHW +
+ 		+ +
src0src1src2dst +
F16F16F16F16 +
F32F32F32F32 +
QASYMM8QASYMM8S32QASYMM8 +
QASYMM8QSYMM8_PER_CHANNELS32QASYMM8 +
QASYMM8_SIGNEDQASYMM8_SIGNEDS32QASYMM8_SIGNED +
QASYMM8_SIGNEDQSYMM8_PER_CHANNELS32QASYMM8_SIGNED +
+
DeconvolutionLayerUpsample + Function to execute deconvolution upsample on OpenCL. + +
    +
  • ANEURALNETWORKS_TRANSPOSE_CONV_2D +
+ 		CLDeconvolutionLayerUpsample + +
    +
  • NHWC +
  • NCHW +
+ 		+ +
srcdst +
AllAll +
+
DepthConvertLayer + Performs a down-scaling depth conversion. + +
    +
  • n/a +
+ 		NEDepthConvertLayer + +
    +
  • All +
+ 		+ +
srcdst +
QASYMM8F16, F32 +
U8U16, S16, S32 +
U16U8, U32 +
S16U8, S32 +
BFLOAT16F32 +
F16QASYMM8, F32 +
F32QASYMM8, F16, BFLOAT16 +
+
CLDepthConvertLayer + +
    +
  • All +
+ 		+ +
srcdst +
U8S8, U16, S16, U32, S32, F16, F32 +
U16U8, S8, S16, U32, S32, F16, F32 +
S16U8, S8, U16, U32, S32, F16, F32 +
U32U8, S8, U16, S16, S32, F16, F32 +
S32U8, S8, U16, S16, U32, F16, F32 +
F16U8, S8, U16, S16, U32, F32 +
F32U8, S8, U16, S16, U32, F16 +
+
DepthToSpaceLayer + Depth to Space transformation. + +
    +
  • ANEURALNETWORKS_DEPTH_TO_SPACE +
+ 		NEDepthToSpaceLayer + +
    +
  • NHWC +
  • NCHW +
+ 		+ +
srcdst +
AllAll +
+
CLDepthToSpaceLayer + +
    +
  • NHWC +
  • NCHW +
+ 		+ +
srcdst +
AllAll +
+
DepthwiseConvolutionLayer + Function to perform depthwise separable convolution. + +
    +
  • ANEURALNETWORKS_DEPTHWISE_CONV_2D +
+ 		NEDepthwiseConvolutionLayer + +
    +
  • NHWC +
  • NCHW +
+ 		+ +
src0src1src2dst +
F16F16F16F16 +
F32F32F32F32 +
QASYMM8QASYMM8S32QASYMM8 +
QASYMM8QSYMM8_PER_CHANNELS32QASYMM8 +
QASYMM8_SIGNEDQASYMM8_SIGNEDS32QASYMM8_SIGNED +
QASYMM8_SIGNEDQSYMM8_PER_CHANNELS32QASYMM8_SIGNED +
+
CLDepthwiseConvolutionLayer + +
    +
  • NHWC +
  • NCHW +
+ 		+ +
src0src1src2dst +
F16F16F16F16 +
F32F32F32F32 +
QASYMM8QASYMM8S32QASYMM8 +
QASYMM8QSYMM8_PER_CHANNELS32QASYMM8 +
QASYMM8_SIGNEDQASYMM8_SIGNEDS32QASYMM8_SIGNED +
QASYMM8_SIGNEDQSYMM8_PER_CHANNELS32QASYMM8_SIGNED +
+
DequantizationLayer + Function to dequantize the values in a tensor. + +
    +
  • ANEURALNETWORKS_DEQUANTIZE +
+ 		NEDequantizationLayer + +
    +
  • All +
+ 		+ +
srcdst +
QASYMM8F16, F32 +
QASYMM8_SIGNEDF16, F32 +
QSYMM8_PER_CHANNELF16, F32 +
QSYMM8F16, F32 +
QSYMM16F16, F32 +
+
CLDequantizationLayer + +
    +
  • All +
+ 		+ +
srcdst +
QASYMM8F16, F32 +
QASYMM8_SIGNEDF16, F32 +
QSYMM8_PER_CHANNELF16, F32 +
QSYMM8F16, F32 +
QSYMM16F16, F32 +
+
DetectionPostProcessLayer + Function to generate the detection output based on center size encoded boxes, class prediction and anchors by doing non maximum suppression (NMS). + +
    +
  • ANEURALNETWORKS_DETECTION_POSTPROCESSING +
+ 		NEDetectionPostProcessLayer + +
    +
  • All +
+ 		+ +
src0 - src2dst0 - dst3 +
QASYMM8F32 +
QASYMM8_SIGNEDF32 +
F32F32 +
+
DirectConvolutionLayer + Function to compute direct convolution. + +
    +
  • ANEURALNETWORKS_CONV_2D +
+ 		NEDirectConvolutionLayer + +
    +
  • NHWC +
  • NCHW +
+ 		+ +
src0src1src2dst +
F16F16F16F16 +
F32F32F32F32 +
+
CLDirectConvolutionLayer + +
    +
  • NHWC +
  • NCHW +
+ 		+ +
src0src1src2dst +
F16F16F16F16 +
F32F32F32F32 +
QASYMM8QASYMM8S32QASYMM8 +
QASYMM8_SIGNEDQASYMM8_SIGNEDS32QASYMM8_SIGNED +
+
DirectDeconvolutionLayer + Function to run the deconvolution layer. + +
    +
  • ANEURALNETWORKS_TRANSPOSE_CONV_2D +
+ 		CLDirectDeconvolutionLayer + +
    +
  • NHWC +
  • NCHW +
+ 		+ +
src0src1src2dst +
F16F16F16F16 +
F32F32F32F32 +
QASYMM8QASYMM8S32QASYMM8 +
QASYMM8_SIGNEDQASYMM8_SIGNEDS32QASYMM8_SIGNED +
QASYMM8QSYMM8_PER_CHANNELS32QASYMM8 +
QASYMM8_SIGNEDQSYMM8_PER_CHANNELS32QASYMM8_SIGNED +
+
ElementWiseOperations + Function to perform in Cpu: - Div - Max - Min - Pow - SquaredDiff - Comparisons (Equal, greater, greater_equal, less, less_equal, not_equal) Function to perform in CL: - Add - Sub - Div - Max - Min - Pow - SquaredDiff + +
    +
  • ANEURALNETWORKS_MAXIMUM +
  • ANEURALNETWORKS_MINIMUM +
  • ANEURALNETWORKS_POW +
  • ANEURALNETWORKS_DIV +
  • ANEURALNETWORKS_ADD +
  • ANEURALNETWORKS_SUB +
  • ANEURALNETWORKS_EQUAL +
  • ANEURALNETWORKS_GREATER +
  • ANEURALNETWORKS_GREATER_EQUAL +
  • ANEURALNETWORKS_LESS +
  • ANEURALNETWORKS_LESS_EQUAL +
  • ANEURALNETWORKS_NOT_EQUAL +
+ 		NEElementwiseMax + +
    +
  • All +
+ 		+ +
src0src1dst +
QASYMM8QASYMM8QASYMM8 +
QASYMM8_SIGNEDQASYMM8_SIGNEDQASYMM8_SIGNED +
S32S32S32 +
S16S16S16 +
F16F16F16 +
F32F32F32 +
+
NEElementwiseMin + +
    +
  • All +
+ 		+ +
src0src1dst +
QASYMM8QASYMM8QASYMM8 +
QASYMM8_SIGNEDQASYMM8_SIGNEDQASYMM8_SIGNED +
S32S32S32 +
S16S16S16 +
F16F16F16 +
F32F32F32 +
+
NEElementwiseSquaredDiff + +
    +
  • All +
+ 		+ +
src0src1dst +
QASYMM8QASYMM8QASYMM8 +
QASYMM8_SIGNEDQASYMM8_SIGNEDQASYMM8_SIGNED +
S32S32S32 +
S16S16S16 +
F16F16F16 +
F32F32F32 +
+
NEElementwiseDivision + +
    +
  • All +
+ 		+ +
src0src1dst +
F16F16F16 +
F32F32F32 +
+
NEElementwisePower + +
    +
  • All +
+ 		+ +
src0src1dst +
F16F16F16 +
F32F32F32 +
+
NEElementwiseComparison + +
    +
  • All +
+ 		+ +
src0src1dst +
QASYMM8QASYMM8U8 +
QASYMM8_SIGNEDQASYMM8_SIGNEDU8 +
S32S32U8 +
U8U8U8 +
S16S16U8 +
F16F16U8 +
F32F32U8 +
+
CLArithmeticAddition + +
    +
  • All +
+ 		+ +
src0src1dst +
QASYMM8QASYMM8QASYMM8 +
QASYMM8_SIGNEDQASYMM8_SIGNEDQASYMM8_SIGNED +
QSYMM16QSYMM16QASYMM16 +
U8U8U8 +
U8U8S16 +
U8S16S16 +
S16U8S16 +
S16S16S16 +
S32S32S32 +
F16F16F16 +
F32F32F32 +
+
CLArithmeticSubtraction + +
    +
  • All +
+ 		+ +
src0src1dst +
QASYMM8QASYMM8QASYMM8 +
QASYMM8_SIGNEDQASYMM8_SIGNEDQASYMM8_SIGNED +
QSYMM16QSYMM16QASYMM16 +
U8U8U8 +
U8U8S16 +
U8S16S16 +
S16U8S16 +
S16S16S16 +
S32S32S32 +
F16F16F16 +
F32F32F32 +
+
CLArithmeticDivision + +
    +
  • All +
+ 		+ +
src0src1dst +
F16F16F16 +
F32F32F32 +
+
CLElementwiseMax + +
    +
  • All +
+ 		+ +
src0src1dst +
QASYMM8QASYMM8QASYMM8 +
QASYMM8_SIGNEDQASYMM8_SIGNEDQASYMM8_SIGNED +
QSYMM16QSYMM16QASYMM16 +
U8U8U8 +
S16S16S16 +
S32S32S32 +
U32U32U32 +
F16F16F16 +
F32F32F32 +
+
CLElementwiseMin + +
    +
  • All +
+ 		+ +
src0src1dst +
QASYMM8QASYMM8QASYMM8 +
QASYMM8_SIGNEDQASYMM8_SIGNEDQASYMM8_SIGNED +
QSYMM16QSYMM16QASYMM16 +
U8U8U8 +
S16S16S16 +
S32S32S32 +
U32U32U32 +
F16F16F16 +
F32F32F32 +
+
CLElementwiseSquaredDiff + +
    +
  • All +
+ 		+ +
src0src1dst +
QASYMM8QASYMM8QASYMM8 +
QASYMM8_SIGNEDQASYMM8_SIGNEDQASYMM8_SIGNED +
QSYMM16QSYMM16QASYMM16 +
U8U8U8 +
S16S16S16 +
F16F16F16 +
F32F32F32 +
+
CLElementwisePower + +
    +
  • All +
+ 		+ +
src0src1dst +
F16F16F16 +
F32F32F32 +
+
ElementwiseUnaryLayer + Function to perform: - Rsqrt - Exp - Neg - Log - Abs - Round - Sin + +
    +
  • ANEURALNETWORKS_ABS +
  • ANEURALNETWORKS_EXP +
  • ANEURALNETWORKS_LOG +
  • ANEURALNETWORKS_NEG +
  • ANEURALNETWORKS_RSQRT +
  • ANEURALNETWORKS_SIN +
+ 		NEElementwiseUnaryLayer + +
    +
  • All +
+ 		+ +
srcdst +
F16F16 +
F32F32 +
S32S32 +
+
CLRsqrtLayer + +
    +
  • All +
+ 		+ +
srcdst +
F16F16 +
F32F32 +
+
CLExpLayer + +
    +
  • All +
+ 		+ +
srcdst +
F16F16 +
F32F32 +
+
CLNegLayer + +
    +
  • All +
+ 		+ +
srcdst +
F16F16 +
F32F32 +
+
CLSinLayer + +
    +
  • All +
+ 		+ +
srcdst +
F16F16 +
F32F32 +
+
CLLogLayer + +
    +
  • All +
+ 		+ +
srcdst +
F16F16 +
F32F32 +
+
CLAbsLayer + +
    +
  • All +
+ 		+ +
srcdst +
F16F16 +
F32F32 +
+
CLRoundLayer + +
    +
  • All +
+ 		+ +
srcdst +
F16F16 +
F32F32 +
+
FFT1D + Fast Fourier Transform 1D. + +
    +
  • n/a +
+ 		NEFFT1D + +
    +
  • All +
+ 		+ +
srcdst +
F32F32 +
+
CLFFT1D + +
    +
  • All +
+ 		+ +
srcdst +
F32F32 +
F16F16 +
+
FFT2D + Fast Fourier Transform 2D. + +
    +
  • n/a +
+ 		NEFFT2D + +
    +
  • All +
+ 		+ +
srcdst +
F32F32 +
+
CLFFT2D + +
    +
  • All +
+ 		+ +
srcdst +
F32F32 +
F16F16 +
+
FFTConvolutionLayer + Fast Fourier Transform Convolution. + +
    +
  • ANEURALNETWORKS_CONV_2D +
+ 		NEFFTConvolutionLayer + +
    +
  • All +
+ 		+ +
srcdst +
F32F32 +
+
CLFFTConvolutionLayer + +
    +
  • All +
+ 		+ +
srcdst +
F32F32 +
F16F16 +
+
Fill + Set the values of a tensor with a given value. + +
    +
  • ANEURALNETWORKS_FILL +
+ 		NEFill + +
    +
  • All +
+ 		+ +
srcdst +
AllAll +
+
CLFill + +
    +
  • All +
+ 		+ +
srcdst +
AllAll +
+
FillBorder + Function to . + +
    +
  • n/a +
+ 		NEFillBorder + +
    +
  • All +
+ 		+ +
srcdst +
AllAll +
+
CLFillBorder + +
    +
  • All +
+ 		+ +
srcdst +
AllAll +
+
FlattenLayer + Reshape a tensor to be 1D + +
    +
  • ANEURALNETWORKS_RESHAPE +
+ 		NEFlattenLayer + +
    +
  • All +
+ 		+ +
srcdst +
AllAll +
+
CLFlattenLayer + +
    +
  • All +
+ 		+ +
srcdst +
AllAll +
+
Floor + Round the value to the lowest number. + +
    +
  • ANEURALNETWORKS_FLOOR +
+ 		NEFloor + +
    +
  • All +
+ 		+ +
srcdst +
F32F32 +
F16F16 +
+
CLFloor + +
    +
  • All +
+ 		+ +
srcdst +
F32F32 +
F16F16 +
+
FullyConnectedLayer + Function to perform a fully connected / dense layer. + +
    +
  • ANEURALNETWORKS_FULLY_CONNECTED +
+ 		NEFullyConnectedLayer + +
    +
  • NHWC +
  • NCHW +
+ 		+ +
src0src1src2dst +
F16F16F16F16 +
F32F32F32F32 +
QASYMM8QASYMM8S32QASYMM8 +
QASYMM8_SIGNEDQASYMM8_SIGNEDS32QASYMM8_SIGNED +
+
CLFullyConnectedLayer + +
    +
  • NHWC +
  • NCHW +
+ 		+ +
src0src1src2dst +
F16F16F16F16 +
F32F32F32F32 +
QASYMM8QASYMM8S32QASYMM8 +
QASYMM8_SIGNEDQASYMM8_SIGNEDS32QASYMM8_SIGNED +
+
FuseBatchNormalization + Function to fuse the batch normalization node to a preceding convolution node. + +
    +
  • n/a +
+ 		NEFuseBatchNormalization + +
    +
  • NHWC +
  • NCHW +
+ 		+ +
srcdst +
F32F32 +
F16F16 +
+
CLFuseBatchNormalization + +
    +
  • NHWC +
  • NCHW +
+ 		+ +
srcdst +
F32F32 +
F16F16 +
+
Gather + Performs the Gather operation along the chosen axis. + +
    +
  • ANEURALNETWORKS_GATHER +
+ 		NEGather + +
    +
  • All +
+ 		+ +
srcdst +
AllAll +
+
CLGather + +
    +
  • All +
+ 		+ +
srcdst +
AllAll +
+
GEMM + General Matrix Multiplication. + +
    +
  • n/a +
+ 		NEGEMM + +
    +
  • All +
+ 		+ +
src0src1src2dst +
F32F32F32F32 +
F16F16F16F16 +
BFLOAT16BFLOAT16BFLOAT16BFLOAT16 +
+
CLGEMM + +
    +
  • All +
+ 		+ +
src0src1src2dst +
F32F32F32F32 +
F16F16F16F16 +
+
GEMMConv2D + General Matrix Multiplication. + +
    +
  • ANEURALNETWORKS_CONV_2D +
+ 		NEGEMMConv2d + +
    +
  • All +
+ 		+ +
src0src1src2dst +
QASYMM8QASYMM8S32QASYMM8 +
QASYMM8_SIGNEDQASYMM8_SIGNEDS32QASYMM8_SIGNED +
F16F16F16F16 +
F32F32F32F32 +
BFLOAT16BFLOAT16BFLOAT16BFLOAT16 +
+
GEMMConvolutionLayer + General Matrix Multiplication. + +
    +
  • ANEURALNETWORKS_CONV_2D +
+ 		NEGEMMConvolutionLayer + +
    +
  • NHWC +
  • NCHW +
+ 		+ +
src0src1src2dst +
F16F16F16F16 +
F32F32F32F32 +
BFLOAT16BFLOAT16BFLOAT16BFLOAT16 +
QASYMM8QASYMM8S32QASYMM8 +
QASYMM8QSYMM8_PER_CHANNELS32QASYMM8 +
QASYMM8_SIGNEDQASYMM8_SIGNEDS32QASYMM8_SIGNED +
QASYMM8_SIGNEDQSYMM8_PER_CHANNELS32QASYMM8_SIGNED +
+
CLGEMMConvolutionLayer + +
    +
  • NHWC +
  • NCHW +
+ 		+ +
src0src1src2dst +
F16F16F16F16 +
F32F32F32F32 +
QASYMM8QASYMM8S32QASYMM8 +
QASYMM8QSYMM8_PER_CHANNELS32QASYMM8 +
QASYMM8_SIGNEDQASYMM8_SIGNEDS32QASYMM8_SIGNED +
QASYMM8_SIGNEDQSYMM8_PER_CHANNELS32QASYMM8_SIGNED +
+
GEMMDeconvolutionLayer + General Matrix Multiplication. + +
    +
  • ANEURALNETWORKS_TRANSPOSE_CONV_2D +
+ 		CLGEMMDeconvolutionLayer + +
    +
  • NHWC +
+ 		+ +
src0src1src2dst +
F16F16F16F16 +
F32F32F32F32 +
QASYMM8QASYMM8S32QASYMM8 +
QASYMM8_SIGNEDQASYMM8_SIGNEDS32QASYMM8_SIGNED +
+
GEMMLowpMatrixMultiplyCore + General Matrix Multiplication. + +
    +
  • n/a +
+ 		NEGEMMLowpMatrixMultiplyCore + +
    +
  • NHWC +
  • NCHW +
+ 		+ +
src0src1src2dst +
QASYMM8QASYMM8S32QASYMM8 +
QASYMM8QSYMM8_PER_CHANNELS32QASYMM8 +
QASYMM8QSYMM8S32QASYMM8 +
QASYMM8QASYMM8S32S32 +
QASYMM8QSYMM8_PER_CHANNELS32S32 +
QASYMM8QSYMM8S32S32 +
QASYMM8_SIGNEDQASYMM8_SIGNEDS32QASYMM8_SIGNED +
QASYMM8_SIGNEDQSYMM8_PER_CHANNELS32QASYMM8_SIGNED +
QASYMM8_SIGNEDQSYMM8S32QASYMM8_SIGNED +
QASYMM8_SIGNEDQASYMM8_SIGNEDS32S32 +
QASYMM8_SIGNEDQSYMM8_PER_CHANNELS32S32 +
QASYMM8_SIGNEDQSYMM8S32S32 +
+
CLGEMMLowpMatrixMultiplyCore + +
    +
  • NHWC +
  • NCHW +
+ 		+ +
src0src1src2dst +
QASYMM8QASYMM8S32QASYMM8 +
QASYMM8QSYMM8_PER_CHANNELS32QASYMM8 +
QASYMM8QSYMM8S32QASYMM8 +
QASYMM8QASYMM8S32S32 +
QASYMM8QSYMM8_PER_CHANNELS32S32 +
QASYMM8QSYMM8S32S32 +
QASYMM8_SIGNEDQASYMM8_SIGNEDS32QASYMM8_SIGNED +
QASYMM8_SIGNEDQSYMM8_PER_CHANNELS32QASYMM8_SIGNED +
QASYMM8_SIGNEDQSYMM8S32QASYMM8_SIGNED +
QASYMM8_SIGNEDQASYMM8_SIGNEDS32S32 +
QASYMM8_SIGNEDQSYMM8_PER_CHANNELS32S32 +
QASYMM8_SIGNEDQSYMM8S32S32 +
+
GEMMLowpOutputStage + General Matrix Multiplication. + +
    +
  • n/a +
+ 		NEGEMMLowpOutputStage + +
    +
  • All +
+ 		+ +
src0src1dst +
S32S32QASYMM8 +
S32S32QASYMM8_SIGNED +
S32S32QSYMM16 +
+
CLGEMMLowpOutputStage + +
    +
  • All +
+ 		+ +
src0src1dst +
S32S32QASYMM8 +
S32S32QASYMM8_SIGNED +
S32S32QSYMM16 +
+
GenerateProposalsLayer + Function to generate proposals for a RPN (Region Proposal Network). + +
    +
  • ANEURALNETWORKS_GENERATE_PROPOSALS +
+ 		NEGenerateProposalsLayer + +
    +
  • All +
+ 		+ +
src0src1src2dst +
F16F16F16F16 +
F32F32F32F32 +
QASYMM8QSYMM8QSYMM16QASYMM8 +
+
CLGenerateProposalsLayer + +
    +
  • All +
+ 		+ +
src0src1src2dst +
F16F16F16F16 +
F32F32F32F32 +
QASYMM8QSYMM8QSYMM16QASYMM8 +
+
InstanceNormalizationLayer + Function to perform a Instance normalization on a given axis. + +
    +
  • ANEURALNETWORKS_INSTANCE_NORMALIZATION +
+ 		NEInstanceNormalizationLayer + +
    +
  • NHWC +
  • NCHW +
+ 		+ +
srcdst +
F16F16 +
F32F32 +
+
CLInstanceNormalizationLayer + +
    +
  • NHWC +
  • NCHW +
+ 		+ +
srcdst +
F16F16 +
F32F32 +
+
L2NormalizeLayer + Function to perform a L2 normalization on a given axis. + +
    +
  • ANEURALNETWORKS_L2_NORMALIZATION +
+ 		NEL2NormalizeLayer + +
    +
  • NHWC +
  • NCHW +
+ 		+ +
srcdst +
F16F16 +
F32F32 +
+
CLL2NormalizeLayer + +
    +
  • NHWC +
  • NCHW +
+ 		+ +
srcdst +
F16F16 +
F32F32 +
+
Logical + Function to perform: - Logical AND - Logical OR - Logical NOT + +
    +
  • n/a +
+ 		NELogicalAnd + +
    +
  • All +
+ 		+ +
src0src1dst +
U8U8U8 +
+
NELogicalOr + +
    +
  • All +
+ 		+ +
src0src1dst +
U8U8U8 +
+
NELogicalNot + +
    +
  • All +
+ 		+ +
srcdst +
U8U8 +
+
LogicalAnd + Function to perform Logical AND. + +
    +
  • n/a +
+ 		CLLogicalAnd + +
    +
  • All +
+ 		+ +
src0src1dst +
U8U8U8 +
+
LogicalOr + Function to perform Logical OR. + +
    +
  • n/a +
+ 		CLLogicalOr + +
    +
  • All +
+ 		+ +
src0src1dst +
U8U8U8 +
+
LogicalNot + Function to perform Logical NOT. + +
    +
  • n/a +
+ 		CLLogicalNot + +
    +
  • All +
+ 		+ +
srcdst +
U8U8 +
+
LSTMLayer + Function to perform a single time step in a Long Short-Term Memory (LSTM) layer. + +
    +
  • ANEURALNETWORKS_LSTM +
+ 		NELSTMLayer + +
    +
  • All +
+ 		+ +
src0 - src13dst0 - dst3 +
F16F16 +
F32F32 +
+
CLLSTMLayer + +
    +
  • All +
+ 		+ +
src0 - src13dst0 - dst3 +
F16F16 +
F32F32 +
+
LSTMLayerQuantized + Function to perform quantized LSTM (Long Short-Term Memory) + +
    +
  • ANEURALNETWORKS_QUANTIZED_LSTM +
  • ANEURALNETWORKS_QUANTIZED_16BIT_LSTM +
+ 		NELSTMLayerQuantized + +
    +
  • All +
+ 		+ +
src0 - src8src9 - src12src13src14dst0dst1 +
QASYMM8S32QSYMM16QASYMM8QSYMM16QASYMM8 +
+
CLLSTMLayerQuantized + +
    +
  • All +
+ 		+ +
src0 - src8src9 - src12src13src14dst0dst1 +
QASYMM8S32QSYMM16QASYMM8QSYMM16QASYMM8 +
+
MaxUnpoolingLayer + Function to perform MaxUnpooling. + +
    +
  • n/a +
+ 		NEMaxUnpoolingLayer + +
    +
  • NHWC +
  • NCHW +
+ 		+ +
srcdst +
QASYMM8QASYMM8 +
QASYMM8_SIGNEDQASYMM8_SIGNED +
F16F16 +
F32F32 +
+
CLMaxUnpoolingLayer + +
    +
  • NHWC +
  • NCHW +
+ 		+ +
srcdst +
QASYMM8QASYMM8 +
QASYMM8_SIGNEDQASYMM8_SIGNED +
F16F16 +
F32F32 +
+
MeanStdDevNormalizationLayer + Function to execute mean and standard deviation normalization. + +
    +
  • n/a +
+ 		NEMeanStdDevNormalizationLayer + +
    +
  • NHWC +
  • NCHW +
+ 		+ +
srcdst +
F32F32 +
F16F16 +
+
CLMeanStdDevNormalizationLayer + +
    +
  • NHWC +
  • NCHW +
+ 		+ +
srcdst +
F32F32 +
F16F16 +
+
NormalizationLayer + Function to compute normalization layer. + +
    +
  • ANEURALNETWORKS_LOCAL_RESPONSE_NORMALIZATION +
+ 		NENormalizationLayer + +
    +
  • NHWC +
  • NCHW +
+ 		+ +
srcdst +
F32F32 +
F16F16 +
+
CLNormalizationLayer + +
    +
  • NHWC +
  • NCHW +
+ 		+ +
srcdst +
F32F32 +
F16F16 +
+
PadLayer + Function to pad a tensor. + +
    +
  • ANEURALNETWORKS_PAD +
  • ANEURALNETWORKS_PAD_V2 +
+ 		NEPadLayer + +
    +
  • NHWC +
  • NCHW +
+ 		+ +
srcdst +
AllAll +
+
CLPadLayer + +
    +
  • NHWC +
  • NCHW +
+ 		+ +
srcdst +
AllAll +
+
Permute + Function to transpose an ND tensor. + +
    +
  • ANEURALNETWORKS_TRANSPOSE +
+ 		NEPermute + +
    +
  • NHWC +
  • NCHW +
+ 		+ +
srcdst +
AllAll +
+
CLPermute + +
    +
  • NHWC +
  • NCHW +
+ 		+ +
srcdst +
AllAll +
+
PixelWiseMultiplication + Function to performe a multiplication. + +
    +
  • ANEURALNETWORKS_MUL +
+ 		NEPixelWiseMultiplication + +
    +
  • All +
+ 		+ +
src0src1dst +
QASYMM8QASYMM8QASYMM8 +
QASYMM8_SIGNEDQASYMM8_SIGNEDQASYMM8_SIGNED +
QSYMM16QSYMM16QASYMM16 +
QSYMM16QSYMM16S32 +
U8U8U8 +
U8U8S16 +
U8S16S16 +
S16U8S16 +
S16S16S16 +
F16F16F16 +
F32S32F32 +
+
CLPixelWiseMultiplication + +
    +
  • All +
+ 		+ +
src0src1dst +
QASYMM8QASYMM8QASYMM8 +
QASYMM8_SIGNEDQASYMM8_SIGNEDQASYMM8_SIGNED +
QSYMM16QSYMM16QASYMM16 +
QSYMM16QSYMM16S32 +
U8U8U8 +
U8U8S16 +
U8S16S16 +
S16U8S16 +
S16S16S16 +
F16F16F16 +
F32S32F32 +
+
PoolingLayer + Function to performe pooling with the specified pooling operation. + +
    +
  • ANEURALNETWORKS_AVERAGE_POOL_2D +
  • ANEURALNETWORKS_L2_POOL_2D +
  • ANEURALNETWORKS_MAX_POOL_2D +
+ 		NEPoolingLayer + +
    +
  • NHWC +
  • NCHW +
+ 		+ +
srcdst +
QASYMM8QASYMM8 +
QASYMM8_SIGNEDQASYMM8_SIGNED +
F16F16 +
F32F32 +
+
CLPoolingLayer + +
    +
  • NHWC +
  • NCHW +
+ 		+ +
srcdst +
QASYMM8QASYMM8 +
QASYMM8_SIGNEDQASYMM8_SIGNED +
F16F16 +
F32F32 +
+
PReluLayer + Function to compute the activation layer with the PRELU activation function. + +
    +
  • ANEURALNETWORKS_PRELU +
+ 		NEPReluLayer + +
    +
  • All +
+ 		+ +
srcdst +
QASYMM8QASYMM8 +
QASYMM8_SIGNEDQASYMM8_SIGNED +
F16F16 +
F32F32 +
+
CLPReluLayer + +
    +
  • All +
+ 		+ +
srcdst +
QASYMM8QASYMM8 +
QASYMM8_SIGNEDQASYMM8_SIGNED +
F16F16 +
F32F32 +
+
PriorBoxLayer + Function to compute prior boxes and clip. + +
    +
  • n/a +
+ 		NEPriorBoxLayer + +
    +
  • NHWC +
  • NCHW +
+ 		+ +
src0src1dst +
F32F32F32 +
+
CLPriorBoxLayer + +
    +
  • NHWC +
  • NCHW +
+ 		+ +
src0src1dst +
F32F32F32 +
+
QLSTMLayer + Function to perform quantized LSTM (Long Short-Term Memory). + +
    +
  • ANEURALNETWORKS_QUANTIZED_LSTM +
  • ANEURALNETWORKS_QUANTIZED_16BIT_LSTM +
+ 		NEQLSTMLayer + +
    +
  • All +
+ 		+ +
src0src1 - src6src7 -src9src10src11dst0dst1 - dst2 +
QASYMM8_SIGNEDQASYMM8S32QSYMM16QASYMM8_SIGNEDQSYMM16QASYMM8_SIGNED +
+
CLQLSTMLayer + +
    +
  • All +
+ 		+ +
src0src1 - src6src7 -src9src10src11dst0dst1 - dst2 +
QASYMM8_SIGNEDQASYMM8S32QSYMM16QASYMM8_SIGNEDQSYMM16QASYMM8_SIGNED +
+
QuantizationLayer + Function to perform quantization layer + +
    +
  • ANEURALNETWORKS_QUANTIZE +
+ 		NEQuantizationLayer + +
    +
  • All +
+ 		+ +
srcdst +
QASYMM8QASYMM8, QASYMM8_SIGNED, QASYMM16 +
QASYMM8_SIGNEDQASYMM8, QASYMM8_SIGNED, QASYMM16 +
F16QASYMM8, QASYMM8_SIGNED, QASYMM16 +
F32QASYMM8, QASYMM8_SIGNED, QASYMM16 +
+
CLQuantizationLayer + +
    +
  • All +
+ 		+ +
srcdst +
QASYMM8QASYMM8, QASYMM8_SIGNED, QASYMM16 +
QASYMM8_SIGNEDQASYMM8, QASYMM8_SIGNED, QASYMM16 +
F16QASYMM8, QASYMM8_SIGNED, QASYMM16 +
F32QASYMM8, QASYMM8_SIGNED, QASYMM16 +
+
Range + Function to generates a sequence of numbers starting from START and extends by increments of 'STEP' up to but not including 'END'. + +
    +
  • n/a +
+ 		NERange + +
    +
  • All +
+ 		+ +
dst +
U8 +
S8 +
U16 +
S16 +
U32 +
S32 +
F16 +
F32 +
+
CLRange + +
    +
  • All +
+ 		+ +
dst +
U8 +
S8 +
QASYMM8 +
U16 +
S16 +
U32 +
S32 +
F16 +
F32 +
+
ReduceMean + Function to performe reduce mean operation. + +
    +
  • ANEURALNETWORKS_MEAN +
+ 		NEReduceMean + +
    +
  • All +
+ 		+ +
srcdst +
QASYMM8QASYMM8 +
QASYMM8_SIGNEDQASYMM8_SIGNED +
F16F16 +
F32F32 +
+
CLReduceMean + +
    +
  • All +
+ 		+ +
srcdst +
QASYMM8QASYMM8 +
QASYMM8_SIGNEDQASYMM8_SIGNED +
F16F16 +
F32F32 +
+
ReductionOperation + Function to performe reduce with the following operations - ARG_IDX_MAX: Index of the max value - ARG_IDX_MIN: Index of the min value - MEAN_SUM: Mean of sum - PROD: Product - SUM_SQUARE: Sum of squares - SUM: Sum - MIN: Min - MAX: Max + +
    +
  • ANEURALNETWORKS_REDUCE_ALL +
  • ANEURALNETWORKS_REDUCE_ANY +
  • ANEURALNETWORKS_REDUCE_MAX +
  • ANEURALNETWORKS_REDUCE_MIN +
  • ANEURALNETWORKS_REDUCE_PROD +
  • ANEURALNETWORKS_REDUCE_SUM +
+ 		NEReductionOperation + +
    +
  • All +
+ 		+ +
srcdst +
QASYMM8QASYMM8 +
QASYMM8_SIGNEDQASYMM8_SIGNED +
F16F16 +
F32F32 +
S32S32 +
+
CLReductionOperation + +
    +
  • All +
+ 		+ +
srcdst +
QASYMM8QASYMM8 +
QASYMM8_SIGNEDQASYMM8_SIGNED +
F16F16 +
F32F32 +
S32S32 +
+
ReorgLayer + Performs a reorganization layer of input tensor to the output tensor. + +
    +
  • n/a +
+ 		NEReorgLayer + +
    +
  • NHWC +
  • NCHW +
+ 		+ +
srcdst +
AllAll +
+
CLReorgLayer + +
    +
  • NHWC +
  • NCHW +
+ 		+ +
srcdst +
AllAll +
+
ReshapeLayer + Function to reshape a tensor. + +
    +
  • ANEURALNETWORKS_RESHAPE +
  • ANEURALNETWORKS_SQUEEZE +
+ 		NEReshapeLayer + +
    +
  • All +
+ 		+ +
srcdst +
AllAll +
+
CLReshapeLayer + +
    +
  • All +
+ 		+ +
srcdst +
AllAll +
+
Reverse + Function to reverse tensor according to axis. + +
    +
  • n/a +
+ 		NEReverse + +
    +
  • All +
+ 		+ +
src0src1dst +
AllU32All +
+
CLReverse + +
    +
  • All +
+ 		+ +
src0src1dst +
AllU32All +
+
RNNLayer + Function to perform recurrent neural network layer. + +
    +
  • ANEURALNETWORKS_RNN +
+ 		NERNNLayer + +
    +
  • NHWC +
  • NCHW +
+ 		+ +
src0src1src2src3dst0dst1 +
F16F16F16F16F16F16 +
F32F32F32F32F32F32 +
+
CLRNNLayer + +
    +
  • NHWC +
  • NCHW +
+ 		+ +
src0src1src2src3dst0dst1 +
F16F16F16F16F16F16 +
F32F32F32F32F32F32 +
+
ROIAlignLayer + Function to perform ROI alignment. + +
    +
  • ANEURALNETWORKS_ROI_ALIGN +
+ 		NEROIAlignLayer + +
    +
  • All +
+ 		+ +
src0src1dst +
F16F16F16 +
F32F32F32 +
QASYMM8QASYMM16QASYMM8 +
QASYMM8_SIGNEDQASYMM16QASYMM8_SIGNED +
+
CLROIAlignLayer + +
    +
  • All +
+ 		+ +
src0src1dst +
F16F16F16 +
F32F32F32 +
QASYMM8QASYMM16QASYMM8 +
QASYMM8_SIGNEDQASYMM16QASYMM8_SIGNED +
+
ROIPoolingLayer + Function to perform ROI pooling. + +
    +
  • ANEURALNETWORKS_ROI_POOLING +
+ 		NEROIPoolingLayer + +
    +
  • All +
+ 		+ +
src0src1dst +
F32U16F32 +
QASYMM8U16QASYMM8 +
+
CLROIPoolingLayer + +
    +
  • All +
+ 		+ +
src0src1dst +
F16U16F16 +
F32U16F32 +
QASYMM8U16QASYMM8 +
+
Scale + Function to perform resize a tensor using to interpolate: - Bilinear - Nearest neighbor + +
    +
  • ANEURALNETWORKS_RESIZE_BILINEAR +
  • ANEURALNETWORKS_RESIZE_NEAREST_NEIGHBOR +
+ 		NEScale + +
    +
  • NHWC +
  • NCHW +
+ 		+ +
srcdst +
QASYMM8QASYMM8 +
QASYMM8_SIGNEDQASYMM8_SIGNED +
F16F16 +
F32F32 +
U8U8 +
S16S16 +
+
CLScale + +
    +
  • NHWC +
  • NCHW +
+ 		+ +
srcdst +
QASYMM8QASYMM8 +
QASYMM8_SIGNEDQASYMM8_SIGNED +
F16F16 +
F32F32 +
U8U8 +
S16S16 +
+
Select + Function to select values from 2 tensors depending on an input tensor of booleans. + +
    +
  • ANEURALNETWORKS_SELECT +
+ 		NESelect + +
    +
  • All +
+ 		+ +
src0src1src2dst +
U8AllAllAll +
+
CLSelect + +
    +
  • All +
+ 		+ +
src0src1src2dst +
U8AllAllAll +
+
Slice + Function to perform tensor slicing. + +
    +
  • ANEURALNETWORKS_SLICE +
+ 		NESlice + +
    +
  • All +
+ 		+ +
srcdst +
AllAll +
+
CLSlice + +
    +
  • All +
+ 		+ +
srcdst +
AllAll +
+
SoftmaxLayer + Function to compute a SoftmaxLayer and a Log SoftmaxLayer. + +
    +
  • ANEURALNETWORKS_LOG_SOFTMAX +
  • ANEURALNETWORKS_SOFTMAX +
+ 		NESoftmaxLayerGeneric + +
    +
  • All +
+ 		+ +
srcdst +
QASYMM8QASYMM8 +
QASYMM8_SIGNEDQASYMM8_SIGNED +
F16F16 +
F32F32 +
+
CLSoftmaxLayerGeneric + +
    +
  • All +
+ 		+ +
srcdst +
QASYMM8QASYMM8 +
QASYMM8_SIGNEDQASYMM8_SIGNED +
F16F16 +
F32F32 +
+
SpaceToBatchLayer + Function to divide a tensor spatially. + +
    +
  • ANEURALNETWORKS_SPACE_TO_BATCH_ND +
+ 		NESpaceToBatchLayer + +
    +
  • NHWC +
  • NCHW +
+ 		+ +
src0src1src2dst +
AllS32S32All +
+
CLSpaceToBatchLayer + +
    +
  • NHWC +
  • NCHW +
+ 		+ +
src0src1src2dst +
AllS32S32All +
+
SpaceToDepthLayer + Function to rearrange blocks of spatial data into depth. + +
    +
  • ANEURALNETWORKS_SPACE_TO_DEPTH +
+ 		NESpaceToDepthLayer + +
    +
  • NHWC +
  • NCHW +
+ 		+ +
srcdst +
AllAll +
+
CLSpaceToDepthLayer + +
    +
  • NHWC +
  • NCHW +
+ 		+ +
srcdst +
AllAll +
+
Split + Function to split a tensor along a given axis. + +
    +
  • ANEURALNETWORKS_SPLIT +
+ 		NESplit + +
    +
  • All +
+ 		+ +
srcdst +
AllAll +
+
CLSplit + +
    +
  • All +
+ 		+ +
srcdst +
AllAll +
+
StackLayer + Function to stack tensors along an axis. + +
    +
  • n/a +
+ 		NEStackLayer + +
    +
  • All +
+ 		+ +
srcdst +
AllAll +
+
CLStackLayer + +
    +
  • All +
+ 		+ +
srcdst +
AllAll +
+
StridedSlice + Function to extract a strided slice of a tensor. + +
    +
  • ANEURALNETWORKS_STRIDED_SLICE +
+ 		NEStridedSlice + +
    +
  • All +
+ 		+ +
srcdst +
AllAll +
+
CLStridedSlice + +
    +
  • All +
+ 		+ +
srcdst +
AllAll +
+
Tile + Function to construct a tensor by tiling a given tensor. + +
    +
  • ANEURALNETWORKS_TILE +
+ 		NETile + +
    +
  • All +
+ 		+ +
srcdst +
AllAll +
+
CLTile + +
    +
  • All +
+ 		+ +
srcdst +
AllAll +
+
Transpose + Function to transpose a 2D tensor. + +
    +
  • ANEURALNETWORKS_TRANSPOSE +
+ 		NETranspose + +
    +
  • All +
+ 		+ +
srcdst +
AllAll +
+
CLTranspose + +
    +
  • All +
+ 		+ +
srcdst +
AllAll +
+
Unstack + Function to unpack a rank-R tensor into rank-(R-1) tensors. + +
    +
  • n/a +
+ 		NEUnstack + +
    +
  • All +
+ 		+ +
srcdst +
AllAll +
+
CLUnstack + +
    +
  • All +
+ 		+ +
srcdst +
AllAll +
+
WinogradConvolutionLayer + Function to do Winograd Convolution. + +
    +
  • ANEURALNETWORKS_CONV_2D +
+ 		NEWinogradConvolutionLayer + +
    +
  • NHWC +
  • NCHW +
+ 		+ +
src0src1src2dst +
F16F16F16F16 +
F32F32F32F32 +
+
CLWinogradConvolutionLayer + +
    +
  • NHWC +
  • NCHW +
+ 		+ +
src0src1src2dst +
F16F16F16F16 +
F32F32F32F32 +
+
WinogradInputTransform + Function to. + +
    +
  • n/a +
+ 		CLWinogradInputTransform + +
    +
  • NHWC +
  • NCHW +
+ 		+ +
srcdst +
F16F16 +
F32F32 +
+
+ +*/ +} // namespace \ No newline at end of file -- cgit v1.2.1