Compute Library
18.05
|
This file contains all available output stages for GEMMLowp on OpenCL. More...
Namespaces | |
detail | |
gles | |
graph | |
graph_utils | |
io | |
logging | |
misc | |
quantization | |
strong_type | |
support | |
test | |
traits | |
tuners | |
utility | |
utils | |
wrapper | |
Data Structures | |
class | AccessWindowAutoPadding |
Dummy access window. More... | |
class | AccessWindowHorizontal |
Implementation of a row access pattern. More... | |
class | AccessWindowRectangle |
Implementation of a rectangular access pattern. More... | |
class | AccessWindowStatic |
Implementation of a static rectangular access pattern. More... | |
class | AccessWindowTranspose |
Implementation of a XY-transpose access pattern. More... | |
class | AccessWindowVertical |
Implementation of a column access pattern. More... | |
class | ActivationLayerInfo |
Activation Layer Information class. More... | |
class | Allocator |
Default malloc allocator implementation. More... | |
class | Array |
Basic implementation of the IArray interface which allocates a static number of T values. More... | |
class | AssemblyKernelGlue |
Assembly kernel glue. More... | |
class | BlobLifetimeManager |
Concrete class that tracks the lifetime of registered tensors and calculates the systems memory requirements in terms of blobs. More... | |
class | BlobMemoryPool |
Blob memory pool. More... | |
struct | BorderSize |
Container for 2D border size. More... | |
class | CLAbsoluteDifference |
Basic function to run CLAbsoluteDifferenceKernel. More... | |
class | CLAbsoluteDifferenceKernel |
Interface for the absolute difference kernel. More... | |
class | CLAccumulate |
Basic function to run CLAccumulateKernel. More... | |
class | CLAccumulateKernel |
Interface for the accumulate kernel. More... | |
class | CLAccumulateSquared |
Basic function to run CLAccumulateSquaredKernel. More... | |
class | CLAccumulateSquaredKernel |
Interface for the accumulate squared kernel. More... | |
class | CLAccumulateWeighted |
Basic function to run CLAccumulateWeightedKernel. More... | |
class | CLAccumulateWeightedKernel |
Interface for the accumulate weighted kernel. More... | |
class | CLActivationLayer |
Basic function to run CLActivationLayerKernel. More... | |
class | CLActivationLayerKernel |
Interface for the activation layer kernel. More... | |
class | CLArithmeticAddition |
Basic function to run CLArithmeticAdditionKernel. More... | |
class | CLArithmeticAdditionKernel |
Interface for the arithmetic addition kernel. More... | |
class | CLArithmeticSubtraction |
Basic function to run CLArithmeticSubtractionKernel. More... | |
class | CLArithmeticSubtractionKernel |
Interface for the arithmetic subtraction kernel. More... | |
class | CLArray |
CLArray implementation. More... | |
class | CLBatchNormalizationLayer |
Basic function to run CLNormalizationLayerKernel and simulate a batch normalization layer. More... | |
class | CLBatchNormalizationLayerKernel |
Interface for the BatchNormalization layer kernel. More... | |
class | CLBitwiseAnd |
Basic function to run CLBitwiseAndKernel. More... | |
class | CLBitwiseAndKernel |
Interface for the bitwise AND operation kernel. More... | |
class | CLBitwiseNot |
Basic function to run CLBitwiseNotKernel. More... | |
class | CLBitwiseNotKernel |
Interface for the bitwise NOT operation kernel. More... | |
class | CLBitwiseOr |
Basic function to run CLBitwiseOrKernel. More... | |
class | CLBitwiseOrKernel |
Interface for the bitwise OR operation kernel. More... | |
class | CLBitwiseXor |
Basic function to run CLBitwiseXorKernel. More... | |
class | CLBitwiseXorKernel |
Interface for the bitwise XOR operation kernel. More... | |
class | CLBox3x3 |
Basic function to execute box filter 3x3. More... | |
class | CLBox3x3Kernel |
Interface for the box 3x3 filter kernel. More... | |
class | CLBufferAllocator |
Default OpenCL cl buffer allocator implementation. More... | |
class | CLBufferMemoryRegion |
OpenCL buffer memory region implementation. More... | |
class | CLBuildOptions |
Build options. More... | |
class | CLCannyEdge |
Basic function to execute canny edge on OpenCL. More... | |
class | CLChannelCombine |
Basic function to run CLChannelCombineKernel to perform channel combination. More... | |
class | CLChannelCombineKernel |
Interface for the channel combine kernel. More... | |
class | CLChannelExtract |
Basic function to run CLChannelExtractKernel to perform channel extraction. More... | |
class | CLChannelExtractKernel |
Interface for the channel extract kernel. More... | |
class | CLChannelShuffleLayer |
Basic function to run CLChannelShuffleLayerKernel. More... | |
class | CLChannelShuffleLayerKernel |
Interface for the channel shuffle kernel. More... | |
class | CLCoarseSVMMemoryRegion |
OpenCL coarse-grain SVM memory region implementation. More... | |
struct | CLCoefficientTable |
Structure for storing Spatial Gradient Matrix and the minimum eigenvalue for each keypoint. More... | |
class | CLCol2ImKernel |
Interface for the col2im reshaping kernel. More... | |
class | CLColorConvert |
Basic function to run CLColorConvertKernel. More... | |
class | CLColorConvertKernel |
Interface for the color convert kernel. More... | |
class | CLConvertFullyConnectedWeights |
Basic function to run CLConvertFullyConnectedWeightsKernel. More... | |
class | CLConvertFullyConnectedWeightsKernel |
Interface to convert the 2D Fully Connected weights from NCHW to NHWC or vice versa. More... | |
class | CLConvolution3x3 |
Basic function to execute convolution of size 3x3. More... | |
class | CLConvolutionKernel |
Interface for the kernel to run an arbitrary size convolution on a tensor. More... | |
class | CLConvolutionLayer |
Basic function to compute the convolution layer. More... | |
class | CLConvolutionLayerReshapeWeights |
Function to reshape and transpose the weights. More... | |
class | CLConvolutionRectangle |
Basic function to execute non-square convolution. More... | |
class | CLConvolutionRectangleKernel |
Kernel for the running convolution on a rectangle matrix. More... | |
class | CLConvolutionSquare |
Basic function to execute square convolution.Currently it supports 5x5, 7x7, 9x9. More... | |
class | CLCopy |
class | CLCopyKernel |
OpenCL kernel to perform a copy between two tensors. More... | |
class | CLCopyToArrayKernel |
CL kernel to copy keypoints information to ICLKeyPointArray and counts the number of key points. More... | |
class | CLDeconvolutionLayer |
Function to run the deconvolution layer. More... | |
class | CLDeconvolutionLayerUpsample |
Basic function to run CLDeconvolutionLayerUpsampleKernel. More... | |
class | CLDeconvolutionLayerUpsampleKernel |
Interface for the Deconvolution layer kernel on OpenCL. More... | |
class | CLDepthConcatenateLayer |
Basic function to execute concatenate tensors along z axis. More... | |
class | CLDepthConcatenateLayerKernel |
Interface for the depth concatenate kernel. More... | |
class | CLDepthConvertLayer |
Basic function to run CLDepthConvertLayerKernel. More... | |
class | CLDepthConvertLayerKernel |
Interface for the depth conversion kernel. More... | |
class | CLDepthwiseConvolutionLayer |
Basic function to execute a generic depthwise convolution. More... | |
class | CLDepthwiseConvolutionLayer3x3 |
Basic function to execute a depthwise convolution for kernel size 3x3xC (when data layout NCHW) or Cx3x3 (when data layout NHWC). More... | |
class | CLDepthwiseConvolutionLayer3x3NCHWKernel |
Interface for the kernel to run a 3x3 depthwise convolution on a tensor when the data layout is NCHW. More... | |
class | CLDepthwiseConvolutionLayer3x3NHWCKernel |
Interface for the kernel to run a 3x3 depthwise convolution on a tensor when the data layout is NHWC. More... | |
class | CLDepthwiseIm2ColKernel |
Interface for the depthwise im2col reshape kernel. More... | |
class | CLDepthwiseSeparableConvolutionLayer |
Basic function to execute depthwise convolution. More... | |
class | CLDepthwiseVectorToTensorKernel |
Interface for the depthwise vector to tensor kernel. More... | |
class | CLDepthwiseWeightsReshapeKernel |
Interface for the depthwise weights reshape kernel. More... | |
class | CLDequantizationLayer |
Basic function to simulate a dequantization layer. More... | |
class | CLDequantizationLayerKernel |
Interface for the dequantization layer kernel. More... | |
class | CLDerivative |
Basic function to execute first order derivative operator. More... | |
class | CLDerivativeKernel |
Interface for the derivative kernel. More... | |
struct | CLDeviceOptions |
OpenCL device options. More... | |
class | CLDilate |
Basic function to execute dilate. More... | |
class | CLDilateKernel |
Interface for the dilate kernel. More... | |
class | CLDirectConvolutionLayer |
Basic function to execute direct convolution function: More... | |
class | CLDirectConvolutionLayerKernel |
Interface for the direct convolution kernel. More... | |
class | CLDirectConvolutionLayerOutputStageKernel |
OpenCL kernel to accumulate the biases, if provided, or downscale in case of quantized input. More... | |
class | CLDistribution1D |
CLDistribution1D object class. More... | |
class | CLEdgeNonMaxSuppressionKernel |
OpenCL kernel to perform Non-Maxima suppression for Canny Edge. More... | |
class | CLEdgeTraceKernel |
OpenCL kernel to perform Edge tracing. More... | |
class | CLEqualizeHistogram |
Basic function to execute histogram equalization. More... | |
class | CLErode |
Basic function to execute erode. More... | |
class | CLErodeKernel |
Interface for the erode kernel. More... | |
class | CLFastCorners |
Basic function to execute fast corners. More... | |
class | CLFastCornersKernel |
CL kernel to perform fast corners. More... | |
class | CLFillBorder |
Basic function to run CLFillBorderKernel. More... | |
class | CLFillBorderKernel |
Interface for filling the border of a kernel. More... | |
class | CLFineSVMMemoryRegion |
OpenCL fine-grain SVM memory region implementation. More... | |
class | CLFlattenLayer |
Basic function to execute flatten. More... | |
class | CLFloor |
Basic function to run CLFloorKernel. More... | |
class | CLFloorKernel |
OpenCL kernel to perform a floor operation. More... | |
class | CLFullyConnectedLayer |
Basic function to compute a Fully Connected layer on OpenCL. More... | |
class | CLFullyConnectedLayerReshapeWeights |
Basic function to reshape the weights of Fully Connected layer with OpenCL. More... | |
class | CLGaussian3x3 |
Basic function to execute gaussian filter 3x3. More... | |
class | CLGaussian3x3Kernel |
Interface for the Gaussian 3x3 filter kernel. More... | |
class | CLGaussian5x5 |
Basic function to execute gaussian filter 5x5. More... | |
class | CLGaussian5x5HorKernel |
Interface for the kernel to run the horizontal pass of 5x5 Gaussian filter on a tensor. More... | |
class | CLGaussian5x5VertKernel |
Interface for the kernel to run the vertical pass of 5x5 Gaussian filter on a tensor. More... | |
class | CLGaussianPyramid |
Common interface for all Gaussian pyramid functions. More... | |
class | CLGaussianPyramidHalf |
Basic function to execute gaussian pyramid with HALF scale factor. More... | |
class | CLGaussianPyramidHorKernel |
OpenCL kernel to perform a Gaussian filter and half scaling across width (horizontal pass) More... | |
class | CLGaussianPyramidOrb |
Basic function to execute gaussian pyramid with ORB scale factor. More... | |
class | CLGaussianPyramidVertKernel |
OpenCL kernel to perform a Gaussian filter and half scaling across height (vertical pass) More... | |
class | CLGEMM |
Basic function to execute GEMM on OpenCL. More... | |
class | CLGEMMConvolutionLayer |
Basic function to compute the convolution layer. More... | |
class | CLGEMMInterleave4x4 |
Basic function to execute CLGEMMInterleave4x4Kernel. More... | |
class | CLGEMMInterleave4x4Kernel |
OpenCL kernel which interleaves the elements of a matrix A in chunk of 4x4. More... | |
class | CLGEMMLowpMatrixAReductionKernel |
OpenCL kernel used to compute the row-vectors of sums of all the entries in each row of Matrix A. More... | |
class | CLGEMMLowpMatrixBReductionKernel |
OpenCL kernel used to compute the row-vectors of sums of all the entries in each column of Matrix B. More... | |
class | CLGEMMLowpMatrixMultiplyCore |
Basic function to execute GEMMLowpMatrixMultiplyCore on OpenCL. More... | |
class | CLGEMMLowpMatrixMultiplyKernel |
OpenCL kernel to multiply matrices. More... | |
class | CLGEMMLowpOffsetContributionKernel |
OpenCL kernel used to add the offset contribution after CLGEMMLowpMatrixMultiplyKernel. More... | |
class | CLGEMMLowpQuantizeDownInt32ToUint8Scale |
Basic function to execute CLGEMMLowpQuantizeDownInt32ToUint8Scale on OpenCL. More... | |
class | CLGEMMLowpQuantizeDownInt32ToUint8ScaleByFixedPoint |
Basic function to execute CLGEMMLowpQuantizeDownInt32ToUint8ScaleByFixedPoint on OpenCL. More... | |
class | CLGEMMLowpQuantizeDownInt32ToUint8ScaleByFixedPointKernel |
OpenCL kernel used to quantize down the int32 accumulator values of GEMMLowp to QASYMM8. More... | |
class | CLGEMMLowpQuantizeDownInt32ToUint8ScaleKernel |
OpenCL kernel used to quantize down the int32 accumulator values of GEMMLowp to QASYMM8. More... | |
class | CLGEMMMatrixAccumulateBiasesKernel |
Interface to add a bias to each row of the input tensor. More... | |
class | CLGEMMMatrixAdditionKernel |
OpenCL kernel to perform the in-place matrix addition between 2 matrices, taking into account that the second matrix might be weighted by a scalar value beta. More... | |
class | CLGEMMMatrixMultiplyKernel |
OpenCL kernel to multiply two input matrices "A" and "B" . More... | |
class | CLGEMMMatrixVectorMultiplyKernel |
Interface for the GEMM matrix vector multiply kernel. More... | |
class | CLGEMMTranspose1xW |
Basic function to execute CLGEMMTranspose1xWKernel. More... | |
class | CLGEMMTranspose1xWKernel |
OpenCL kernel which transposes the elements of a matrix in chunks of 1xW, where W is equal to (16 / element size of the tensor) More... | |
class | CLGradientKernel |
OpenCL kernel to perform Gradient computation. More... | |
class | CLHarrisCorners |
Basic function to execute harris corners detection. More... | |
class | CLHarrisScoreKernel |
Interface for the harris score kernel. More... | |
class | CLHistogram |
Basic function to execute histogram. More... | |
class | CLHistogramBorderKernel |
Interface to run the histogram kernel to handle the leftover part of image. More... | |
class | CLHistogramKernel |
Interface to run the histogram kernel. More... | |
class | CLHOG |
OpenCL implementation of HOG data-object. More... | |
class | CLHOGBlockNormalizationKernel |
OpenCL kernel to perform HOG block normalization. More... | |
class | CLHOGDescriptor |
Basic function to calculate HOG descriptor. More... | |
class | CLHOGDetector |
Basic function to execute HOG detector based on linear SVM. More... | |
class | CLHOGDetectorKernel |
OpenCL kernel to perform HOG detector kernel using linear SVM. More... | |
class | CLHOGGradient |
Basic function to calculate the gradient for HOG. More... | |
class | CLHOGMultiDetection |
Basic function to detect multiple objects (or the same object at different scales) on the same input image using HOG. More... | |
class | CLHOGOrientationBinningKernel |
OpenCL kernel to perform HOG Orientation Binning. More... | |
class | CLIm2ColKernel |
Interface for the im2col reshape kernel. More... | |
class | CLIntegralImage |
Basic function to execute integral image. More... | |
class | CLIntegralImageHorKernel |
Interface to run the horizontal pass of the integral image kernel. More... | |
class | CLIntegralImageVertKernel |
Interface to run the vertical pass of the integral image kernel. More... | |
class | CLKernelLibrary |
CLKernelLibrary class. More... | |
class | CLL2NormalizeLayer |
Basic function to perform a L2 normalization on a given axis. More... | |
class | CLL2NormalizeLayerKernel |
Interface for performing a L2 normalize on a given axis given the square sum of it in this axis. More... | |
class | CLLaplacianPyramid |
Basic function to execute laplacian pyramid. More... | |
class | CLLaplacianReconstruct |
Basic function to execute laplacian reconstruction. More... | |
struct | CLLKInternalKeypoint |
Internal keypoint structure for Lucas-Kanade Optical Flow. More... | |
class | CLLKTrackerFinalizeKernel |
Interface to run the finalize step of LKTracker, where it truncates the coordinates stored in new_points array. More... | |
class | CLLKTrackerInitKernel |
Interface to run the initialization step of LKTracker. More... | |
class | CLLKTrackerStage0Kernel |
Interface to run the first stage of LKTracker, where A11, A12, A22, min_eig, ival, ixval and iyval are computed. More... | |
class | CLLKTrackerStage1Kernel |
Interface to run the second stage of LKTracker, where the motion vectors of the given points are computed. More... | |
class | CLLocallyConnectedLayer |
Basic function to compute the locally connected layer. More... | |
class | CLLocallyConnectedMatrixMultiplyKernel |
OpenCL kernel to multiply each row of first tensor with low 2 dimensions of second tensor. More... | |
class | CLLogits1DMaxKernel |
Interface for the identifying the max value of 1D Logits. More... | |
class | CLLogits1DMaxShiftExpSumKernel |
Interface for max, shifting, exponentiating and summing the logits. More... | |
class | CLLogits1DNormKernel |
Interface for calculating the final step of the Softmax Layer where each logit value is multiplied by the inverse of the sum of the logits. More... | |
class | CLLogits1DShiftExpSumKernel |
Interface for shifting, exponentiating and summing the logits. More... | |
class | CLLSTMLayer |
This function performs a single time step in a Long Short-Term Memory (LSTM) layer. More... | |
class | CLLut |
Basic implementation of the OpenCL lut interface. More... | |
class | CLLutAllocator |
Basic implementation of a CL memory LUT allocator. More... | |
class | CLMagnitude |
Basic function to run CLMagnitudePhaseKernel. More... | |
class | CLMagnitudePhaseKernel |
Template interface for the kernel to compute magnitude and phase. More... | |
class | CLMeanStdDev |
Basic function to execute mean and standard deviation by calling CLMeanStdDevKernel. More... | |
class | CLMeanStdDevKernel |
Interface for the kernel to calculate mean and standard deviation of input image pixels. More... | |
class | CLMedian3x3 |
Basic function to execute median filter. More... | |
class | CLMedian3x3Kernel |
Interface for the median 3x3 filter kernel. More... | |
class | CLMemory |
OpenCL implementation of memory object. More... | |
class | CLMinMaxKernel |
Interface for the kernel to perform min max search on an image. More... | |
class | CLMinMaxLayerKernel |
Interface for the kernel to perform min max search on a 3D tensor. More... | |
class | CLMinMaxLocation |
Basic function to execute min and max location. More... | |
class | CLMinMaxLocationKernel |
Interface for the kernel to find min max locations of an image. More... | |
class | CLMultiHOG |
Basic implementation of the CL multi HOG data-objects. More... | |
class | CLMultiImage |
Basic implementation of the CL multi-planar image interface. More... | |
class | CLNonLinearFilter |
Basic function to execute non linear filter. More... | |
class | CLNonLinearFilterKernel |
Interface for the kernel to apply a non-linear filter. More... | |
class | CLNonMaximaSuppression3x3 |
Basic function to execute non-maxima suppression over a 3x3 window. More... | |
class | CLNonMaximaSuppression3x3Kernel |
Interface to perform Non-Maxima suppression over a 3x3 window using OpenCL. More... | |
class | CLNormalizationLayer |
Basic function to compute a normalization layer. More... | |
class | CLNormalizationLayerKernel |
Interface for the normalization layer kernel. More... | |
struct | CLOldValue |
Structure for storing ival, ixval and iyval for each point inside the window. More... | |
class | CLOpticalFlow |
Basic function to execute optical flow. More... | |
class | CLPermute |
Basic function to execute an CLPermuteKernel. More... | |
class | CLPermuteKernel |
OpenCL kernel to perform tensor permutation. More... | |
class | CLPhase |
Basic function to execute an CLMagnitudePhaseKernel. More... | |
class | CLPixelWiseMultiplication |
Basic function to run CLPixelWiseMultiplicationKernel. More... | |
class | CLPixelWiseMultiplicationKernel |
Interface for the pixelwise multiplication kernel. More... | |
class | CLPoolingLayer |
Basic function to simulate a pooling layer with the specified pooling operation. More... | |
class | CLPoolingLayerKernel |
Interface for the pooling layer kernel. More... | |
class | CLPyramid |
Basic implementation of the OpenCL pyramid interface. More... | |
class | CLQuantizationLayer |
Basic function to simulate a quantization layer. More... | |
class | CLQuantizationLayerKernel |
Interface for the quantization layer kernel. More... | |
class | CLReductionOperation |
Perform reduction operation. More... | |
class | CLReductionOperationKernel |
Interface for the reduction operation kernel. More... | |
class | CLRemap |
Basic function to execute remap. More... | |
class | CLRemapKernel |
OpenCL kernel to perform a remap on a tensor. More... | |
class | CLReshapeLayer |
Basic function to run CLReshapeLayerKernel. More... | |
class | CLReshapeLayerKernel |
Interface for the kernel to perform tensor reshaping. More... | |
class | CLRNNLayer |
Basic function to run CLRNNLayer. More... | |
class | CLROIPoolingLayer |
Basic function to run CLROIPoolingLayerKernel. More... | |
class | CLROIPoolingLayerKernel |
Interface for the ROI pooling layer kernel. More... | |
class | CLScale |
Basic function to run CLScaleKernel. More... | |
class | CLScaleKernel |
Interface for the scale kernel. More... | |
class | CLScharr3x3 |
Basic function to execute scharr 3x3 filter. More... | |
class | CLScharr3x3Kernel |
Interface for the kernel to run a 3x3 Scharr filter on a tensor. More... | |
class | CLScheduler |
Provides global access to a CL context and command queue. More... | |
class | CLSeparableConvolutionHorKernel |
Kernel for the Horizontal pass of a Separable Convolution. More... | |
class | CLSeparableConvolutionVertKernel |
Kernel for the Vertical pass of a Separable Convolution. More... | |
class | CLSobel3x3 |
Basic function to execute sobel 3x3 filter. More... | |
class | CLSobel3x3Kernel |
Interface for the kernel to run a 3x3 Sobel filter on a tensor. More... | |
class | CLSobel5x5 |
Basic function to execute sobel 5x5 filter. More... | |
class | CLSobel5x5HorKernel |
Interface for the kernel to run the horizontal pass of 5x5 Sobel filter on a tensor. More... | |
class | CLSobel5x5VertKernel |
Interface for the kernel to run the vertical pass of 5x5 Sobel filter on a tensor. More... | |
class | CLSobel7x7 |
Basic function to execute sobel 7x7 filter. More... | |
class | CLSobel7x7HorKernel |
Interface for the kernel to run the horizontal pass of 7x7 Sobel filter on a tensor. More... | |
class | CLSobel7x7VertKernel |
Interface for the kernel to run the vertical pass of 7x7 Sobel filter on a tensor. More... | |
class | CLSoftmaxLayer |
Basic function to compute a SoftmaxLayer. More... | |
class | CLSubTensor |
Basic implementation of the OpenCL sub-tensor interface. More... | |
class | CLSymbols |
Class for loading OpenCL symbols. More... | |
class | CLTableLookup |
Basic function to run CLTableLookupKernel. More... | |
class | CLTableLookupKernel |
Interface for the kernel to perform table lookup calculations. More... | |
class | CLTensor |
Basic implementation of the OpenCL tensor interface. More... | |
class | CLTensorAllocator |
Basic implementation of a CL memory tensor allocator. More... | |
class | CLThreshold |
Basic function to run CLThresholdKernel. More... | |
class | CLThresholdKernel |
Interface for the thresholding kernel. More... | |
class | CLTranspose |
Basic function to transpose a matrix on OpenCL. More... | |
class | CLTransposeKernel |
OpenCL kernel which transposes the elements of a matrix. More... | |
class | CLTuner |
Basic implementation of the OpenCL tuner interface. More... | |
class | CLWarpAffine |
Basic function to run CLWarpAffineKernel for AFFINE transformation. More... | |
class | CLWarpAffineKernel |
Interface for the warp affine kernel. More... | |
class | CLWarpPerspective |
Basic function to run CLWarpPerspectiveKernel for PERSPECTIVE transformation. More... | |
class | CLWarpPerspectiveKernel |
Interface for the warp perspective kernel. More... | |
class | CLWeightsReshapeKernel |
OpenCL kernel to perform reshaping on the weights used by convolution and locally connected layer. More... | |
class | CLWidthConcatenateLayer |
Basic function to execute concatenate tensors along x axis. More... | |
class | CLWidthConcatenateLayerKernel |
Interface for the width concatenate kernel. More... | |
class | CLWinogradConvolutionLayer |
Basic function to execute Winograd-based convolution on OpenCL. More... | |
class | CLWinogradFilterTransformKernel |
Interface for the Winograd filter transform kernel. More... | |
class | CLWinogradInputTransform |
Basic function to execute a CLWinogradInputTransformKernel. More... | |
class | CLWinogradInputTransformKernel |
OpenCL kernel to perform Winograd input transform. More... | |
class | CLWinogradOutputTransformKernel |
Interface for the Winograd output transform kernel. More... | |
class | Coordinates |
Coordinates of an item. More... | |
struct | Coordinates2D |
Coordinate type. More... | |
struct | Coordinates3D |
Coordinate type. More... | |
class | CPPCornerCandidatesKernel |
CPP kernel to perform corner candidates. More... | |
class | CPPDetectionWindowNonMaximaSuppressionKernel |
CPP kernel to perform in-place computation of euclidean distance on IDetectionWindowArray. More... | |
class | CPPPermute |
Basic function to run CPPPermuteKernel. More... | |
class | CPPPermuteKernel |
CPP kernel to perform tensor permutation. More... | |
class | CPPScheduler |
C++11 implementation of a pool of threads to automatically split a kernel's execution among several threads. More... | |
class | CPPSortEuclideanDistanceKernel |
CPP kernel to perform sorting and euclidean distance. More... | |
class | CPPUpsample |
Basic function to run CPPUpsample. More... | |
class | CPPUpsampleKernel |
CPP kernel to perform tensor upsample. More... | |
class | CPUInfo |
struct | DetectionWindow |
Detection window used for the object detection. More... | |
class | Dimensions |
Dimensions with dimensionality. More... | |
class | Distribution1D |
Basic implementation of the 1D distribution interface. More... | |
struct | enable_bitwise_ops |
Disable bitwise operations by default. More... | |
struct | enable_bitwise_ops< arm_compute::GPUTarget > |
Enable bitwise operations on GPUTarget enumerations. More... | |
class | GCAbsoluteDifference |
Basic function to run GCAbsoluteDifferenceKernel. More... | |
class | GCAbsoluteDifferenceKernel |
Interface for the absolute difference kernel. More... | |
class | GCActivationLayer |
Basic function to run GCActivationLayerKernel. More... | |
class | GCActivationLayerKernel |
Interface for the activation layer kernel. More... | |
class | GCArithmeticAddition |
Basic function to run GCArithmeticAdditionKernel. More... | |
class | GCArithmeticAdditionKernel |
Interface for the arithmetic addition kernel. More... | |
class | GCBatchNormalizationLayer |
Basic function to run GCBatchNormalizationLayerKernel and simulate a batch normalization layer. More... | |
class | GCBatchNormalizationLayerKernel |
Interface for the BatchNormalization layer kernel. More... | |
class | GCBufferAllocator |
Default GLES buffer allocator implementation. More... | |
class | GCCol2ImKernel |
Interface for the col2im reshaping kernel. More... | |
class | GCConvolutionLayer |
Basic function to compute the convolution layer. More... | |
class | GCConvolutionLayerReshapeWeights |
Function to reshape and transpose the weights. More... | |
class | GCDepthConcatenateLayer |
Basic function to execute concatenate tensors along z axis. More... | |
class | GCDepthConcatenateLayerKernel |
Interface for the depth concatenate kernel. More... | |
class | GCDepthwiseConvolutionLayer3x3 |
Basic function to execute a depthwise convolution for kernel size 3x3xC. More... | |
class | GCDepthwiseConvolutionLayer3x3Kernel |
Interface for the kernel to run a 3x3 depthwise convolution on a tensor. More... | |
class | GCDirectConvolutionLayer |
Basic function to execute direct convolution function. More... | |
class | GCDirectConvolutionLayerKernel |
Interface for the direct convolution kernel. More... | |
class | GCDropoutLayer |
Basic function to do dropout op. More... | |
class | GCDropoutLayerKernel |
Interface for the dropout layer kernel. More... | |
class | GCFillBorder |
Basic function to run GCFillBorderKernel. More... | |
class | GCFillBorderKernel |
Interface for filling the border of a kernel. More... | |
class | GCFullyConnectedLayer |
Basic function to compute a Fully Connected layer on OpenGL ES. More... | |
class | GCFullyConnectedLayerReshapeWeights |
Basic function to reshape the weights of Fully Connected layer with OpenGL ES. More... | |
class | GCGEMM |
Basic function to execute GEMM on OpenGLES Compute. More... | |
class | GCGEMMInterleave4x4 |
Basic function to execute GCGEMMInterleave4x4Kernel. More... | |
class | GCGEMMInterleave4x4Kernel |
OpenGL ES kernel which interleaves the elements of a matrix A in chunk of 4x4. More... | |
class | GCGEMMMatrixAccumulateBiasesKernel |
Interface to add a bias to each row of the input tensor. More... | |
class | GCGEMMMatrixAdditionKernel |
OpenGL ES kernel to perform the in-place matrix addition between 2 matrices, taking into account that the second matrix might be weighted by a scalar value beta. More... | |
class | GCGEMMMatrixMultiplyKernel |
GLES Compute kernel to multiply two input matrices "A" and "B" or to multiply a vector "A" by a matrix "B". More... | |
class | GCGEMMTranspose1xW |
Basic function to execute GCGEMMTranspose1xWKernel. More... | |
class | GCGEMMTranspose1xWKernel |
OpenGLES kernel which transposes the elements of a matrix in chunks of 1xW, where W is equal to (16 / element size of the tensor) More... | |
class | GCIm2ColKernel |
Interface for the im2col reshape kernel. More... | |
class | GCKernel |
GCKernel class. More... | |
class | GCKernelLibrary |
GCKernelLibrary class. More... | |
class | GCLogits1DMaxKernel |
Interface for the identifying the max value of 1D Logits. More... | |
class | GCLogits1DNormKernel |
Interface for calculating the final step of the Softmax Layer where each logit value is multiplied by the inverse of the sum of the logits. More... | |
class | GCLogits1DShiftExpSumKernel |
Interface for shifting the logits values around the max value and exponentiating the result. More... | |
class | GCNormalizationLayer |
Basic function to compute a normalization layer. More... | |
class | GCNormalizationLayerKernel |
Interface for the normalization layer kernel. More... | |
class | GCNormalizePlanarYUVLayer |
Basic function to run GCNormalizePlanarYUVLayerKernel. More... | |
class | GCNormalizePlanarYUVLayerKernel |
Interface for the NormalizePlanarYUV layer kernel. More... | |
class | GCPixelWiseMultiplication |
Basic function to run GCPixelWiseMultiplicationKernel. More... | |
class | GCPixelWiseMultiplicationKernel |
Interface for the pixelwise multiplication kernel. More... | |
class | GCPoolingLayer |
Basic function to simulate a pooling layer with the specified pooling operation. More... | |
class | GCPoolingLayerKernel |
Interface for the pooling layer kernel. More... | |
class | GCProgram |
GCProgram class. More... | |
class | GCScale |
Basic function to run GCScaleKernel. More... | |
class | GCScaleKernel |
Interface for the scale kernel. More... | |
class | GCScheduler |
Provides global access to a OpenGL ES context and command queue. More... | |
class | GCSoftmaxLayer |
Basic function to compute a SoftmaxLayer. More... | |
class | GCTensor |
Interface for OpenGL ES tensor. More... | |
class | GCTensorAllocator |
Basic implementation of a GLES memory tensor allocator. More... | |
class | GCTensorShift |
Basic function to execute shift function for tensor. More... | |
class | GCTensorShiftKernel |
Interface for the kernel to shift valid data on a tensor. More... | |
class | GCTranspose |
Basic function to transpose a matrix on OpenGL ES. More... | |
class | GCTransposeKernel |
OpenGL ES kernel which transposes the elements of a matrix. More... | |
class | GCWeightsReshapeKernel |
GLES Compute kernel to perform reshaping on the weights used by convolution and locally connected layer. More... | |
class | GEMMInfo |
GEMM information class. More... | |
class | GEMMReshapeInfo |
GEMM reshape information class. More... | |
class | GLBufferWrapper |
class | HOG |
CPU implementation of HOG data-object. More... | |
class | HOGInfo |
Store the HOG's metadata. More... | |
class | IAccessWindow |
Interface describing methods to update access window and padding based on kernel parameters. More... | |
class | IAllocator |
Allocator interface. More... | |
class | IArray |
Array of type T. More... | |
class | ICLArray |
Interface for OpenCL Array. More... | |
class | ICLDepthwiseConvolutionLayer3x3Kernel |
Interface for the kernel to run a 3x3 depthwise convolution on a tensor. More... | |
class | ICLDistribution1D |
ICLDistribution1D interface class. More... | |
class | ICLGEMMLowpReductionKernel |
Common interface for all OpenCL reduction kernels. More... | |
class | ICLHOG |
Interface for OpenCL HOG data-object. More... | |
class | ICLKernel |
Common interface for all the OpenCL kernels. More... | |
class | ICLLut |
Interface for OpenCL LUT. More... | |
class | ICLMemoryRegion |
OpenCL memory region interface. More... | |
class | ICLMultiHOG |
Interface for storing multiple HOG data-objects. More... | |
class | ICLMultiImage |
Interface for OpenCL multi-planar images. More... | |
class | ICLSimple2DKernel |
Interface for simple OpenCL kernels having 1 tensor input and 1 tensor output. More... | |
class | ICLSimple3DKernel |
Interface for simple OpenCL kernels having 1 tensor input and 1 tensor output. More... | |
class | ICLSimpleFunction |
Basic interface for functions which have a single OpenCL kernel. More... | |
class | ICLSimpleKernel |
Interface for simple OpenCL kernels having 1 tensor input and 1 tensor output. More... | |
class | ICLSVMMemoryRegion |
OpenCL SVM memory region interface. More... | |
class | ICLTensor |
Interface for OpenCL tensor. More... | |
class | ICLTuner |
Basic interface for tuning the OpenCL kernels. More... | |
class | ICPPKernel |
Common interface for all kernels implemented in C++. More... | |
class | ICPPSimpleFunction |
Basic interface for functions which have a single CPP kernel. More... | |
class | ICPPSimpleKernel |
Interface for simple C++ kernels having 1 tensor input and 1 tensor output. More... | |
class | IDistribution |
Interface for distribution objects. More... | |
class | IDistribution1D |
1D Distribution interface More... | |
class | IFunction |
Base class for all functions. More... | |
class | IGCKernel |
Common interface for all the GLES kernels. More... | |
class | IGCSimple2DKernel |
Interface for simple OpenGL ES kernels having 1 tensor input and 1 tensor output. More... | |
class | IGCSimple3DKernel |
Interface for simple GLES kernels having 1 tensor input and 1 tensor output. More... | |
class | IGCSimpleFunction |
Basic interface for functions which have a single OpenGL ES kernel. More... | |
class | IGCSimpleKernel |
Interface for simple OpenGL ES kernels having 1 tensor input and 1 tensor output. More... | |
class | IGCTensor |
Interface for GLES Compute tensor. More... | |
class | IHOG |
Interface for HOG data-object. More... | |
class | IKernel |
Common information for all the kernels. More... | |
class | ILifetimeManager |
Interface for managing the lifetime of objects. More... | |
class | ILut |
Lookup Table object interface. More... | |
class | ILutAllocator |
Basic interface to allocate LUTs'. More... | |
class | IMemoryGroup |
Memory group interface. More... | |
class | IMemoryManager |
Memory manager interface to handle allocations of backing memory. More... | |
class | IMemoryPool |
Memory Pool Inteface. More... | |
class | IMemoryRegion |
Memory region interface. More... | |
class | IMultiHOG |
Interface for storing multiple HOG data-objects. More... | |
class | IMultiImage |
Interface for multi-planar images. More... | |
class | INEGEMMLowpReductionKernel |
Common interface for all NEON reduction kernels. More... | |
class | INEHarrisScoreKernel |
Common interface for all Harris Score kernels. More... | |
class | INESimpleFunction |
Basic interface for functions which have a single NEON kernel. More... | |
class | INEWarpKernel |
Common interface for warp affine and warp perspective. More... | |
class | INEWinogradLayerBatchedGEMMKernel |
Interface for the NEON kernel to perform Winograd. More... | |
class | INEWinogradLayerTransformInputKernel |
Interface for the NEON kernel to perform Winograd input transform. More... | |
class | INEWinogradLayerTransformOutputKernel |
Interface for the NEON kernel to perform Winograd output transform. More... | |
class | INEWinogradLayerTransformWeightsKernel |
Interface for the NEON kernel to perform Winograd weights transform. More... | |
struct | InternalKeyPoint |
Internal keypoint class for Lucas-Kanade Optical Flow. More... | |
struct | IOFormatInfo |
IO formatting information class. More... | |
class | IPoolManager |
Memory pool manager interface. More... | |
class | IPyramid |
Interface for pyramid data-object. More... | |
class | IScheduler |
Scheduler interface to run kernels. More... | |
class | ISimpleLifetimeManager |
Abstract class of the simple lifetime manager interface. More... | |
class | ITensor |
Interface for NEON tensor. More... | |
class | ITensorAllocator |
Interface to allocate tensors. More... | |
class | ITensorInfo |
Store the tensor's metadata. More... | |
class | Iterator |
Iterator updated by execute_window_loop for each window element. More... | |
class | Kernel |
Kernel class. More... | |
struct | KeyPoint |
Keypoint type. More... | |
class | LSTMParams |
class | Lut |
Basic implementation of the LUT interface. More... | |
class | LutAllocator |
Basic implementation of a CPU memory LUT allocator. More... | |
class | Memory |
CPU implementation of memory object. More... | |
class | MemoryGroupBase |
Memory group. More... | |
class | MemoryManagerOnDemand |
On-demand memory manager. More... | |
class | MemoryRegion |
Memory region CPU implementation. More... | |
struct | MinMaxLocationValues |
Min and max values and locations. More... | |
class | MultiHOG |
CPU implementation of multi HOG data-object. More... | |
class | MultiImage |
Basic implementation of the multi-planar image interface. More... | |
class | MultiImageInfo |
Store the multi-planar image's metadata. More... | |
class | NEAbsoluteDifference |
Basic function to run NEAbsoluteDifferenceKernel. More... | |
class | NEAbsoluteDifferenceKernel |
Interface for the absolute difference kernel. More... | |
class | NEAccumulate |
Basic function to run NEAccumulateKernel. More... | |
class | NEAccumulateKernel |
Interface for the accumulate kernel. More... | |
class | NEAccumulateSquared |
Basic function to run NEAccumulateSquaredKernel. More... | |
class | NEAccumulateSquaredKernel |
Interface for the accumulate squared kernel. More... | |
class | NEAccumulateWeighted |
Basic function to run NEAccumulateWeightedKernel. More... | |
class | NEAccumulateWeightedKernel |
Interface for the accumulate weighted kernel. More... | |
class | NEActivationLayer |
Basic function to run NEActivationLayerKernel. More... | |
class | NEActivationLayerKernel |
Interface for the activation layer kernel. More... | |
class | NEArithmeticAddition |
Basic function to run NEArithmeticAdditionKernel. More... | |
class | NEArithmeticAdditionKernel |
Interface for the kernel to perform addition between two tensors. More... | |
class | NEArithmeticSubtraction |
Basic function to run NEArithmeticSubtractionKernel. More... | |
class | NEArithmeticSubtractionKernel |
Interface for the kernel to perform subtraction between two tensors. More... | |
class | NEBatchNormalizationLayer |
Basic function to run NENormalizationLayerKernel and simulate a batch normalization layer. More... | |
class | NEBatchNormalizationLayerKernel |
Interface for the batch normalization layer kernel. More... | |
class | NEBitwiseAnd |
Basic function to run NEBitwiseAndKernel. More... | |
class | NEBitwiseAndKernel |
Interface for the kernel to perform bitwise AND between XY-planes of two tensors. More... | |
class | NEBitwiseNot |
Basic function to run NEBitwiseNotKernel. More... | |
class | NEBitwiseNotKernel |
Interface for the kernel to perform bitwise NOT operation. More... | |
class | NEBitwiseOr |
Basic function to run NEBitwiseOrKernel. More... | |
class | NEBitwiseOrKernel |
Interface for the kernel to perform bitwise inclusive OR between two tensors. More... | |
class | NEBitwiseXor |
Basic function to run NEBitwiseXorKernel. More... | |
class | NEBitwiseXorKernel |
Interface for the kernel to perform bitwise exclusive OR (XOR) between two tensors. More... | |
class | NEBox3x3 |
Basic function to execute box filter 3x3. More... | |
class | NEBox3x3Kernel |
NEON kernel to perform a Box 3x3 filter. More... | |
class | NECannyEdge |
Basic function to execute canny edge on NEON. More... | |
class | NEChannelCombine |
Basic function to run NEChannelCombineKernel to perform channel combination. More... | |
class | NEChannelCombineKernel |
Interface for the channel combine kernel. More... | |
class | NEChannelExtract |
Basic function to run NEChannelExtractKernel to perform channel extraction. More... | |
class | NEChannelExtractKernel |
Interface for the channel extract kernel. More... | |
class | NECol2Im |
Basic function to run NECol2Im. More... | |
class | NECol2ImKernel |
NEON kernel to perform col2im reshaping. More... | |
class | NEColorConvert |
Basic function to run NEColorConvertKernel to perform color conversion. More... | |
class | NEColorConvertKernel |
Interface for the color convert kernel. More... | |
class | NEConvertFullyConnectedWeights |
Basic function to run NEConvertFullyConnectedWeightsKernel. More... | |
class | NEConvertFullyConnectedWeightsKernel |
Interface to convert the 2D Fully Connected weights from NCHW to NHWC or vice versa. More... | |
class | NEConvolution3x3 |
Basic function to execute convolution of size 3x3. More... | |
class | NEConvolutionKernel |
Interface for the kernel to run an arbitrary size convolution on a tensor. More... | |
class | NEConvolutionLayer |
Basic function to simulate a convolution layer. More... | |
class | NEConvolutionLayerReshapeWeights |
Function to reshape and perform 1xW transposition on the weights. More... | |
class | NEConvolutionRectangle |
Basic function to execute non-square convolution. More... | |
class | NEConvolutionRectangleKernel |
Kernel for the running convolution on a rectangle matrix. More... | |
class | NEConvolutionSquare |
Basic function to execute convolution of size 5x5, 7x7, 9x9. More... | |
class | NECumulativeDistributionKernel |
Interface for the cumulative distribution (cummulative summmation) calculation kernel. More... | |
class | NEDeconvolutionLayer |
Function to run the deconvolution layer. More... | |
class | NEDepthConcatenateLayer |
Basic function to execute concatenate tensors along z axis. More... | |
class | NEDepthConcatenateLayerKernel |
Interface for the depth concatenate kernel. More... | |
class | NEDepthConvertLayer |
Basic function to run NEDepthConvertLayerKernel. More... | |
class | NEDepthConvertLayerKernel |
Depth conversion kernel. More... | |
class | NEDepthwiseConvolutionLayer |
Basic function to execute a generic depthwise convolution. More... | |
class | NEDepthwiseConvolutionLayer3x3 |
Basic function to execute a depthwise convolution for kernel size 3x3xC. More... | |
class | NEDepthwiseConvolutionLayer3x3Kernel |
Interface for the kernel to run a 3x3 depthwise convolution on a tensor. More... | |
class | NEDepthwiseIm2ColKernel |
Interface for the depthwise im2col reshape kernel. More... | |
class | NEDepthwiseSeparableConvolutionLayer |
Basic function to execute depthwise convolution. More... | |
class | NEDepthwiseVectorToTensorKernel |
Interface for the depthwise vector to tensor kernel. More... | |
class | NEDepthwiseWeightsReshapeKernel |
Interface for the depthwise weights reshape kernel. More... | |
class | NEDequantizationLayer |
Basic function to simulate a dequantization layer. More... | |
class | NEDequantizationLayerKernel |
Interface for the dequantization layer kernel. More... | |
class | NEDerivative |
Basic function to execute first order derivative operator. More... | |
class | NEDerivativeKernel |
Interface for the kernel to run the derivative along the X/Y directions on a tensor. More... | |
class | NEDilate |
Basic function to execute dilate. More... | |
class | NEDilateKernel |
Interface for the kernel to perform boolean image dilatation. More... | |
class | NEDirectConvolutionLayer |
Function to run the direct convolution. More... | |
class | NEDirectConvolutionLayerKernel |
NEON interface for Direct Convolution Layer kernel. More... | |
class | NEDirectConvolutionLayerOutputStageKernel |
NEON kernel to accumulate the biases, if provided, or downscale in case of quantized input. More... | |
class | NEEdgeNonMaxSuppressionKernel |
NEON kernel to perform Non-Maxima suppression for Canny Edge. More... | |
class | NEEdgeTraceKernel |
NEON kernel to perform Edge tracing. More... | |
class | NEEqualizeHistogram |
Basic function to execute histogram equalization. More... | |
class | NEErode |
Basic function to execute erode. More... | |
class | NEErodeKernel |
Interface for the kernel to perform boolean image erosion. More... | |
class | NEFastCorners |
Basic function to execute fast corners. More... | |
class | NEFastCornersKernel |
NEON kernel to perform fast corners. More... | |
class | NEFillArrayKernel |
This kernel adds all texels greater than or equal to the threshold value to the keypoint array. More... | |
class | NEFillBorder |
Basic function to run NEFillBorderKernel. More... | |
class | NEFillBorderKernel |
Interface for the kernel to fill borders. More... | |
class | NEFillInnerBorderKernel |
Interface for the kernel to fill the interior borders. More... | |
class | NEFlattenLayer |
Basic function to execute flatten. More... | |
class | NEFloor |
Basic function to run NEFloorKernel. More... | |
class | NEFloorKernel |
NEON kernel to perform a floor operation. More... | |
class | NEFullyConnectedLayer |
Basic function to compute a Fully Connected layer on NEON. More... | |
class | NEFullyConnectedLayerReshapeWeights |
Basic function to reshape the weights of Fully Connected layer with NEON. More... | |
class | NEGaussian3x3 |
Basic function to execute gaussian filter 3x3. More... | |
class | NEGaussian3x3Kernel |
NEON kernel to perform a Gaussian 3x3 filter. More... | |
class | NEGaussian5x5 |
Basic function to execute gaussian filter 5x5. More... | |
class | NEGaussian5x5HorKernel |
NEON kernel to perform a Gaussian 5x5 filter (horizontal pass) More... | |
class | NEGaussian5x5VertKernel |
NEON kernel to perform a Gaussian 5x5 filter (vertical pass) More... | |
class | NEGaussianPyramid |
Common interface for all Gaussian pyramid functions. More... | |
class | NEGaussianPyramidHalf |
Basic function to execute gaussian pyramid with HALF scale factor. More... | |
class | NEGaussianPyramidHorKernel |
NEON kernel to perform a GaussianPyramid (horizontal pass) More... | |
class | NEGaussianPyramidOrb |
Basic function to execute gaussian pyramid with ORB scale factor. More... | |
class | NEGaussianPyramidVertKernel |
NEON kernel to perform a GaussianPyramid (vertical pass) More... | |
class | NEGEMM |
Basic function to execute GEMM on NEON. More... | |
class | NEGEMMAssemblyBaseKernel |
Base class for GEMM NEON kernels implemented in Assembly. More... | |
class | NEGEMMConvolutionLayer |
Basic function to simulate a convolution layer. More... | |
class | NEGEMMInterleave4x4 |
Basic function to execute NEGEMMInterleave4x4Kernel. More... | |
class | NEGEMMInterleave4x4Kernel |
NEON kernel to interleave the elements of a matrix. More... | |
class | NEGEMMLowpAssemblyMatrixMultiplyCore |
Basic function to execute matrix multiply assembly kernels. More... | |
class | NEGEMMLowpMatrixAReductionKernel |
NEON kernel used to compute the row-vectors of sums of all the entries in each row of Matrix A. More... | |
class | NEGEMMLowpMatrixBReductionKernel |
NEON kernel used to compute the row-vectors of sums of all the entries in each column of Matrix B. More... | |
class | NEGEMMLowpMatrixMultiplyCore |
Basic function to execute GEMMLowpMatrixMultiplyCore on NEON. More... | |
class | NEGEMMLowpMatrixMultiplyKernel |
NEON kernel to multiply matrices. More... | |
class | NEGEMMLowpOffsetContributionKernel |
NEON kernel used to add the offset contribution after NEGEMMLowpMatrixMultiplyKernel. More... | |
class | NEGEMMLowpQuantizeDownInt32ToUint8Scale |
Basic function to execute NEGEMMLowpQuantizeDownInt32ToUint8Scale on NEON. More... | |
class | NEGEMMLowpQuantizeDownInt32ToUint8ScaleByFixedPoint |
Basic function to execute NEGEMMLowpQuantizeDownInt32ToUint8ScaleByFixedPoint on NEON. More... | |
class | NEGEMMLowpQuantizeDownInt32ToUint8ScaleByFixedPointKernel |
NEON kernel used to quantize down the int32 accumulator values of GEMMLowp to QASYMM8. More... | |
class | NEGEMMLowpQuantizeDownInt32ToUint8ScaleKernel |
NEON kernel used to quantize down the int32 accumulator values of GEMMLowp to QASYMM8. More... | |
class | NEGEMMMatrixAccumulateBiasesKernel |
NEON kernel to add a bias to each row of the input tensor. More... | |
class | NEGEMMMatrixAdditionKernel |
NEON kernel to perform the in-place matrix addition between 2 matrices taking into account that the second matrix might be weighted by a scalar value beta: More... | |
class | NEGEMMMatrixMultiplyKernel |
NEON kernel to multiply two input matrices "A" and "B". More... | |
class | NEGEMMMatrixVectorMultiplyKernel |
Interface for the GEMM matrix vector multiply kernel. More... | |
class | NEGEMMTranspose1xW |
Basic function to execute NEGEMMTranspose1xWKernel. More... | |
class | NEGEMMTranspose1xWKernel |
NEON kernel which transposes the elements of a matrix in chunks of 1xW, where W is equal to (16 / element size of the tensor) More... | |
class | NEGradientKernel |
Computes magnitude and quantised phase from inputs gradients. More... | |
class | NEHarrisCorners |
Basic function to execute harris corners detection. More... | |
class | NEHarrisScoreKernel |
Template NEON kernel to perform Harris Score. More... | |
class | NEHistogram |
Basic function to run NEHistogramKernel. More... | |
class | NEHistogramKernel |
Interface for the histogram kernel. More... | |
class | NEHOGBlockNormalizationKernel |
NEON kernel to perform HOG block normalization. More... | |
class | NEHOGDescriptor |
Basic function to calculate HOG descriptor. More... | |
class | NEHOGDetector |
Basic function to execute HOG detector based on linear SVM. More... | |
class | NEHOGDetectorKernel |
NEON kernel to perform HOG detector kernel using linear SVM. More... | |
class | NEHOGGradient |
Basic function to calculate the gradient for HOG. More... | |
class | NEHOGMultiDetection |
Basic function to detect multiple objects (or the same object at different scales) on the same input image using HOG. More... | |
class | NEHOGOrientationBinningKernel |
NEON kernel to perform HOG Orientation Binning. More... | |
class | NEIm2Col |
Basic function to run NEIm2ColKernel. More... | |
class | NEIm2ColKernel |
Interface for the im2col reshape kernel. More... | |
class | NEIntegralImage |
Basic function to run a NEIntegralImageKernel. More... | |
class | NEIntegralImageKernel |
Kernel to perform an image integral on an image. More... | |
class | NEL2NormalizeLayer |
Basic function to perform a L2 normalization on a given axis. More... | |
class | NEL2NormalizeLayerKernel |
Interface for performing a L2 normalize on a given axis given the square sum of it in this axis. More... | |
class | NELaplacianPyramid |
Basic function to execute laplacian pyramid. More... | |
class | NELaplacianReconstruct |
Basic function to execute laplacian reconstruction. More... | |
struct | NELKInternalKeypoint |
Internal keypoint class for Lucas-Kanade Optical Flow. More... | |
class | NELKTrackerKernel |
Interface for the Lucas-Kanade tracker kernel. More... | |
class | NELocallyConnectedLayer |
Basic function to compute the locally connected layer. More... | |
class | NELocallyConnectedMatrixMultiplyKernel |
NEON kernel to multiply each row of first tensor with low 2 dimensions of second tensor. More... | |
class | NELogits1DMaxKernel |
Interface for the identifying the max value of 1D Logits. More... | |
class | NELogits1DSoftmaxKernel |
Interface for softmax computation for QASYMM8 with pre-computed max. More... | |
class | NEMagnitude |
Basic function to run NEMagnitudePhaseKernel. More... | |
class | NEMagnitudePhaseKernel |
Template interface for the kernel to compute magnitude and phase. More... | |
class | NEMeanStdDev |
Basic function to execute mean and std deviation. More... | |
class | NEMeanStdDevKernel |
Interface for the kernel to calculate mean and standard deviation of input image pixels. More... | |
class | NEMedian3x3 |
Basic function to execute median filter. More... | |
class | NEMedian3x3Kernel |
Kernel to perform a median filter on a tensor. More... | |
class | NEMinMaxKernel |
Interface for the kernel to perform min max search on an image. More... | |
class | NEMinMaxLayerKernel |
Interface for the kernel to perform min max search on a 3D tensor. More... | |
class | NEMinMaxLocation |
Basic function to execute min and max location. More... | |
class | NEMinMaxLocationKernel |
Interface for the kernel to find min max locations of an image. More... | |
class | NENonLinearFilter |
Basic function to execute non linear filter. More... | |
class | NENonLinearFilterKernel |
Interface for the kernel to apply a non-linear filter. More... | |
class | NENonMaximaSuppression3x3 |
Basic function to execute non-maxima suppression over a 3x3 window. More... | |
class | NENonMaximaSuppression3x3Kernel |
Interface to perform Non-Maxima suppression over a 3x3 window using NEON. More... | |
class | NENormalizationLayer |
Basic function to compute a normalization layer. More... | |
class | NENormalizationLayerKernel |
Interface for the normalization layer kernel. More... | |
class | NEOpticalFlow |
Basic function to execute optical flow. More... | |
class | NEPermute |
Basic function to run NEPermuteKernel. More... | |
class | NEPermuteKernel |
NEON kernel to perform tensor permutation. More... | |
class | NEPhase |
Basic function to run NEMagnitudePhaseKernel. More... | |
class | NEPixelWiseMultiplication |
Basic function to run NEPixelWiseMultiplicationKernel. More... | |
class | NEPixelWiseMultiplicationKernel |
Interface for the kernel to perform addition between two tensors. More... | |
class | NEPoolingLayer |
Basic function to simulate a pooling layer with the specified pooling operation. More... | |
class | NEPoolingLayerKernel |
Interface for the pooling layer kernel. More... | |
class | NEQuantizationLayer |
Basic function to simulate a quantization layer. More... | |
class | NEQuantizationLayerKernel |
Interface for the quantization layer kernel. More... | |
class | NEReductionOperation |
Basic function to simulate a reduction operation. More... | |
class | NEReductionOperationKernel |
NEON kernel to perform a reduction operation. More... | |
class | NERemap |
Basic function to execute remap. More... | |
class | NERemapKernel |
NEON kernel to perform a remap on a tensor. More... | |
class | NEReshapeLayer |
Basic function to run NEReshapeLayerKernel. More... | |
class | NEReshapeLayerKernel |
Interface for the kernel to perform tensor reshaping. More... | |
class | NEROIPoolingLayer |
Basic function to run NEROIPoolingLayerKernel. More... | |
class | NEROIPoolingLayerKernel |
Interface for the ROI pooling layer kernel. More... | |
class | NEScale |
Basic function to run NEScaleKernel. More... | |
class | NEScaleKernel |
NEON kernel to perform scaling on a tensor. More... | |
class | NEScharr3x3 |
Basic function to execute scharr 3x3 filter. More... | |
class | NEScharr3x3Kernel |
Interface for the kernel to run a 3x3 Scharr filter on a tensor. More... | |
class | NESeparableConvolutionHorKernel |
Kernel for the Horizontal pass of a Separable Convolution. More... | |
class | NESeparableConvolutionVertKernel |
Kernel for the Vertical pass of a Separable Convolution. More... | |
class | NESobel3x3 |
Basic function to execute sobel 3x3 filter. More... | |
class | NESobel3x3Kernel |
Interface for the kernel to run a 3x3 Sobel X filter on a tensor. More... | |
class | NESobel5x5 |
Basic function to execute sobel 5x5 filter. More... | |
class | NESobel5x5HorKernel |
Interface for the kernel to run the horizontal pass of 5x5 Sobel filter on a tensor. More... | |
class | NESobel5x5VertKernel |
Interface for the kernel to run the vertical pass of 5x5 Sobel Y filter on a tensor. More... | |
class | NESobel7x7 |
Basic function to execute sobel 7x7 filter. More... | |
class | NESobel7x7HorKernel |
Interface for the kernel to run the horizontal pass of 7x7 Sobel filter on a tensor. More... | |
class | NESobel7x7VertKernel |
Interface for the kernel to run the vertical pass of 7x7 Sobel Y filter on a tensor. More... | |
class | NESoftmaxLayer |
Basic function to compute a SoftmaxLayer. More... | |
class | NETableLookup |
Basic function to run NETableLookupKernel. More... | |
class | NETableLookupKernel |
Interface for the kernel to perform table lookup calculations. More... | |
class | NEThreshold |
Basic function to run NEThresholdKernel. More... | |
class | NEThresholdKernel |
Interface for the thresholding kernel. More... | |
class | NETranspose |
Basic function to transpose a matrix on NEON. More... | |
class | NETransposeKernel |
NEON kernel which transposes the elements of a matrix. More... | |
class | NEWarpAffine |
Basic function to run NEWarpAffineKernel. More... | |
class | NEWarpAffineKernel |
Template interface for the kernel to compute warp affine. More... | |
class | NEWarpPerspective |
Basic function to run NEWarpPerspectiveKernel. More... | |
class | NEWarpPerspectiveKernel |
Template interface for the kernel to compute warp perspective. More... | |
class | NEWeightsReshapeKernel |
NEON kernel to perform reshaping on the weights used by convolution and locally connected layer. More... | |
class | NEWinogradConvolutionLayer |
Basic function to simulate a convolution layer. More... | |
class | NEWinogradLayerBatchedGEMMKernel |
NEON kernel to perform Winograd. More... | |
class | NEWinogradLayerTransformInputKernel |
NEON kernel to perform Winograd input transform. More... | |
class | NEWinogradLayerTransformOutputKernel |
NEON kernel to perform Winograd output transform. More... | |
class | NEWinogradLayerTransformWeightsKernel |
NEON kernel to perform Winograd weights transform. More... | |
class | NormalizationLayerInfo |
Normalization Layer Information class. More... | |
class | OffsetLifetimeManager |
Concrete class that tracks the lifetime of registered tensors and calculates the systems memory requirements in terms of a single blob and a list of offsets. More... | |
class | OffsetMemoryPool |
Offset based memory pool. More... | |
class | OMPScheduler |
Pool of threads to automatically split a kernel's execution among several threads. More... | |
struct | OpticalFlowParameters |
Parameters of Optical Flow algorithm. More... | |
class | PadStrideInfo |
Padding and stride information class. More... | |
class | PixelValue |
Class describing the value of a pixel for any image format. More... | |
class | PoolingLayerInfo |
Pooling Layer Information class. More... | |
class | PoolManager |
Memory pool manager. More... | |
class | Program |
Program class. More... | |
class | Pyramid |
Basic implementation of the pyramid interface. More... | |
class | PyramidInfo |
Store the Pyramid's metadata. More... | |
struct | QuantizationInfo |
Quantization settings (used for QASYMM8 data type) More... | |
struct | Rectangle |
Rectangle type. More... | |
struct | ROI |
Region of interest. More... | |
class | ROIPoolingLayerInfo |
ROI Pooling Layer Information class. More... | |
class | Scheduler |
Configurable scheduler which supports multiple multithreading APIs and choosing between different schedulers at runtime. More... | |
class | Semaphore |
Semamphore class. More... | |
class | SingleThreadScheduler |
Pool of threads to automatically split a kernel's execution among several threads. More... | |
class | Size2D |
Class for specifying the size of an image or rectangle. More... | |
class | Status |
Status class. More... | |
class | Steps |
Class to describe a number of elements in each dimension. More... | |
class | Strides |
Strides of an item in bytes. More... | |
class | SubTensor |
Basic implementation of the sub-tensor interface. More... | |
class | SubTensorInfo |
Store the sub tensor's metadata. More... | |
class | Tensor |
Basic implementation of the tensor interface. More... | |
class | TensorAllocator |
Basic implementation of a CPU memory tensor allocator. More... | |
class | TensorInfo |
Store the tensor's metadata. More... | |
class | TensorShape |
Shape of a tensor. More... | |
struct | ThreadInfo |
Information about executing thread and CPU. More... | |
struct | ValidRegion |
Container for valid region of a window. More... | |
class | WeightsInfo |
Convolution Layer Weights Information class. More... | |
class | Window |
Describe a multidimensional execution window. More... | |
struct | WinogradInfo |
Winograd information. More... | |
Typedefs | |
using | ICLKeyPointArray = ICLArray< KeyPoint > |
Interface for OpenCL Array of Key Points. More... | |
using | ICLCoordinates2DArray = ICLArray< Coordinates2D > |
Interface for OpenCL Array of 2D Coordinates. More... | |
using | ICLDetectionWindowArray = ICLArray< DetectionWindow > |
Interface for OpenCL Array of Detection Windows. More... | |
using | ICLROIArray = ICLArray< ROI > |
Interface for OpenCL Array of ROIs. More... | |
using | ICLSize2DArray = ICLArray< Size2D > |
Interface for OpenCL Array of 2D Sizes. More... | |
using | ICLUInt8Array = ICLArray< cl_uchar > |
Interface for OpenCL Array of uint8s. More... | |
using | ICLUInt16Array = ICLArray< cl_ushort > |
Interface for OpenCL Array of uint16s. More... | |
using | ICLUInt32Array = ICLArray< cl_uint > |
Interface for OpenCL Array of uint32s. More... | |
using | ICLInt16Array = ICLArray< cl_short > |
Interface for OpenCL Array of int16s. More... | |
using | ICLInt32Array = ICLArray< cl_int > |
Interface for OpenCL Array of int32s. More... | |
using | ICLFloatArray = ICLArray< cl_float > |
Interface for OpenCL Array of floats. More... | |
using | ICLImage = ICLTensor |
Interface for OpenCL images. More... | |
using | CLConvolution3x3Kernel = CLConvolutionKernel< 3 > |
Interface for the kernel which applies a 3x3 convolution to a tensor. More... | |
using | CLConvolution5x5Kernel = CLConvolutionKernel< 5 > |
Interface for the kernel which applies a 5x5 convolution to a tensor. More... | |
using | CLConvolution7x7Kernel = CLConvolutionKernel< 7 > |
Interface for the kernel which applies a 7x7 convolution to a tensor. More... | |
using | CLConvolution9x9Kernel = CLConvolutionKernel< 9 > |
Interface for the kernel which applies a 9x9 convolution to a tensor. More... | |
using | CLSeparableConvolution5x5HorKernel = CLSeparableConvolutionHorKernel< 5 > |
Interface for the kernel which applies a horizontal pass of 5x5 convolution to a tensor. More... | |
using | CLSeparableConvolution7x7HorKernel = CLSeparableConvolutionHorKernel< 7 > |
Interface for the kernel which applies a horizontal pass of 7x7 convolution to a tensor. More... | |
using | CLSeparableConvolution9x9HorKernel = CLSeparableConvolutionHorKernel< 9 > |
Interface for the kernel which applies a horizontal pass of 9x9 convolution to a tensor. More... | |
using | CLSeparableConvolution5x5VertKernel = CLSeparableConvolutionVertKernel< 5 > |
Interface for the kernel which applies a vertical pass of 5x5 convolution to a tensor. More... | |
using | CLSeparableConvolution7x7VertKernel = CLSeparableConvolutionVertKernel< 7 > |
Interface for the kernel which applies a vertical pass of 7x7 convolution to a tensor. More... | |
using | CLSeparableConvolution9x9VertKernel = CLSeparableConvolutionVertKernel< 9 > |
Interface for the kernel which applies a vertical pass of 9x9 convolution to a tensor. More... | |
using | ICLLKInternalKeypointArray = ICLArray< CLLKInternalKeypoint > |
Interface for OpenCL Array of Internal Key Points. More... | |
using | ICLCoefficientTableArray = ICLArray< CLCoefficientTable > |
Interface for OpenCL Array of Coefficient Tables. More... | |
using | ICLOldValArray = ICLArray< CLOldValue > |
Interface for OpenCL Array of Old Values. More... | |
using | IImage = ITensor |
Interface for CPP Images. More... | |
using | qint8_t = int8_t |
8 bit fixed point scalar value More... | |
using | qint16_t = int16_t |
16 bit fixed point scalar value More... | |
using | qint32_t = int32_t |
32 bit fixed point scalar value More... | |
using | qint64_t = int64_t |
64 bit fixed point scalar value More... | |
using | IGCImage = IGCTensor |
Interface for GLES Compute image. More... | |
using | GCDirectConvolutionLayer1x1Kernel = GCDirectConvolutionLayerKernel< 1 > |
Interface for the 1x1 direct convolution kernel. More... | |
using | GCDirectConvolutionLayer3x3Kernel = GCDirectConvolutionLayerKernel< 3 > |
Interface for the 3x3 direct convolution kernel. More... | |
using | GCDirectConvolutionLayer5x5Kernel = GCDirectConvolutionLayerKernel< 5 > |
Interface for the 5x5 direct convolution kernel. More... | |
using | IKeyPointArray = IArray< KeyPoint > |
Interface for Array of Key Points. More... | |
using | ICoordinates2DArray = IArray< Coordinates2D > |
Interface for Array of 2D Coordinates. More... | |
using | IDetectionWindowArray = IArray< DetectionWindow > |
Interface for Array of Detection Windows. More... | |
using | IROIArray = IArray< ROI > |
Interface for Array of ROIs. More... | |
using | ISize2DArray = IArray< Size2D > |
Interface for Array of 2D Sizes. More... | |
using | IUInt8Array = IArray< uint8_t > |
Interface for Array of uint8s. More... | |
using | IUInt16Array = IArray< uint16_t > |
Interface for Array of uint16s. More... | |
using | IUInt32Array = IArray< uint32_t > |
Interface for Array of uint32s. More... | |
using | IInt16Array = IArray< int16_t > |
Interface for Array of int16s. More... | |
using | IInt32Array = IArray< int32_t > |
Interface for Array of int32s. More... | |
using | IFloatArray = IArray< float > |
Interface for Array of floats. More... | |
using | INEKernel = ICPPKernel |
Common interface for all kernels implemented in NEON. More... | |
using | INESimpleKernel = ICPPSimpleKernel |
Interface for simple NEON kernels having 1 tensor input and 1 tensor output. More... | |
using | NEAccumulateWeightedFP16Kernel = NEAccumulateWeightedKernel |
Interface for the accumulate weighted kernel using F16. More... | |
using | NEBox3x3FP16Kernel = NEBox3x3Kernel |
NEON kernel to perform a Box 3x3 filter for FP16 datatype. More... | |
using | NEGradientFP16Kernel = NEGradientKernel |
NEON kernel to perform Gradient computation for FP16 datatype. More... | |
using | NEConvolution3x3Kernel = NEConvolutionKernel< 3 > |
Interface for the kernel which applied a 3x3 convolution to a tensor. More... | |
using | NEConvolution5x5Kernel = NEConvolutionKernel< 5 > |
Interface for the kernel which applied a 5x5 convolution to a tensor. More... | |
using | NEConvolution7x7Kernel = NEConvolutionKernel< 7 > |
Interface for the kernel which applied a 7x7 convolution to a tensor. More... | |
using | NEConvolution9x9Kernel = NEConvolutionKernel< 9 > |
Interface for the kernel which applied a 9x9 convolution to a tensor. More... | |
using | NESeparableConvolution5x5HorKernel = NESeparableConvolutionHorKernel< 5 > |
Interface for the kernel which applied a 5x1 horizontal convolution to a tensor. More... | |
using | NESeparableConvolution7x7HorKernel = NESeparableConvolutionHorKernel< 7 > |
Interface for the kernel which applied a 7x1 horizontal convolution to a tensor. More... | |
using | NESeparableConvolution9x9HorKernel = NESeparableConvolutionHorKernel< 9 > |
Interface for the kernel which applied a 9x1 horizontal convolution to a tensor. More... | |
using | NESeparableConvolution5x5VertKernel = NESeparableConvolutionVertKernel< 5 > |
Interface for the kernel which applied a 1x5 vertical convolution to a tensor. More... | |
using | NESeparableConvolution7x7VertKernel = NESeparableConvolutionVertKernel< 7 > |
Interface for the kernel which applied a 1x7 vertical convolution to a tensor. More... | |
using | NESeparableConvolution9x9VertKernel = NESeparableConvolutionVertKernel< 9 > |
Interface for the kernel which applied a 1x9 vertical convolution to a tensor. More... | |
template<int32_t block_size> | |
using | NEHarrisScoreFP16Kernel = NEHarrisScoreKernel< block_size > |
Interface for the accumulate Weighted kernel using FP16. More... | |
using | INELKInternalKeypointArray = IArray< NELKInternalKeypoint > |
Interface for NEON Array of Internal Key Points. More... | |
template<MagnitudeType mag_type, PhaseType phase_type> | |
using | NEMagnitudePhaseFP16Kernel = NEMagnitudePhaseKernel< mag_type, phase_type > |
Template interface for the kernel to compute magnitude and phase. More... | |
using | NENonMaximaSuppression3x3FP16Kernel = NENonMaximaSuppression3x3Kernel |
NEON kernel to perform Non-Maxima suppression 3x3 with intermediate results in FP16 if the input data type is FP32. More... | |
using | qasymm8x8_t = uint8x8_t |
8 bit quantized asymmetric vector with 8 elements More... | |
using | qasymm8x8x2_t = uint8x8x2_t |
8 bit quantized asymmetric vector with 16 elements More... | |
using | qasymm8x8x3_t = uint8x8x3_t |
8 bit quantized asymmetric vector with 24 elements More... | |
using | qasymm8x8x4_t = uint8x8x4_t |
8 bit quantized asymmetric vector with 32 elements More... | |
using | qasymm8x16_t = uint8x16_t |
8 bit quantized asymmetric vector with 16 elements More... | |
using | qint8x8_t = int8x8_t |
8 bit fixed point vector with 8 elements More... | |
using | qint8x8x2_t = int8x8x2_t |
8 bit fixed point vector with 16 elements More... | |
using | qint8x8x3_t = int8x8x3_t |
8 bit fixed point vector with 24 elements More... | |
using | qint8x8x4_t = int8x8x4_t |
8 bit fixed point vector with 32 elements More... | |
using | qint8x16_t = int8x16_t |
8 bit fixed point vector with 16 elements More... | |
using | qint8x16x2_t = int8x16x2_t |
8 bit fixed point vector with 32 elements More... | |
using | qint8x16x3_t = int8x16x3_t |
8 bit fixed point vector with 48 elements More... | |
using | qint8x16x4_t = int8x16x4_t |
8 bit fixed point vector with 64 elements More... | |
using | qint16x4_t = int16x4_t |
16 bit fixed point vector with 4 elements More... | |
using | qint16x4x2_t = int16x4x2_t |
16 bit fixed point vector with 8 elements More... | |
using | qint16x4x3_t = int16x4x3_t |
16 bit fixed point vector with 12 elements More... | |
using | qint16x4x4_t = int16x4x4_t |
16 bit fixed point vector with 16 elements More... | |
using | qint16x8_t = int16x8_t |
16 bit fixed point vector with 8 elements More... | |
using | qint16x8x2_t = int16x8x2_t |
16 bit fixed point vector with 16 elements More... | |
using | qint16x8x3_t = int16x8x3_t |
16 bit fixed point vector with 24 elements More... | |
using | qint16x8x4_t = int16x8x4_t |
16 bit fixed point vector with 32 elements More... | |
using | qint32x2_t = int32x2_t |
32 bit fixed point vector with 2 elements More... | |
using | qint32x4_t = int32x4_t |
32 bit fixed point vector with 4 elements More... | |
using | qint32x4x2_t = int32x4x2_t |
32 bit fixed point vector with 8 elements More... | |
using | qasymm8_t = uint8_t |
8 bit quantized asymmetric scalar value More... | |
using | half = half_float::half |
16-bit floating point type More... | |
using | PermutationVector = Strides |
Permutation vector. More... | |
using | PaddingSize = BorderSize |
Container for 2D padding size. More... | |
using | InternalKeypoint = std::tuple< float, float, float > |
Internal key point. More... | |
using | KeyPointArray = Array< KeyPoint > |
Array of Key Points. More... | |
using | Coordinates2DArray = Array< Coordinates2D > |
Array of 2D Coordinates. More... | |
using | DetectionWindowArray = Array< DetectionWindow > |
Array of Detection Windows. More... | |
using | ROIArray = Array< ROI > |
Array of ROIs. More... | |
using | Size2DArray = Array< Size2D > |
Array of 2D Sizes. More... | |
using | UInt8Array = Array< uint8_t > |
Array of uint8s. More... | |
using | UInt16Array = Array< uint16_t > |
Array of uint16s. More... | |
using | UInt32Array = Array< uint32_t > |
Array of uint32s. More... | |
using | Int16Array = Array< int16_t > |
Array of int16s. More... | |
using | Int32Array = Array< int32_t > |
Array of int32s. More... | |
using | FloatArray = Array< float > |
Array of floats. More... | |
using | CLKeyPointArray = CLArray< KeyPoint > |
OpenCL Array of Key Points. More... | |
using | CLCoordinates2DArray = CLArray< Coordinates2D > |
OpenCL Array of 2D Coordinates. More... | |
using | CLDetectionWindowArray = CLArray< DetectionWindow > |
OpenCL Array of Detection Windows. More... | |
using | CLROIArray = CLArray< ROI > |
OpenCL Array of ROIs. More... | |
using | CLSize2DArray = CLArray< Size2D > |
OpenCL Array of 2D Sizes. More... | |
using | CLUInt8Array = CLArray< cl_uchar > |
OpenCL Array of uint8s. More... | |
using | CLUInt16Array = CLArray< cl_ushort > |
OpenCL Array of uint16s. More... | |
using | CLUInt32Array = CLArray< cl_uint > |
OpenCL Array of uint32s. More... | |
using | CLInt16Array = CLArray< cl_short > |
OpenCL Array of int16s. More... | |
using | CLInt32Array = CLArray< cl_int > |
OpenCL Array of int32s. More... | |
using | CLFloatArray = CLArray< cl_float > |
OpenCL Array of floats. More... | |
using | CLMemoryGroup = MemoryGroupBase< CLTensor > |
Memory Group in OpenCL. More... | |
using | CLImage = CLTensor |
OpenCL Image. More... | |
using | CLConvolution5x5 = CLConvolutionSquare< 5 > |
Basic function to run 5x5 convolution. More... | |
using | CLConvolution7x7 = CLConvolutionSquare< 7 > |
Basic function to run 7x7 convolution. More... | |
using | CLConvolution9x9 = CLConvolutionSquare< 9 > |
Basic function to run 9x9 convolution. More... | |
using | CLLKInternalKeypointArray = CLArray< CLLKInternalKeypoint > |
OpenCL Array of Internal Keypoints. More... | |
using | CLCoefficientTableArray = CLArray< CLCoefficientTable > |
OpenCL Array of Coefficient Tables. More... | |
using | CLOldValueArray = CLArray< CLOldValue > |
OpenCL Array of Old Values. More... | |
using | GCMemoryGroup = MemoryGroupBase< GCTensor > |
using | GCImage = GCTensor |
OpenGL ES Image. More... | |
using | MemoryGroup = MemoryGroupBase< Tensor > |
Memory Group. More... | |
using | AssemblyKernelGlueF32 = AssemblyKernelGlue< float, float > |
Float 32 assembly kernel glue. More... | |
using | AssemblyKernelGlueU8U32 = AssemblyKernelGlue< uint8_t, uint32_t > |
Uint 8 to Uint 32 kernel glue. More... | |
using | AssemblyKernelGlueS8S32 = AssemblyKernelGlue< int8_t, int32_t > |
Int 8 to Int 32 kernel glue. More... | |
using | NEConvolution5x5 = NEConvolutionSquare< 5 > |
Basic function to run 5x5 convolution. More... | |
using | NEConvolution7x7 = NEConvolutionSquare< 7 > |
Basic function to run 7x7 convolution. More... | |
using | NEConvolution9x9 = NEConvolutionSquare< 9 > |
Basic function to run 9x9 convolution. More... | |
using | LKInternalKeypointArray = Array< NELKInternalKeypoint > |
Array of LK Internel Keypoints. More... | |
using | NEScheduler = Scheduler |
NEON Scheduler. More... | |
using | Image = Tensor |
Image. More... | |
using | MemoryMappings = std::map< void **, size_t > |
A map of (handle, index/offset), where handle is the memory handle of the object to provide the memory for and index/offset is the buffer/offset from the pool that should be used. More... | |
using | GroupMappings = std::map< size_t, MemoryMappings > |
A map of the groups and memory mappings. More... | |
using | Mutex = std::mutex |
Wrapper of Mutex data-object. More... | |
Enumerations | |
enum | CLVersion { CL10, CL11, CL12, CL20, UNKNOWN } |
Available OpenCL Version. More... | |
enum | CPUModel { GENERIC, A53, A55r0, A55r1 } |
CPU models - we only need to detect CPUs we have microarchitecture-specific code for. More... | |
enum | ErrorCode { OK, RUNTIME_ERROR } |
Available error codes. More... | |
enum | GPUTarget { UNKNOWN = 0x101, GPU_ARCH_MASK = 0xF00, MIDGARD = 0x100, BIFROST = 0x200, T600 = 0x110, T700 = 0x120, T800 = 0x130, G71 = 0x210, G72 = 0x220, G51 = 0x230, G51BIG = 0x231, G51LIT = 0x232, TNOX = 0x240, TTRX = 0x250, TBOX = 0x260 } |
Available GPU Targets. More... | |
enum | RoundingPolicy { TO_ZERO, TO_NEAREST_UP, TO_NEAREST_EVEN } |
Rounding method. More... | |
enum | Format { UNKNOWN, U8, S16, U16, S32, U32, F16, F32, UV88, RGB888, RGBA8888, YUV444, YUYV422, NV12, NV21, IYUV, UYVY422 } |
Image colour formats. More... | |
enum | DataType { UNKNOWN, U8, S8, QS8, QASYMM8, U16, S16, QS16, U32, S32, QS32, U64, S64, F16, F32, F64, SIZET } |
Available data types. More... | |
enum | SamplingPolicy { CENTER, TOP_LEFT } |
Available Sampling Policies. More... | |
enum | DataLayout { UNKNOWN, NCHW, NHWC } |
Supported tensor data layouts. More... | |
enum | DataLayoutDimension { CHANNEL, HEIGHT, WIDTH, BATCHES } |
Supported tensor data layout dimensions. More... | |
enum | BorderMode { UNDEFINED, CONSTANT, REPLICATE } |
Methods available to handle borders. More... | |
enum | ConvertPolicy { WRAP, SATURATE } |
Policy to handle overflow. More... | |
enum | InterpolationPolicy { NEAREST_NEIGHBOR, BILINEAR, AREA } |
Interpolation method. More... | |
enum | BilinearInterpolation { BILINEAR_OLD_NEW, BILINEAR_SCHARR } |
Bilinear Interpolation method used by LKTracker. More... | |
enum | ThresholdType { BINARY, RANGE } |
Threshold mode. More... | |
enum | Termination { TERM_CRITERIA_EPSILON, TERM_CRITERIA_ITERATIONS, TERM_CRITERIA_BOTH } |
Termination criteria. More... | |
enum | MagnitudeType { L1NORM, L2NORM } |
Magnitude calculation type. More... | |
enum | PhaseType { SIGNED, UNSIGNED } |
Phase calculation type. More... | |
enum | Channel { UNKNOWN, C0, C1, C2, C3, R, G, B, A, Y, U, V } |
Available channels. More... | |
enum | MatrixPattern { BOX, CROSS, DISK, OTHER } |
Available matrix patterns. More... | |
enum | NonLinearFilterFunction : unsigned { MEDIAN = 0, MIN = 1, MAX = 2 } |
Available non linear functions. More... | |
enum | ReductionOperation { SUM_SQUARE, SUM } |
Available reduction operations. More... | |
enum | NormType { IN_MAP_1D, IN_MAP_2D, CROSS_MAP } |
The normalization type used for the normalization layer. More... | |
enum | HOGNormType { L2_NORM = 1, L2HYS_NORM = 2, L1_NORM = 3 } |
Normalization type for Histogram of Oriented Gradients (HOG) More... | |
enum | DimensionRoundingType { FLOOR, CEIL } |
Dimension rounding type when down-scaling on CNNs. More... | |
enum | PoolingType { MAX, AVG, L2 } |
Available pooling types. More... | |
enum | ConvolutionMethod { GEMM, DIRECT, WINOGRAD } |
Available ConvolutionMethod. More... | |
enum | MappingType { BLOBS, OFFSETS } |
Mapping type. More... | |
enum | FixedPointOp { ADD, SUB, MUL, EXP, LOG, INV_SQRT, RECIPROCAL } |
Fixed point operation. More... | |
enum | GradientDimension { GRAD_XY } |
Gradient dimension type. More... | |
Functions | |
std::string | get_cl_type_from_data_type (const DataType &dt) |
Translates a tensor data type to the appropriate OpenCL type. More... | |
std::string | get_data_size_from_data_type (const DataType &dt) |
Get the size of a data type in number of bits. More... | |
std::string | get_underlying_cl_type_from_data_type (const DataType &dt) |
Translates fixed point tensor data type to the underlying OpenCL type. More... | |
GPUTarget | get_target_from_device (cl::Device &device) |
Helper function to get the GPU target from CL device. More... | |
CLVersion | get_cl_version (const cl::Device &device) |
Helper function to get the highest OpenCL version supported. More... | |
bool | device_supports_extension (const cl::Device &device, const char *extension_name) |
Helper function to check whether a given extension is supported. More... | |
bool | fp16_supported (const cl::Device &device) |
Helper function to check whether the cl_khr_fp16 extension is supported. More... | |
bool | arm_non_uniform_workgroup_supported (const cl::Device &device) |
Helper function to check whether the arm_non_uniform_work_group_size extension is supported. More... | |
void | enqueue (cl::CommandQueue &queue, ICLKernel &kernel, const Window &window, const cl::NDRange &lws_hint=CLKernelLibrary::get().default_ndrange()) |
Add the kernel to the command queue with the given window. More... | |
bool | opencl_is_available () |
Check if OpenCL is available. More... | |
template<typename T > | |
bool | operator== (const Dimensions< T > &lhs, const Dimensions< T > &rhs) |
Check that given dimensions are equal. More... | |
template<typename T > | |
bool | operator!= (const Dimensions< T > &lhs, const Dimensions< T > &rhs) |
Check that given dimensions are not equal. More... | |
template<typename... T> | |
void | ignore_unused (T &&...) |
Ignores unused arguments. More... | |
Status | create_error_va_list (ErrorCode error_code, const char *function, const char *file, const int line, const char *msg, va_list args) |
Creates an error containing the error message from variable argument list. More... | |
Status | create_error (ErrorCode error_code, const char *function, const char *file, const int line, const char *msg,...) |
Creates an error containing the error message. More... | |
void | error (const char *function, const char *file, const int line, const char *msg,...) |
Print an error message then throw an std::runtime_error. More... | |
qint8_t | sqshl_qs8 (qint8_t a, int shift) |
8 bit fixed point scalar saturating shift left More... | |
qint8_t | sshr_qs8 (qint8_t a, int shift) |
8 bit fixed point scalar shift right More... | |
qint16_t | sshr_qs16 (qint16_t a, int shift) |
16 bit fixed point scalar shift right More... | |
qint16_t | sqshl_qs16 (qint16_t a, int shift) |
16 bit fixed point scalar saturating shift left More... | |
qint8_t | sabs_qs8 (qint8_t a) |
8 bit fixed point scalar absolute value More... | |
qint16_t | sabs_qs16 (qint16_t a) |
16 bit fixed point scalar absolute value More... | |
qint8_t | sadd_qs8 (qint8_t a, qint8_t b) |
8 bit fixed point scalar add More... | |
qint16_t | sadd_qs16 (qint16_t a, qint16_t b) |
16 bit fixed point scalar add More... | |
qint8_t | sqadd_qs8 (qint8_t a, qint8_t b) |
8 bit fixed point scalar saturating add More... | |
qint16_t | sqadd_qs16 (qint16_t a, qint16_t b) |
16 bit fixed point scalar saturating add More... | |
qint32_t | sqadd_qs32 (qint32_t a, qint32_t b) |
32 bit fixed point scalar saturating add More... | |
qint8_t | ssub_qs8 (qint8_t a, qint8_t b) |
8 bit fixed point scalar subtraction More... | |
qint16_t | ssub_qs16 (qint16_t a, qint16_t b) |
16 bit fixed point scalar subtraction More... | |
qint8_t | sqsub_qs8 (qint8_t a, qint8_t b) |
8 bit fixed point scalar saturating subtraction More... | |
qint16_t | sqsub_qs16 (qint16_t a, qint16_t b) |
16 bit fixed point scalar saturating subtraction More... | |
qint8_t | smul_qs8 (qint8_t a, qint8_t b, int fixed_point_position) |
8 bit fixed point scalar multiply More... | |
qint16_t | smul_qs16 (qint16_t a, qint16_t b, int fixed_point_position) |
16 bit fixed point scalar multiply More... | |
qint8_t | sqmul_qs8 (qint8_t a, qint8_t b, int fixed_point_position) |
8 bit fixed point scalar saturating multiply More... | |
qint16_t | sqmul_qs16 (qint16_t a, qint16_t b, int fixed_point_position) |
16 bit fixed point scalar saturating multiply More... | |
qint16_t | sqmull_qs8 (qint8_t a, qint8_t b, int fixed_point_position) |
8 bit fixed point scalar multiply long More... | |
qint32_t | sqmull_qs16 (qint16_t a, qint16_t b, int fixed_point_position) |
16 bit fixed point scalar multiply long More... | |
qint8_t | sinvsqrt_qs8 (qint8_t a, int fixed_point_position) |
8 bit fixed point scalar inverse square root More... | |
qint16_t | sinvsqrt_qs16 (qint16_t a, int fixed_point_position) |
16 bit fixed point scalar inverse square root More... | |
qint8_t | sdiv_qs8 (qint8_t a, qint8_t b, int fixed_point_position) |
8 bit fixed point scalar division More... | |
qint16_t | sdiv_qs16 (qint16_t a, qint16_t b, int fixed_point_position) |
16 bit fixed point scalar division More... | |
qint8_t | sqexp_qs8 (qint8_t a, int fixed_point_position) |
8 bit fixed point scalar exponential More... | |
qint16_t | sqexp_qs16 (qint16_t a, int fixed_point_position) |
16 bit fixed point scalar exponential More... | |
qint16_t | sexp_qs16 (qint16_t a, int fixed_point_position) |
16 bit fixed point scalar exponential More... | |
qint8_t | slog_qs8 (qint8_t a, int fixed_point_position) |
8 bit fixed point scalar logarithm More... | |
qint16_t | slog_qs16 (qint16_t a, int fixed_point_position) |
16 bit fixed point scalar logarithm More... | |
float | scvt_f32_qs8 (qint8_t a, int fixed_point_position) |
Convert an 8 bit fixed point to float. More... | |
qint8_t | sqcvt_qs8_f32 (float a, int fixed_point_position) |
Convert a float to 8 bit fixed point. More... | |
float | scvt_f32_qs16 (qint16_t a, int fixed_point_position) |
Convert a 16 bit fixed point to float. More... | |
qint16_t | sqcvt_qs16_f32 (float a, int fixed_point_position) |
Convert a float to 16 bit fixed point. More... | |
qint8_t | sqmovn_qs16 (qint16_t a) |
Scalar saturating move and narrow. More... | |
qint16_t | sqmovn_qs32 (qint32_t a) |
Scalar saturating move and narrow. More... | |
GPUTarget | get_target_from_device () |
Helper function to get the GPU target from GLES using GL_RENDERER enum. More... | |
void | enqueue (IGCKernel &kernel, const Window &window, const gles::NDRange &lws=gles::NDRange(1U, 1U, 1U)) |
Add the kernel to the command queue with the given window. More... | |
bool | opengles31_is_available () |
Check if the OpenGL ES 3.1 API is available at runtime. More... | |
const std::string & | string_from_target (GPUTarget target) |
Translates a given gpu device target to string. More... | |
GPUTarget | get_target_from_name (const std::string &device_name) |
Helper function to get the GPU target from a device name. More... | |
GPUTarget | get_arch_from_target (GPUTarget target) |
Helper function to get the GPU arch. More... | |
template<typename... Args> | |
bool | gpu_target_is_in (GPUTarget target_to_check, GPUTarget target, Args...targets) |
Helper function to check whether a gpu target is equal to the provided targets. More... | |
bool | gpu_target_is_in (GPUTarget target_to_check, GPUTarget target) |
Variant of gpu_target_is_in for comparing two targets. More... | |
template<typename Kernel , typename... T> | |
std::unique_ptr< Kernel > | create_configure_kernel (T &&...args) |
Helper function to create and return a unique_ptr pointed to a CL/GLES kernel object It also calls the kernel's configuration. More... | |
template<typename Kernel > | |
std::unique_ptr< Kernel > | create_kernel () |
Helper function to create and return a unique_ptr pointed to a CL/GLES kernel object. More... | |
template<typename T > | |
T | delta_bilinear_c1 (const T *pixel_ptr, size_t stride, float dx, float dy) |
Computes bilinear interpolation using the pointer to the top-left pixel and the pixel's distance between the real coordinates and the smallest following integer coordinates. More... | |
template<typename T > | |
T | delta_linear_c1_y (const T *pixel_ptr, size_t stride, float dy) |
Computes linear interpolation using the pointer to the top pixel and the pixel's distance between the real coordinates and the smallest following integer coordinates. More... | |
template<typename T > | |
T | delta_linear_c1_x (const T *pixel_ptr, float dx) |
Computes linear interpolation using the pointer to the left pixel and the pixel's distance between the real coordinates and the smallest following integer coordinates. More... | |
template<typename T > | |
T | pixel_bilinear_c1 (const T *first_pixel_ptr, size_t stride, float x, float y) |
Return the pixel at (x,y) using bilinear interpolation. More... | |
template<typename T > | |
uint8_t | pixel_bilinear_c1_clamp (const T *first_pixel_ptr, size_t stride, size_t width, size_t height, float x, float y) |
Return the pixel at (x,y) using bilinear interpolation by clamping when out of borders. More... | |
uint8_t | pixel_area_c1u8_clamp (const uint8_t *first_pixel_ptr, size_t stride, size_t width, size_t height, float wr, float hr, int x, int y) |
Return the pixel at (x,y) using area interpolation by clamping when out of borders. More... | |
template<typename L , typename... Ts> | |
void | execute_window_loop (const Window &w, L &&lambda_function, Ts &&...iterators) |
Iterate through the passed window, automatically adjusting the iterators and calling the lambda_functino for each element. More... | |
template<typename... Ts> | |
bool | update_window_and_padding (Window &win, Ts &&...patterns) |
Update window and padding size for each of the access patterns. More... | |
Window | calculate_max_window (const ValidRegion &valid_region, const Steps &steps=Steps(), bool skip_border=false, BorderSize border_size=BorderSize()) |
Calculate the maximum window for a given tensor shape and border setting. More... | |
Window | calculate_max_window (const ITensorInfo &info, const Steps &steps=Steps(), bool skip_border=false, BorderSize border_size=BorderSize()) |
Calculate the maximum window for a given tensor shape and border setting. More... | |
Window | calculate_max_window_horizontal (const ValidRegion &valid_region, const Steps &steps=Steps(), bool skip_border=false, BorderSize border_size=BorderSize()) |
Calculate the maximum window used by a horizontal kernel for a given tensor shape and border setting. More... | |
Window | calculate_max_window_horizontal (const ITensorInfo &info, const Steps &steps=Steps(), bool skip_border=false, BorderSize border_size=BorderSize()) |
Calculate the maximum window used by a horizontal kernel for a given tensor shape and border setting. More... | |
Window | calculate_max_enlarged_window (const ValidRegion &valid_region, const Steps &steps=Steps(), BorderSize border_size=BorderSize()) |
Calculate the maximum window for a given tensor shape and border setting. More... | |
Window | calculate_max_enlarged_window (const ITensorInfo &info, const Steps &steps=Steps(), BorderSize border_size=BorderSize()) |
Calculate the maximum window for a given tensor shape and border setting. More... | |
template<typename... Ts> | |
ValidRegion | intersect_valid_regions (const Ts &...regions) |
Intersect multiple valid regions. More... | |
template<typename T , typename... Ts> | |
Strides | compute_strides (const ITensorInfo &info, T stride_x, Ts &&...fixed_strides) |
Create a strides object based on the provided strides and the tensor dimensions. More... | |
template<typename... Ts> | |
Strides | compute_strides (const ITensorInfo &info) |
Create a strides object based on the tensor dimensions. More... | |
template<typename T > | |
void | permute (Dimensions< T > &dimensions, const PermutationVector &perm) |
Permutes given Dimensions according to a permutation vector. More... | |
void | permute (TensorShape &shape, const PermutationVector &perm) |
Permutes given TensorShape according to a permutation vector. More... | |
bool | auto_init_if_empty (ITensorInfo &info, const TensorShape &shape, int num_channels, DataType data_type, int fixed_point_position, QuantizationInfo quantization_info=QuantizationInfo()) |
Auto initialize the tensor info (shape, number of channels, data type and fixed point position) if the current assignment is empty. More... | |
bool | auto_init_if_empty (ITensorInfo &info_sink, const ITensorInfo &info_source) |
Auto initialize the tensor info using another tensor info. More... | |
bool | set_shape_if_empty (ITensorInfo &info, const TensorShape &shape) |
Set the shape to the specified value if the current assignment is empty. More... | |
bool | set_format_if_unknown (ITensorInfo &info, Format format) |
Set the format, data type and number of channels to the specified value if the current data type is unknown. More... | |
bool | set_data_type_if_unknown (ITensorInfo &info, DataType data_type) |
Set the data type and number of channels to the specified value if the current data type is unknown. More... | |
bool | set_data_layout_if_unknown (ITensorInfo &info, DataLayout data_layout) |
Set the data layout to the specified value if the current data layout is unknown. More... | |
bool | set_fixed_point_position_if_zero (ITensorInfo &info, int fixed_point_position) |
Set the fixed point position to the specified value if the current fixed point position is 0 and the data type is QS8 or QS16. More... | |
bool | set_quantization_info_if_empty (ITensorInfo &info, QuantizationInfo quantization_info) |
Set the quantization info to the specified value if the current quantization info is empty and the data type of asymmetric quantized type. More... | |
ValidRegion | calculate_valid_region_scale (const ITensorInfo &src_info, const TensorShape &dst_shape, InterpolationPolicy interpolate_policy, SamplingPolicy sampling_policy, bool border_undefined) |
Helper function to calculate the Valid Region for Scale. More... | |
Coordinates | index2coords (const TensorShape &shape, int index) |
Convert a linear index into n-dimensional coordinates. More... | |
int | coords2index (const TensorShape &shape, const Coordinates &coord) |
Convert n-dimensional coordinates into a linear index. More... | |
size_t | get_data_layout_dimension_index (const DataLayout data_layout, const DataLayoutDimension data_layout_dimension) |
Get the index of the given dimension. More... | |
int | adjust_down (int required, int available, int step) |
Decrease required in steps of step until it's less than available . More... | |
int | adjust_up (int required, int available, int step) |
Increase required in steps of step until it's greater than available . More... | |
int32x4_t | rounding_divide_by_pow2 (int32x4_t x, int exponent) |
Round to the nearest division by a power-of-two using exponent. More... | |
uint8x16_t | vmlaq_qasymm8 (qasymm8x16_t vd, float32x4_t vs, float32x4_t vo) |
Perform a multiply-accumulate on all 16 components of a QASYMM8 vector. More... | |
template<bool is_bounded_relu> | |
uint8x16_t | finalize_quantization (int32x4x4_t &in_s32, int result_fixedpoint_multiplier, int32_t result_shift, int32x4_t result_offset_after_shift_s32, uint8x16_t min_u8, uint8x16_t max_u8) |
Performs final quantization step on 16 elements. More... | |
void | colorconvert_rgb_to_rgbx (const void *__restrict input, void *__restrict output, const Window &win) |
Convert RGB to RGBX. More... | |
void | colorconvert_rgbx_to_rgb (const void *input, void *output, const Window &win) |
Convert RGBX to RGB. More... | |
template<bool yuyv, bool alpha> | |
void | colorconvert_yuyv_to_rgb (const void *__restrict input, void *__restrict output, const Window &win) |
Convert YUYV to RGB. More... | |
template<bool uv, bool alpha> | |
void | colorconvert_nv12_to_rgb (const void *__restrict input, void *__restrict output, const Window &win) |
Convert NV12 to RGB. More... | |
template<bool alpha> | |
void | colorconvert_iyuv_to_rgb (const void *__restrict input, void *__restrict output, const Window &win) |
Convert IYUV to RGB. More... | |
template<bool yuyv> | |
void | colorconvert_yuyv_to_nv12 (const void *__restrict input, void *__restrict output, const Window &win) |
Convert YUYV to NV12. More... | |
void | colorconvert_iyuv_to_nv12 (const void *__restrict input, void *__restrict output, const Window &win) |
Convert IYUV to NV12. More... | |
template<bool uv> | |
void | colorconvert_nv12_to_iyuv (const void *__restrict input, void *__restrict output, const Window &win) |
Convert NV12 to IYUV. More... | |
template<bool yuyv> | |
void | colorconvert_yuyv_to_iyuv (const void *__restrict input, void *__restrict output, const Window &win) |
Convert YUYV to IYUV. More... | |
template<bool uv> | |
void | colorconvert_nv12_to_yuv4 (const void *__restrict input, void *__restrict output, const Window &win) |
Convert NV12 to YUV4. More... | |
void | colorconvert_iyuv_to_yuv4 (const void *__restrict input, void *__restrict output, const Window &win) |
Convert IYUV to YUV4. More... | |
template<bool alpha> | |
void | colorconvert_rgb_to_nv12 (const void *__restrict input, void *__restrict output, const Window &win) |
Convert RGB to NV12. More... | |
template<bool alpha> | |
void | colorconvert_rgb_to_iyuv (const void *__restrict input, void *__restrict output, const Window &win) |
Convert RGB to IYUV. More... | |
template<bool alpha> | |
void | colorconvert_rgb_to_yuv4 (const void *__restrict input, void *__restrict output, const Window &win) |
Convert RGB to YUV4. More... | |
qint8x8_t | vget_low_qs8 (qint8x16_t a) |
Get the lower half of a 16 elements vector. More... | |
qint16x4_t | vget_low_qs16 (qint16x8_t a) |
Get the lower half of a 16 elements vector. More... | |
qint8x8_t | vget_high_qs8 (qint8x16_t a) |
Get the higher half of a 16 elements vector. More... | |
qint16x4_t | vget_high_qs16 (qint16x8_t a) |
Get the higher half of a 16 elements vector. More... | |
qint8x8_t | vld1_qs8 (const qint8_t *addr) |
Load a single 8 bit fixed point vector from memory (8 elements) More... | |
qint16x4_t | vld1_qs16 (const qint16_t *addr) |
Load a single 16 bit fixed point vector from memory (4 elements) More... | |
qint8x16_t | vld1q_qs8 (const qint8_t *addr) |
Load a single 8 bit fixed point vector from memory (16 elements) More... | |
qint16x8_t | vld1q_qs16 (const qint16_t *addr) |
Load a single 16 bit fixed point vector from memory (8 elements) More... | |
qint8x8_t | vld1_dup_qs8 (const qint8_t *addr) |
Load all lanes of 8 bit fixed point vector with same value from memory (8 elements) More... | |
qint16x4_t | vld1_dup_qs16 (const qint16_t *addr) |
Load all lanes of 16 bit fixed point vector with same value from memory (4 elements) More... | |
qint8x16_t | vld1q_dup_qs8 (const qint8_t *addr) |
Load all lanes of 8 bit fixed point vector with same value from memory (16 elements) More... | |
qint16x8_t | vld1q_dup_qs16 (const qint16_t *addr) |
Load all lanes of 16 bit fixed point vector with same value from memory (8 elements) More... | |
qint16x8x2_t | vld2q_qs16 (qint16_t *addr) |
Load two 16 bit fixed point vectors from memory (8x2 elements) More... | |
void | vst1_qs8 (qint8_t *addr, qint8x8_t b) |
Store a single 8 bit fixed point vector to memory (8 elements) More... | |
void | vst1_qs16 (qint16_t *addr, qint16x4_t b) |
Store a single 16 bit fixed point vector to memory (4 elements) More... | |
void | vst1q_qs8 (qint8_t *addr, qint8x16_t b) |
Store a single 8 bit fixed point vector to memory (16 elements) More... | |
void | vst1q_qs16 (qint16_t *addr, qint16x8_t b) |
Store a single 16 bit fixed point vector to memory (8 elements) More... | |
void | vst2q_qs16 (qint16_t *addr, qint16x8x2_t b) |
Store two 16 bit fixed point vector to memory (8x2 elements) More... | |
qint8x8_t | vqmovn_q16 (qint16x8_t a) |
16 bit fixed point vector saturating narrow (8 elements) More... | |
qint16x4_t | vqmovn_q32 (qint32x4_t a) |
32 bit fixed point vector saturating narrow (4 elements) More... | |
qint8x8_t | vdup_n_qs8 (qint8_t a) |
8 bit fixed point vector duplicate (8 elements) More... | |
qint16x4_t | vdup_n_qs16 (qint16_t a) |
16 bit fixed point vector duplicate (4 elements) More... | |
qint8x16_t | vdupq_n_qs8 (qint8_t a) |
8 bit fixed point vector duplicate (16 elements) More... | |
qint8x16_t | vdupq_n_qs8_f32 (float a, int fixed_point_position) |
Duplicate a float and convert it to 8 bit fixed point vector (16 elements) More... | |
qint16x8_t | vdupq_n_qs16_f32 (float a, int fixed_point_position) |
Duplicate a float and convert it to 16 bit fixed point vector (8 elements) More... | |
qint16x8_t | vdupq_n_qs16 (qint16x8_t a) |
16 bit fixed point vector duplicate (8 elements) More... | |
qint8x8_t | vabs_qs8 (qint8x8_t a) |
Absolute value of 8 bit fixed point vector (8 elements) More... | |
qint16x4_t | vabs_qs16 (qint16x4_t a) |
Absolute value of 16 bit fixed point vector (4 elements) More... | |
qint8x16_t | vabsq_qs8 (qint8x16_t a) |
Absolute value of 8 bit fixed point vector (16 elements) More... | |
qint16x8_t | vabsq_qs16 (qint16x8_t a) |
Absolute value of 16 bit fixed point vector (8 elements) More... | |
qint8x8_t | vqabs_qs8 (qint8x8_t a) |
Saturating absolute value of 8 bit fixed point vector (8 elements) More... | |
qint16x4_t | vqabs_qs16 (qint16x4_t a) |
Saturating absolute value of 16 bit fixed point vector (4 elements) More... | |
qint8x16_t | vqabsq_qs8 (qint8x16_t a) |
Saturating absolute value of 8 bit fixed point vector (16 elements) More... | |
qint16x8_t | vqabsq_qs16 (qint16x8_t a) |
Saturating absolute value of 16 bit fixed point vector (8 elements) More... | |
qint8x8_t | vmax_qs8 (qint8x8_t a, qint8x8_t b) |
8 bit fixed point vector max (8 elements) More... | |
qint16x4_t | vmax_qs16 (qint16x4_t a, qint16x4_t b) |
16 bit fixed point vector max (4 elements) More... | |
qint8x16_t | vmaxq_qs8 (qint8x16_t a, qint8x16_t b) |
8 bit fixed point vector max (16 elements) More... | |
qint16x8_t | vmaxq_qs16 (qint16x8_t a, qint16x8_t b) |
16 bit fixed point vector max (8 elements) More... | |
qint8x8_t | vpmax_qs8 (qint8x8_t a, qint8x8_t b) |
8 bit fixed point vector pairwise max (8 elements) More... | |
qint16x4_t | vpmax_qs16 (qint16x4_t a, qint16x4_t b) |
16 bit fixed point vector pairwise max (4 elements) More... | |
qint8x8_t | vmin_qs8 (qint8x8_t a, qint8x8_t b) |
8 bit fixed point vector min (8 elements) More... | |
qint16x4_t | vmin_qs16 (qint16x4_t a, qint16x4_t b) |
16 bit fixed point vector min (4 elements) More... | |
qint8x16_t | vminq_qs8 (qint8x16_t a, qint8x16_t b) |
8 bit fixed point vector min (16 elements) More... | |
qint16x8_t | vminq_qs16 (qint16x8_t a, qint16x8_t b) |
16 bit fixed point vector min (8 elements) More... | |
qint8x8_t | vpmin_qs8 (qint8x8_t a, qint8x8_t b) |
8 bit fixed point vector pairwise min (8 elements) More... | |
qint16x4_t | vpmin_qs16 (qint16x4_t a, qint16x4_t b) |
16 bit fixed point vector pairwise min (4 elements) More... | |
qint8x8_t | vadd_qs8 (qint8x8_t a, qint8x8_t b) |
8 bit fixed point vector add (8 elements) More... | |
qint16x4_t | vadd_qs16 (qint16x4_t a, qint16x4_t b) |
16 bit fixed point vector add (4 elements) More... | |
qint8x16_t | vaddq_qs8 (qint8x16_t a, qint8x16_t b) |
8 bit fixed point vector add (16 elements) More... | |
qint16x8_t | vaddq_qs16 (qint16x8_t a, qint16x8_t b) |
16 bit fixed point vector add (8 elements) More... | |
qint8x8_t | vqadd_qs8 (qint8x8_t a, qint8x8_t b) |
8 bit fixed point vector saturating add (8 elements) More... | |
qint16x4_t | vqadd_qs16 (qint16x4_t a, qint16x4_t b) |
16 bit fixed point vector saturating add (4 elements) More... | |
qint8x16_t | vqaddq_qs8 (qint8x16_t a, qint8x16_t b) |
8 bit fixed point vector saturating add (16 elements) More... | |
qint16x8_t | vqaddq_qs16 (qint16x8_t a, qint16x8_t b) |
16 bit fixed point vector saturating add (8 elements) More... | |
int16x4_t | vpaddl_qs8 (qint8x8_t a) |
8 bit fixed point vector saturating pairwise add (8 elements) More... | |
qint8x8_t | vsub_qs8 (qint8x8_t a, qint8x8_t b) |
8 bit fixed point vector subtraction (8 elements) More... | |
qint16x4_t | vsub_qs16 (qint16x4_t a, qint16x4_t b) |
16 bit fixed point vector subtraction (4 elements) More... | |
qint8x16_t | vsubq_qs8 (qint8x16_t a, qint8x16_t b) |
8 bit fixed point vector subtraction (16 elements) More... | |
qint16x8_t | vsubq_qs16 (qint16x8_t a, qint16x8_t b) |
16 bit fixed point vector subtraction (8 elements) More... | |
qint8x8_t | vqsub_qs8 (qint8x8_t a, qint8x8_t b) |
8 bit fixed point vector saturating subtraction (8 elements) More... | |
qint16x4_t | vqsub_qs16 (qint16x4_t a, qint16x4_t b) |
16 bit fixed point vector saturating subtraction (4 elements) More... | |
qint8x16_t | vqsubq_qs8 (qint8x16_t a, qint8x16_t b) |
8 bit fixed point vector saturating subtraction (16 elements) More... | |
qint16x8_t | vqsubq_qs16 (qint16x8_t a, qint16x8_t b) |
16 bit fixed point vector saturating subtraction (8 elements) More... | |
qint8x8_t | vmul_qs8 (qint8x8_t a, qint8x8_t b, int fixed_point_position) |
8 bit fixed point vector multiply (8 elements) More... | |
qint16x4_t | vmul_qs16 (qint16x4_t a, qint16x4_t b, int fixed_point_position) |
16 bit fixed point vector multiply (4 elements) More... | |
qint8x16_t | vmulq_qs8 (qint8x16_t a, qint8x16_t b, int fixed_point_position) |
8 bit fixed point vector multiply (16 elements) More... | |
qint16x8_t | vmulq_qs16 (qint16x8_t a, qint16x8_t b, int fixed_point_position) |
16 bit fixed point vector multiply (8 elements) More... | |
qint8x8_t | vqmul_qs8 (qint8x8_t a, qint8x8_t b, int fixed_point_position) |
8 bit fixed point vector saturating multiply (8 elements) More... | |
qint16x4_t | vqmul_qs16 (qint16x4_t a, qint16x4_t b, int fixed_point_position) |
16 bit fixed point vector saturating multiply (4 elements) More... | |
qint8x16_t | vqmulq_qs8 (qint8x16_t a, qint8x16_t b, int fixed_point_position) |
8 bit fixed point vector saturating multiply (16 elements) More... | |
qint16x8_t | vqmulq_qs16 (qint16x8_t a, qint16x8_t b, int fixed_point_position) |
16 bit fixed point vector saturating multiply (8 elements) More... | |
qint16x8_t | vmull_qs8 (qint8x8_t a, qint8x8_t b, int fixed_point_position) |
8 bit fixed point vector long multiply (8 elements) More... | |
qint32x4_t | vmull_qs16 (qint16x4_t a, qint16x4_t b, int fixed_point_position) |
16 bit fixed point vector long multiply (4 elements) More... | |
qint8x8_t | vmla_qs8 (qint8x8_t a, qint8x8_t b, qint8x8_t c, int fixed_point_position) |
8 bit fixed point vector multiply-accumulate (8 elements). More... | |
qint16x4_t | vmla_qs16 (qint16x4_t a, qint16x4_t b, qint16x4_t c, int fixed_point_position) |
16 bit fixed point vector multiply-accumulate (4 elements). More... | |
qint8x16_t | vmlaq_qs8 (qint8x16_t a, qint8x16_t b, qint8x16_t c, int fixed_point_position) |
8 bit fixed point vector multiply-accumulate (16 elements). More... | |
qint16x8_t | vmlaq_qs16 (qint16x8_t a, qint16x8_t b, qint16x8_t c, int fixed_point_position) |
16 bit fixed point vector multiply-accumulate (16 elements). More... | |
qint8x8_t | vqmla_qs8 (qint8x8_t a, qint8x8_t b, qint8x8_t c, int fixed_point_position) |
8 bit fixed point vector saturating multiply-accumulate (8 elements). More... | |
qint16x4_t | vqmla_qs16 (qint16x4_t a, qint16x4_t b, qint16x4_t c, int fixed_point_position) |
16 bit fixed point vector saturating multiply-accumulate (4 elements). More... | |
qint8x16_t | vqmlaq_qs8 (qint8x16_t a, qint8x16_t b, qint8x16_t c, int fixed_point_position) |
8 bit fixed point vector saturating multiply-accumulate (16 elements). More... | |
qint16x8_t | vqmlaq_qs16 (qint16x8_t a, qint16x8_t b, qint16x8_t c, int fixed_point_position) |
16 bit fixed point vector saturating multiply-accumulate (8 elements). More... | |
qint16x8_t | vmlal_qs8 (qint16x8_t a, qint8x8_t b, qint8x8_t c, int fixed_point_position) |
8 bit fixed point vector multiply-accumulate long (8 elements). More... | |
qint32x4_t | vmlal_qs16 (qint32x4_t a, qint16x4_t b, qint16x4_t c, int fixed_point_position) |
16 bit fixed point vector multiply-accumulate long (4 elements). More... | |
qint16x8_t | vqmlal_qs8 (qint16x8_t a, qint8x8_t b, qint8x8_t c, int fixed_point_position) |
8 bit fixed point vector saturating multiply-accumulate long (8 elements). More... | |
qint32x4_t | vqmlal_qs16 (qint32x4_t a, qint16x4_t b, qint16x4_t c, int fixed_point_position) |
16 bit fixed point vector saturating multiply-accumulate long (4 elements). More... | |
qint8x8_t | vqcvt_qs8_f32 (const float32x4x2_t a, int fixed_point_position) |
Convert a float vector with 4x2 elements to 8 bit fixed point vector with 8 elements. More... | |
qint16x4_t | vqcvt_qs16_f32 (const float32x4_t a, int fixed_point_position) |
Convert a float vector with 4 elements to 16 bit fixed point vector with 4 elements. More... | |
qint8x16_t | vqcvtq_qs8_f32 (const float32x4x4_t &a, int fixed_point_position) |
Convert a float vector with 4x4 elements to 8 bit fixed point vector with 16 elements. More... | |
qint16x8_t | vqcvtq_qs16_f32 (const float32x4x2_t &a, int fixed_point_position) |
Convert a float vector with 4x2 elements to 16 bit fixed point vector with 8 elements. More... | |
float32x4x2_t | vcvt_f32_qs8 (qint8x8_t a, int fixed_point_position) |
Convert a 8 bit fixed point vector with 8 elements to a float vector with 4x2 elements. More... | |
float32x4_t | vcvt_f32_qs16 (qint16x4_t a, int fixed_point_position) |
Convert a 16 bit fixed point vector with 4 elements to a float vector with 4 elements. More... | |
float32x4x4_t | vcvtq_qs8_f32 (qint8x16_t a, int fixed_point_position) |
Convert a 8 bit fixed point vector with 16 elements to a float vector with 4x4 elements. More... | |
float32x4x2_t | vcvtq_qs16_f32 (qint16x8_t a, int fixed_point_position) |
Convert a 16 bit fixed point vector with 8 elements to a float vector with 4x2 elements. More... | |
qint8x8_t | vrecip_qs8 (qint8x8_t a, int fixed_point_position) |
Calculate reciprocal of a fixed point 8bit number using the Newton-Raphson method. More... | |
qint16x4_t | vrecip_qs16 (qint16x4_t a, int fixed_point_position) |
Calculate reciprocal of a fixed point 8bit number using the Newton-Raphson method. More... | |
qint8x16_t | vrecipq_qs8 (qint8x16_t a, int fixed_point_position) |
Calculate reciprocal of a fixed point 8bit number using the Newton-Raphson method. More... | |
qint16x8_t | vrecipq_qs16 (qint16x8_t a, int fixed_point_position) |
Calculate reciprocal of a fixed point 8bit number using the Newton-Raphson method. More... | |
qint8x8_t | vdiv_qs8 (qint8x8_t a, int8x8_t b, int fixed_point_position) |
Division fixed point 8bit (8 elements) More... | |
qint16x4_t | vdiv_qs16 (qint16x4_t a, qint16x4_t b, int fixed_point_position) |
Division fixed point 16 bit (4 elements) More... | |
qint8x16_t | vdivq_qs8 (qint8x16_t a, qint8x16_t b, int fixed_point_position) |
Division fixed point 8bit (16 elements) More... | |
qint16x8_t | vdivq_qs16 (qint16x8_t a, qint16x8_t b, int fixed_point_position) |
Division fixed point 16 bit (8 elements) More... | |
template<bool islog> | |
qint8x8_t | vtaylor_poly_qs8 (qint8x8_t a, int fixed_point_position) |
Perform a 4th degree polynomial approximation. More... | |
template<bool islog> | |
qint16x4_t | vtaylor_poly_qs16 (qint16x4_t a, int fixed_point_position) |
Perform a 4th degree polynomial approximation. More... | |
template<bool islog> | |
qint8x16_t | vtaylor_polyq_qs8 (qint8x16_t a, int fixed_point_position) |
Perform a 4th degree polynomial approximation. More... | |
template<bool islog> | |
qint16x8_t | vtaylor_polyq_qs16 (qint16x8_t a, int fixed_point_position) |
Perform a 4th degree polynomial approximation. More... | |
qint8x8_t | vqexp_qs8 (qint8x8_t a, int fixed_point_position) |
Calculate saturating exponential fixed point 8bit (8 elements) More... | |
qint16x4_t | vqexp_qs16 (qint16x4_t a, int fixed_point_position) |
Calculate saturating exponential fixed point 16 bit (4 elements) More... | |
qint8x16_t | vqexpq_qs8 (qint8x16_t a, int fixed_point_position) |
Calculate saturating exponential fixed point 8bit (16 elements) More... | |
qint16x8_t | vqexpq_qs16 (qint16x8_t a, int fixed_point_position) |
Calculate saturating exponential fixed point 16 bit (8 elements) More... | |
qint8x8_t | vlog_qs8 (qint8x8_t a, int fixed_point_position) |
Calculate logarithm fixed point 8 bit (8 elements) More... | |
qint16x4_t | vlog_qs16 (qint16x4_t a, int fixed_point_position) |
Calculate logarithm fixed point 16 bit (4 elements) More... | |
qint8x16_t | vlogq_qs8 (qint8x16_t a, int fixed_point_position) |
Calculate logarithm fixed point 16bit (16 elements) More... | |
qint16x8_t | vlogq_qs16 (qint16x8_t a, int fixed_point_position) |
Calculate logarithm fixed point 16 bit (8 elements) More... | |
qint8x8_t | vinvsqrt_qs8 (qint8x8_t a, int fixed_point_position) |
Calculate inverse square root for fixed point 8bit using Newton-Raphosn method (8 elements) More... | |
qint16x4_t | vinvsqrt_qs16 (qint16x4_t a, int fixed_point_position) |
Calculate inverse square root for fixed point 16 bit using Newton-Raphosn method (4 elements) More... | |
qint8x8_t | vqinvsqrt_qs8 (qint8x8_t a, int fixed_point_position) |
Calculate saturating inverse square root for fixed point 8bit using Newton-Raphosn method (8 elements) More... | |
qint16x4_t | vqinvsqrt_qs16 (qint16x4_t a, int fixed_point_position) |
Calculate saturating inverse square root for fixed point 16 bit using Newton-Raphosn method (4 elements) More... | |
qint8x16_t | vinvsqrtq_qs8 (qint8x16_t a, int fixed_point_position) |
Calculate inverse square root for fixed point 8bit using Newton-Raphosn method (16 elements) More... | |
qint16x8_t | vinvsqrtq_qs16 (qint16x8_t a, int fixed_point_position) |
Calculate inverse square root for fixed point 8bit using Newton-Raphosn method (8 elements) More... | |
qint8x16_t | vqinvsqrtq_qs8 (qint8x16_t a, int fixed_point_position) |
Calculate saturating inverse square root for fixed point 8bit using Newton-Raphosn method (16 elements) More... | |
qint16x8_t | vqinvsqrtq_qs16 (qint16x8_t a, int fixed_point_position) |
Calculate saturating inverse square root for fixed point 16 bit using Newton-Raphosn method (8 elements) More... | |
qint8x8_t | vqtanh_qs8 (qint8x8_t a, int fixed_point_position) |
Calculate hyperbolic tangent for fixed point 8bit (8 elements) More... | |
qint16x4_t | vqtanh_qs16 (qint16x4_t a, int fixed_point_position) |
Calculate hyperbolic tangent for fixed point 16 bit (4 elements) More... | |
qint8x16_t | vqtanhq_qs8 (qint8x16_t a, int fixed_point_position) |
Calculate hyperbolic tangent for fixed point 8bit (16 elements) More... | |
qint16x8_t | vqtanhq_qs16 (qint16x8_t a, int fixed_point_position) |
Calculate hyperbolic tangent for fixed point 16bit (8 elements) More... | |
qint8x16_t | vqpowq_qs8 (qint8x16_t a, qint8x16_t b, int fixed_point_position) |
Calculate saturating n power for fixed point 8bit (16 elements). More... | |
qint16x8_t | vqpowq_qs16 (qint16x8_t a, qint16x8_t b, int fixed_point_position) |
Calculate saturating n power for fixed point 16bit (8 elements). More... | |
float32x4x2_t | vmax2q_f32 (float32x4x2_t a, float32x4x2_t b) |
Compute lane-by-lane maximum between elements of a float vector with 4x2 elements. More... | |
float32x4_t | vfloorq_f32 (float32x4_t val) |
Calculate floor of a vector. More... | |
float32x2_t | vinvsqrt_f32 (float32x2_t x) |
Calculate inverse square root. More... | |
float32x4_t | vinvsqrtq_f32 (float32x4_t x) |
Calculate inverse square root. More... | |
float32x2_t | vinv_f32 (float32x2_t x) |
Calculate reciprocal. More... | |
float32x4_t | vinvq_f32 (float32x4_t x) |
Calculate reciprocal. More... | |
float32x4_t | vtaylor_polyq_f32 (float32x4_t x, const std::array< float32x4_t, 8 > &coeffs) |
Perform a 7th degree polynomial approximation using Estrin's method. More... | |
float32x4_t | vexpq_f32 (float32x4_t x) |
Calculate exponential. More... | |
float32x4_t | vlogq_f32 (float32x4_t x) |
Calculate logarithm. More... | |
float32x4_t | vtanhq_f32 (float32x4_t val) |
Calculate hyperbolic tangent. More... | |
float32x4_t | vpowq_f32 (float32x4_t val, float32x4_t n) |
Calculate n power of a number. More... | |
int | round (float x, RoundingPolicy rounding_policy) |
Return a rounded value of x. More... | |
template<typename S , typename T > | |
constexpr auto | DIV_CEIL (S val, T m) -> decltype((val+m-1)/m) |
Calculate the rounded up quotient of val / m. More... | |
template<typename S , typename T > | |
auto | ceil_to_multiple (S value, T divisor) -> decltype(((value+divisor-1)/divisor)*divisor) |
Computes the smallest number larger or equal to value that is a multiple of divisor. More... | |
template<typename S , typename T > | |
auto | floor_to_multiple (S value, T divisor) -> decltype((value/divisor)*divisor) |
Computes the largest number smaller or equal to value that is a multiple of divisor. More... | |
std::string | build_information () |
Returns the arm_compute library build information. More... | |
std::string | read_file (const std::string &filename, bool binary) |
Load an entire file in memory. More... | |
size_t | data_size_from_type (DataType data_type) |
The size in bytes of the data type. More... | |
size_t | pixel_size_from_format (Format format) |
The size in bytes of the pixel format. More... | |
size_t | element_size_from_data_type (DataType dt) |
The size in bytes of the data type. More... | |
DataType | data_type_from_format (Format format) |
Return the data type used by a given single-planar pixel format. More... | |
int | plane_idx_from_channel (Format format, Channel channel) |
Return the plane index of a given channel given an input format. More... | |
int | channel_idx_from_format (Format format, Channel channel) |
Return the channel index of a given channel given an input format. More... | |
size_t | num_planes_from_format (Format format) |
Return the number of planes for a given format. More... | |
size_t | num_channels_from_format (Format format) |
Return the number of channels for a given single-planar pixel format. More... | |
DataType | get_promoted_data_type (DataType dt) |
Return the promoted data type of a given data type. More... | |
bool | has_format_horizontal_subsampling (Format format) |
Return true if the given format has horizontal subsampling. More... | |
bool | has_format_vertical_subsampling (Format format) |
Return true if the given format has vertical subsampling. More... | |
bool | separate_matrix (const int16_t *conv, int16_t *conv_col, int16_t *conv_row, uint8_t size) |
Separate a 2D convolution into two 1D convolutions. More... | |
uint32_t | calculate_matrix_scale (const int16_t *matrix, unsigned int matrix_size) |
Calculate the scale of the given square matrix. More... | |
template<typename T > | |
TensorShape | calculate_depth_concatenate_shape (const std::vector< T * > &inputs_vector) |
Calculate the output shapes of the depth concatenate function. More... | |
TensorShape | adjust_odd_shape (const TensorShape &shape, Format format) |
Adjust tensor shape size if width or height are odd for a given multi-planar format. More... | |
TensorShape | calculate_subsampled_shape (const TensorShape &shape, Format format, Channel channel=Channel::UNKNOWN) |
Calculate subsampled shape for a given format and channel. More... | |
std::pair< DataType, DataType > | data_type_for_convolution (const int16_t *conv_col, const int16_t *conv_row, size_t size) |
Calculate accurary required by the horizontal and vertical convolution computations. More... | |
DataType | data_type_for_convolution_matrix (const int16_t *conv, size_t size) |
Calculate the accuracy required by the squared convolution calculation. More... | |
PadStrideInfo | calculate_same_pad (TensorShape input_shape, TensorShape weights_shape, PadStrideInfo conv_info) |
Calculate padding requirements in case of SAME padding. More... | |
TensorShape | deconvolution_output_shape (const std::pair< unsigned int, unsigned int > &out_dims, TensorShape input, TensorShape weights) |
Returns expected shape for the deconvolution output tensor. More... | |
const std::pair< unsigned int, unsigned int > | deconvolution_output_dimensions (unsigned int in_width, unsigned int in_height, unsigned int kernel_width, unsigned int kernel_height, unsigned int padx, unsigned int pady, unsigned int inner_border_right, unsigned int inner_border_top, unsigned int stride_x, unsigned int stride_y) |
Returns expected width and height of the deconvolution's output tensor. More... | |
const std::pair< unsigned int, unsigned int > | scaled_dimensions (unsigned int width, unsigned int height, unsigned int kernel_width, unsigned int kernel_height, const PadStrideInfo &pad_stride_info, const Size2D &dilation=Size2D(1U, 1U)) |
Returns expected width and height of output scaled tensor depending on dimensions rounding mode. More... | |
const std::string & | string_from_format (Format format) |
Convert a tensor format into a string. More... | |
const std::string & | string_from_channel (Channel channel) |
Convert a channel identity into a string. More... | |
const std::string & | string_from_data_layout (DataLayout dl) |
Convert a data layout identity into a string. More... | |
const std::string & | string_from_data_type (DataType dt) |
Convert a data type identity into a string. More... | |
const std::string & | string_from_matrix_pattern (MatrixPattern pattern) |
Convert a matrix pattern into a string. More... | |
const std::string & | string_from_activation_func (ActivationLayerInfo::ActivationFunction act) |
Translates a given activation function to a string. More... | |
const std::string & | string_from_non_linear_filter_function (NonLinearFilterFunction function) |
Translates a given non linear function to a string. More... | |
const std::string & | string_from_interpolation_policy (InterpolationPolicy policy) |
Translates a given interpolation policy to a string. More... | |
const std::string & | string_from_border_mode (BorderMode border_mode) |
Translates a given border mode policy to a string. More... | |
const std::string & | string_from_norm_type (NormType type) |
Translates a given normalization type to a string. More... | |
const std::string & | string_from_pooling_type (PoolingType type) |
Translates a given pooling type to a string. More... | |
std::string | lower_string (const std::string &val) |
Lower a given string. More... | |
bool | is_data_type_float (DataType dt) |
Check if a given data type is of floating point type. More... | |
bool | is_data_type_quantized (DataType dt) |
Check if a given data type is of quantized type. More... | |
bool | is_data_type_fixed_point (DataType dt) |
Check if a given data type is of fixed point type. More... | |
bool | is_data_type_quantized_asymmetric (DataType dt) |
Check if a given data type is of asymmetric quantized type. More... | |
std::string | float_to_string_with_full_precision (float val) |
Create a string with the float in full precision. More... | |
template<typename T > | |
void | print_consecutive_elements_impl (std::ostream &s, const T *ptr, unsigned int n, int stream_width=0, const std::string &element_delim=" ") |
Print consecutive elements to an output stream. More... | |
template<typename T > | |
int | max_consecutive_elements_display_width_impl (std::ostream &s, const T *ptr, unsigned int n) |
Identify the maximum width of n consecutive elements. More... | |
void | print_consecutive_elements (std::ostream &s, DataType dt, const uint8_t *ptr, unsigned int n, int stream_width, const std::string &element_delim=" ") |
Print consecutive elements to an output stream. More... | |
int | max_consecutive_elements_display_width (std::ostream &s, DataType dt, const uint8_t *ptr, unsigned int n) |
Identify the maximum width of n consecutive elements. More... | |
template<typename... Ts> | |
arm_compute::Status | error_on_nullptr (const char *function, const char *file, const int line, Ts &&...pointers) |
Create an error if one of the pointers is a nullptr. More... | |
arm_compute::Status | error_on_mismatching_windows (const char *function, const char *file, const int line, const Window &full, const Window &win) |
Return an error if the passed window is invalid. More... | |
arm_compute::Status | error_on_invalid_subwindow (const char *function, const char *file, const int line, const Window &full, const Window &sub) |
Return an error if the passed subwindow is invalid. More... | |
arm_compute::Status | error_on_window_not_collapsable_at_dimension (const char *function, const char *file, const int line, const Window &full, const Window &window, const int dim) |
Return an error if the window can't be collapsed at the given dimension. More... | |
arm_compute::Status | error_on_coordinates_dimensions_gte (const char *function, const char *file, const int line, const Coordinates &pos, unsigned int max_dim) |
Return an error if the passed coordinates have too many dimensions. More... | |
arm_compute::Status | error_on_window_dimensions_gte (const char *function, const char *file, const int line, const Window &win, unsigned int max_dim) |
Return an error if the passed window has too many dimensions. More... | |
template<typename T , typename... Ts> | |
arm_compute::Status | error_on_mismatching_dimensions (const char *function, const char *file, int line, const Dimensions< T > &dim1, const Dimensions< T > &dim2, Ts &&...dims) |
Return an error if the passed dimension objects differ. More... | |
template<typename... Ts> | |
arm_compute::Status | error_on_tensors_not_even (const char *function, const char *file, int line, const Format &format, const ITensor *tensor1, Ts...tensors) |
Return an error if the passed tensor objects are not even. More... | |
template<typename... Ts> | |
arm_compute::Status | error_on_tensors_not_subsampled (const char *function, const char *file, int line, const Format &format, const TensorShape &shape, const ITensor *tensor1, Ts...tensors) |
Return an error if the passed tensor objects are not sub-sampled. More... | |
template<typename... Ts> | |
arm_compute::Status | error_on_mismatching_shapes (const char *function, const char *file, const int line, const ITensorInfo *tensor_info_1, const ITensorInfo *tensor_info_2, Ts...tensor_infos) |
Return an error if the passed two tensor infos have different shapes from the given dimension. More... | |
template<typename... Ts> | |
arm_compute::Status | error_on_mismatching_shapes (const char *function, const char *file, const int line, const ITensor *tensor_1, const ITensor *tensor_2, Ts...tensors) |
Return an error if the passed two tensors have different shapes from the given dimension. More... | |
template<typename... Ts> | |
arm_compute::Status | error_on_mismatching_shapes (const char *function, const char *file, const int line, unsigned int upper_dim, const ITensorInfo *tensor_info_1, const ITensorInfo *tensor_info_2, Ts...tensor_infos) |
Return an error if the passed two tensors have different shapes from the given dimension. More... | |
template<typename... Ts> | |
arm_compute::Status | error_on_mismatching_shapes (const char *function, const char *file, const int line, unsigned int upper_dim, const ITensor *tensor_1, const ITensor *tensor_2, Ts...tensors) |
Return an error if the passed two tensors have different shapes from the given dimension. More... | |
template<typename... Ts> | |
arm_compute::Status | error_on_mismatching_data_layouts (const char *function, const char *file, const int line, const ITensorInfo *tensor_info, Ts...tensor_infos) |
Return an error if the passed tensor infos have different data layouts. More... | |
template<typename... Ts> | |
arm_compute::Status | error_on_mismatching_data_layouts (const char *function, const char *file, const int line, const ITensor *tensor, Ts...tensors) |
Return an error if the passed tensors have different data layouts. More... | |
template<typename... Ts> | |
arm_compute::Status | error_on_mismatching_data_types (const char *function, const char *file, const int line, const ITensorInfo *tensor_info, Ts...tensor_infos) |
Return an error if the passed two tensor infos have different data types. More... | |
template<typename... Ts> | |
arm_compute::Status | error_on_mismatching_data_types (const char *function, const char *file, const int line, const ITensor *tensor, Ts...tensors) |
Return an error if the passed two tensors have different data types. More... | |
template<typename... Ts> | |
arm_compute::Status | error_on_mismatching_fixed_point (const char *function, const char *file, const int line, const ITensorInfo *tensor_info_1, const ITensorInfo *tensor_info_2, Ts...tensor_infos) |
Return an error if the passed tensor infos have different fixed point data types or different fixed point positions. More... | |
template<typename... Ts> | |
arm_compute::Status | error_on_mismatching_fixed_point (const char *function, const char *file, const int line, const ITensor *tensor_1, const ITensor *tensor_2, Ts...tensors) |
Return an error if the passed tensor have different fixed point data types or different fixed point positions. More... | |
template<typename... Ts> | |
arm_compute::Status | error_on_mismatching_quantization_info (const char *function, const char *file, const int line, const ITensorInfo *tensor_info_1, const ITensorInfo *tensor_info_2, Ts...tensor_infos) |
Return an error if the passed tensor infos have different asymmetric quantized data types or different quantization info. More... | |
template<typename... Ts> | |
arm_compute::Status | error_on_mismatching_quantization_info (const char *function, const char *file, const int line, const ITensor *tensor_1, const ITensor *tensor_2, Ts...tensors) |
Return an error if the passed tensor have different asymmetric quantized data types or different quantization info. More... | |
template<typename T , typename F , typename... Fs> | |
void | error_on_format_not_in (const char *function, const char *file, const int line, const T *object, F &&format, Fs &&...formats) |
Throw an error if the format of the passed tensor/multi-image does not match any of the formats provided. More... | |
template<typename T , typename... Ts> | |
arm_compute::Status | error_on_data_type_not_in (const char *function, const char *file, const int line, const ITensorInfo *tensor_info, T &&dt, Ts &&...dts) |
Return an error if the data type of the passed tensor info does not match any of the data types provided. More... | |
template<typename T , typename... Ts> | |
arm_compute::Status | error_on_data_type_not_in (const char *function, const char *file, const int line, const ITensor *tensor, T &&dt, Ts &&...dts) |
Return an error if the data type of the passed tensor does not match any of the data types provided. More... | |
template<typename T , typename... Ts> | |
arm_compute::Status | error_on_data_type_channel_not_in (const char *function, const char *file, const int line, const ITensorInfo *tensor_info, size_t num_channels, T &&dt, Ts &&...dts) |
Return an error if the data type or the number of channels of the passed tensor info does not match any of the data types and number of channels provided. More... | |
template<typename T , typename... Ts> | |
arm_compute::Status | error_on_data_type_channel_not_in (const char *function, const char *file, const int line, const ITensor *tensor, size_t num_channels, T &&dt, Ts &&...dts) |
Return an error if the data type or the number of channels of the passed tensor does not match any of the data types and number of channels provided. More... | |
arm_compute::Status | error_on_tensor_not_2d (const char *function, const char *file, const int line, const ITensor *tensor) |
Return an error if the tensor is not 2D. More... | |
template<typename T , typename... Ts> | |
arm_compute::Status | error_on_channel_not_in (const char *function, const char *file, const int line, T cn, T &&channel, Ts &&...channels) |
Return an error if the channel is not in channels. More... | |
arm_compute::Status | error_on_channel_not_in_known_format (const char *function, const char *file, const int line, Format fmt, Channel cn) |
Return an error if the channel is not in format. More... | |
arm_compute::Status | error_on_invalid_multi_hog (const char *function, const char *file, const int line, const IMultiHOG *multi_hog) |
Return an error if the IMultiHOG container is invalid. More... | |
arm_compute::Status | error_on_unconfigured_kernel (const char *function, const char *file, const int line, const IKernel *kernel) |
Return an error if the kernel is not configured. More... | |
arm_compute::Status | error_on_invalid_subtensor (const char *function, const char *file, const int line, const TensorShape &parent_shape, const Coordinates &coords, const TensorShape &shape) |
Return an error if if the coordinates and shape of the subtensor are within the parent tensor. More... | |
arm_compute::Status | error_on_invalid_subtensor_valid_region (const char *function, const char *file, const int line, const ValidRegion &parent_valid_region, const ValidRegion &valid_region) |
Return an error if the valid region of a subtensor is not inside the valid region of the parent tensor. More... | |
template<typename... Ts> | |
arm_compute::Status | error_on_mismatching_fixed_point_position (const char *function, const char *file, const int line, const ITensorInfo *tensor_info_1, const ITensorInfo *tensor_info_2, Ts...tensor_infos) |
Return an error if the input fixed-point positions are different. More... | |
template<typename... Ts> | |
arm_compute::Status | error_on_mismatching_fixed_point_position (const char *function, const char *file, const int line, const ITensor *tensor_1, const ITensor *tensor_2, Ts...tensors) |
Return an error if the input fixed-point positions are different. More... | |
arm_compute::Status | error_on_value_not_representable_in_fixed_point (const char *function, const char *file, int line, float value, const ITensorInfo *tensor_info) |
Return an error if the fixed-point value is not representable in the specified Q format. More... | |
arm_compute::Status | error_on_value_not_representable_in_fixed_point (const char *function, const char *file, int line, float value, const ITensor *tensor) |
Return an error an error if the fixed-point value is not representable in the specified Q format. More... | |
void | get_cpu_configuration (CPUInfo &cpuinfo) |
This function will try to detect the CPU configuration on the system and will fill the cpuinfo object accordingly to reflect this. More... | |
unsigned int | get_threads_hint () |
Some systems have both big and small cores, this fuction computes the minimum number of cores that are exactly the same on the system. More... | |
void | allocate_workspace (size_t workspace_size, Tensor &workspace, MemoryGroup *memory_group, size_t alignment, unsigned int num_threads) |
Allocate a workspace tensor. More... | |
template<typename T > | |
bool | setup_assembly_kernel (const ITensor *a, const ITensor *b, ITensor *d, float alpha, float beta, bool pretranspose_hint, Tensor &workspace, Tensor &B_pretranspose, MemoryGroup &memory_group, T &asm_glue) |
Create a wrapper kernel. More... | |
const std::string & | string_from_scheduler_type (Scheduler::Type t) |
Convert a Scheduler::Type into a string. More... | |
inline::std::istream & | operator>> (::std::istream &is, BorderMode &mode) |
Formatted input of the BorderMode type. More... | |
template<typename T > | |
inline::std::ostream & | operator<< (::std::ostream &os, const Dimensions< T > &dimensions) |
Formatted output of the Dimensions type. More... | |
inline::std::ostream & | operator<< (::std::ostream &os, const NonLinearFilterFunction &function) |
Formatted output of the NonLinearFilterFunction type. More... | |
std::string | to_string (const NonLinearFilterFunction &function) |
Formatted output of the NonLinearFilterFunction type. More... | |
inline::std::ostream & | operator<< (::std::ostream &os, const MatrixPattern &pattern) |
Formatted output of the MatrixPattern type. More... | |
std::string | to_string (const MatrixPattern &pattern) |
Formatted output of the MatrixPattern type. More... | |
inline::std::ostream & | operator<< (::std::ostream &os, const RoundingPolicy &rounding_policy) |
Formatted output of the RoundingPolicy type. More... | |
inline::std::ostream & | operator<< (::std::ostream &os, const WeightsInfo &weights_info) |
Formatted output of the WeightsInfo type. More... | |
inline::std::ostream & | operator<< (::std::ostream &os, const ROIPoolingLayerInfo &pool_info) |
Formatted output of the ROIPoolingInfo type. More... | |
inline::std::ostream & | operator<< (::std::ostream &os, const QuantizationInfo &quantization_info) |
Formatted output of the QuantizationInfo type. More... | |
std::string | to_string (const QuantizationInfo &quantization_info) |
Formatted output of the QuantizationInfo type. More... | |
inline::std::ostream & | operator<< (::std::ostream &os, const FixedPointOp &op) |
Formatted output of the FixedPointOp type. More... | |
std::string | to_string (const FixedPointOp &op) |
Formatted output of the FixedPointOp type. More... | |
inline::std::ostream & | operator<< (::std::ostream &os, const ActivationLayerInfo::ActivationFunction &act_function) |
Formatted output of the activation function type. More... | |
std::string | to_string (const arm_compute::ActivationLayerInfo &info) |
Formatted output of the activation function info type. More... | |
std::string | to_string (const arm_compute::ActivationLayerInfo::ActivationFunction &function) |
Formatted output of the activation function type. More... | |
inline::std::ostream & | operator<< (::std::ostream &os, const NormType &norm_type) |
Formatted output of the NormType type. More... | |
std::string | to_string (const arm_compute::NormalizationLayerInfo &info) |
Formatted output of NormalizationLayerInfo. More... | |
inline::std::ostream & | operator<< (::std::ostream &os, const NormalizationLayerInfo &info) |
Formatted output of NormalizationLayerInfo. More... | |
inline::std::ostream & | operator<< (::std::ostream &os, const PoolingType &pool_type) |
Formatted output of the PoolingType type. More... | |
inline::std::ostream & | operator<< (::std::ostream &os, const PoolingLayerInfo &info) |
Formatted output of PoolingLayerInfo. More... | |
std::string | to_string (const RoundingPolicy &rounding_policy) |
Formatted output of RoundingPolicy. More... | |
inline::std::ostream & | operator<< (::std::ostream &os, const DataLayout &data_layout) |
Formatted output of the DataLayout type. More... | |
std::string | to_string (const arm_compute::DataLayout &data_layout) |
Formatted output of the DataLayout type. More... | |
inline::std::ostream & | operator<< (::std::ostream &os, const DataType &data_type) |
Formatted output of the DataType type. More... | |
std::string | to_string (const arm_compute::DataType &data_type) |
Formatted output of the DataType type. More... | |
inline::std::ostream & | operator<< (::std::ostream &os, const Format &format) |
Formatted output of the Format type. More... | |
std::string | to_string (const Format &format) |
Formatted output of the Format type. More... | |
inline::std::ostream & | operator<< (::std::ostream &os, const Channel &channel) |
Formatted output of the Channel type. More... | |
std::string | to_string (const Channel &channel) |
Formatted output of the Channel type. More... | |
inline::std::ostream & | operator<< (::std::ostream &os, const BorderMode &mode) |
Formatted output of the BorderMode type. More... | |
inline::std::ostream & | operator<< (::std::ostream &os, const BorderSize &border) |
Formatted output of the BorderSize type. More... | |
inline::std::ostream & | operator<< (::std::ostream &os, const InterpolationPolicy &policy) |
Formatted output of the InterpolationPolicy type. More... | |
inline::std::ostream & | operator<< (::std::ostream &os, const SamplingPolicy &policy) |
Formatted output of the SamplingPolicy type. More... | |
std::string | to_string (const TensorInfo &info) |
Formatted output of the TensorInfo type. More... | |
template<typename T > | |
std::string | to_string (const Dimensions< T > &dimensions) |
Formatted output of the Dimensions type. More... | |
std::string | to_string (const Strides &stride) |
Formatted output of the Strides type. More... | |
std::string | to_string (const TensorShape &shape) |
Formatted output of the TensorShape type. More... | |
std::string | to_string (const Coordinates &coord) |
Formatted output of the Coordinates type. More... | |
inline::std::ostream & | operator<< (::std::ostream &os, const Rectangle &rect) |
Formatted output of the Rectangle type. More... | |
inline::std::ostream & | operator<< (::std::ostream &os, const PadStrideInfo &pad_stride_info) |
Formatted output of the PadStrideInfo type. More... | |
std::string | to_string (const PadStrideInfo &pad_stride_info) |
Formatted output of the PadStrideInfo type. More... | |
std::string | to_string (const BorderMode &mode) |
Formatted output of the BorderMode type. More... | |
std::string | to_string (const BorderSize &border) |
Formatted output of the BorderSize type. More... | |
std::string | to_string (const InterpolationPolicy &policy) |
Formatted output of the InterpolationPolicy type. More... | |
std::string | to_string (const SamplingPolicy &policy) |
Formatted output of the SamplingPolicy type. More... | |
inline::std::ostream & | operator<< (::std::ostream &os, const ConvertPolicy &policy) |
Formatted output of the ConvertPolicy type. More... | |
std::string | to_string (const ConvertPolicy &policy) |
inline::std::ostream & | operator<< (::std::ostream &os, const ReductionOperation &op) |
Formatted output of the Reduction Operations. More... | |
std::string | to_string (const ReductionOperation &op) |
Formatted output of the Reduction Operations. More... | |
std::string | to_string (const NormType &type) |
Formatted output of the Norm Type. More... | |
std::string | to_string (const PoolingType &type) |
Formatted output of the Pooling Type. More... | |
std::string | to_string (const PoolingLayerInfo &info) |
Formatted output of the Pooling Layer Info. More... | |
inline::std::ostream & | operator<< (::std::ostream &os, const KeyPoint &point) |
Formatted output of the KeyPoint type. More... | |
inline::std::ostream & | operator<< (::std::ostream &os, const PhaseType &phase_type) |
Formatted output of the PhaseType type. More... | |
std::string | to_string (const arm_compute::PhaseType &type) |
Formatted output of the PhaseType type. More... | |
inline::std::ostream & | operator<< (::std::ostream &os, const MagnitudeType &magnitude_type) |
Formatted output of the MagnitudeType type. More... | |
std::string | to_string (const arm_compute::MagnitudeType &type) |
Formatted output of the MagnitudeType type. More... | |
inline::std::ostream & | operator<< (::std::ostream &os, const GradientDimension &dim) |
Formatted output of the GradientDimension type. More... | |
std::string | to_string (const arm_compute::GradientDimension &type) |
Formatted output of the GradientDimension type. More... | |
inline::std::ostream & | operator<< (::std::ostream &os, const HOGNormType &norm_type) |
Formatted output of the HOGNormType type. More... | |
std::string | to_string (const HOGNormType &type) |
Formatted output of the HOGNormType type. More... | |
inline::std::ostream & | operator<< (::std::ostream &os, const Size2D &size) |
Formatted output of the Size2D type. More... | |
std::string | to_string (const Size2D &type) |
Formatted output of the Size2D type. More... | |
inline::std::ostream & | operator<< (::std::ostream &os, const HOGInfo &hog_info) |
Formatted output of the HOGInfo type. More... | |
std::string | to_string (const HOGInfo &type) |
Formatted output of the HOGInfo type. More... | |
inline::std::ostream & | operator<< (::std::ostream &os, const ConvolutionMethod &conv_method) |
Formatted output of the ConvolutionMethod type. More... | |
std::string | to_string (const ConvolutionMethod &conv_method) |
Formatted output of the ConvolutionMethod type. More... | |
inline::std::ostream & | operator<< (::std::ostream &os, const GPUTarget &gpu_target) |
Formatted output of the GPUTarget type. More... | |
std::string | to_string (const GPUTarget &gpu_target) |
Formatted output of the GPUTarget type. More... | |
inline::std::ostream & | operator<< (::std::ostream &os, const DetectionWindow &detection_window) |
Formatted output of the DetectionWindow type. More... | |
std::string | to_string (const DetectionWindow &detection_window) |
Formatted output of the DetectionWindow type. More... | |
inline::std::ostream & | operator<< (::std::ostream &os, const Termination &termination) |
Formatted output of the Termination type. More... | |
std::string | to_string (const Termination &termination) |
Formatted output of the Termination type. More... | |
inline::std::ostream & | operator<< (::std::ostream &os, const WinogradInfo &info) |
Formatted output of the WinogradInfo type. More... | |
std::string | to_string (const WinogradInfo &type) |
Variables | |
constexpr size_t | MAX_DIMS = 6 |
Constant value used to indicate maximum dimensions of a Window, TensorShape and Coordinates. More... | |
const std::array< float32x4_t, 8 > | exp_tab |
Exponent polynomial coefficients. More... | |
const std::array< float32x4_t, 8 > | log_tab |
Logarithm polynomial coefficients. More... | |
constexpr uint8_t | CONSTANT_BORDER_VALUE = 199 |
Constant value of the border pixels when using BorderMode::CONSTANT. More... | |
constexpr float | SCALE_PYRAMID_HALF = 0.5f |
Constant value used to indicate a half-scale pyramid. More... | |
constexpr float | SCALE_PYRAMID_ORB = 8.408964152537146130583778358414e-01 |
Constant value used to indicate a ORB scaled pyramid. More... | |
This file contains all available output stages for GEMMLowp on OpenCL.
This file contains all available output stages for GEMMLowp on NEON.
In gemmlowp, the "output stage" is the process that takes a final int32 accumulator value (the output of CLGEMMLowpMatrixMultiplyCore), and processes it to obtain the final ASYMM8 value.
More information about the GEMMLowp output stage can be found at https://github.com/google/gemmlowp/blob/master/doc/output.md
In gemmlowp, the "output stage" is the process that takes a final int32 accumulator value (the output of NEGEMMLowpMatrixMultiplyCore), and processes it to obtain the final ASYMM8 value.
More information about the GEMMLowp output stage can be found at https://github.com/google/gemmlowp/blob/master/doc/output.md
using AssemblyKernelGlueF32 = AssemblyKernelGlue<float, float> |
Float 32 assembly kernel glue.
Definition at line 121 of file AssemblyHelper.h.
using AssemblyKernelGlueS8S32 = AssemblyKernelGlue<int8_t, int32_t> |
Int 8 to Int 32 kernel glue.
Definition at line 125 of file AssemblyHelper.h.
using AssemblyKernelGlueU8U32 = AssemblyKernelGlue<uint8_t, uint32_t> |
Uint 8 to Uint 32 kernel glue.
Definition at line 123 of file AssemblyHelper.h.
OpenCL Array of Coefficient Tables.
Definition at line 49 of file CLOpticalFlow.h.
using CLConvolution3x3Kernel = CLConvolutionKernel<3> |
Interface for the kernel which applies a 3x3 convolution to a tensor.
Definition at line 70 of file CLConvolutionKernel.h.
using CLConvolution5x5 = CLConvolutionSquare<5> |
Basic function to run 5x5 convolution.
Definition at line 102 of file CLConvolution.h.
using CLConvolution5x5Kernel = CLConvolutionKernel<5> |
Interface for the kernel which applies a 5x5 convolution to a tensor.
Definition at line 72 of file CLConvolutionKernel.h.
using CLConvolution7x7 = CLConvolutionSquare<7> |
Basic function to run 7x7 convolution.
Definition at line 104 of file CLConvolution.h.
using CLConvolution7x7Kernel = CLConvolutionKernel<7> |
Interface for the kernel which applies a 7x7 convolution to a tensor.
Definition at line 74 of file CLConvolutionKernel.h.
using CLConvolution9x9 = CLConvolutionSquare<9> |
Basic function to run 9x9 convolution.
Definition at line 106 of file CLConvolution.h.
using CLConvolution9x9Kernel = CLConvolutionKernel<9> |
Interface for the kernel which applies a 9x9 convolution to a tensor.
Definition at line 76 of file CLConvolutionKernel.h.
using CLCoordinates2DArray = CLArray<Coordinates2D> |
OpenCL Array of 2D Coordinates.
using CLDetectionWindowArray = CLArray<DetectionWindow> |
using CLFloatArray = CLArray<cl_float> |
OpenCL Image.
Definition at line 80 of file CLTensor.h.
using CLInt16Array = CLArray<cl_short> |
using CLInt32Array = CLArray<cl_int> |
using CLKeyPointArray = CLArray<KeyPoint> |
OpenCL Array of Internal Keypoints.
Definition at line 47 of file CLOpticalFlow.h.
typedef MemoryGroupBase< CLTensor > CLMemoryGroup |
Memory Group in OpenCL.
Definition at line 35 of file CLMemoryGroup.h.
using CLOldValueArray = CLArray<CLOldValue> |
OpenCL Array of Old Values.
Definition at line 51 of file CLOpticalFlow.h.
Interface for the kernel which applies a horizontal pass of 5x5 convolution to a tensor.
Definition at line 106 of file CLConvolutionKernel.h.
Interface for the kernel which applies a vertical pass of 5x5 convolution to a tensor.
Definition at line 133 of file CLConvolutionKernel.h.
Interface for the kernel which applies a horizontal pass of 7x7 convolution to a tensor.
Definition at line 108 of file CLConvolutionKernel.h.
Interface for the kernel which applies a vertical pass of 7x7 convolution to a tensor.
Definition at line 135 of file CLConvolutionKernel.h.
Interface for the kernel which applies a horizontal pass of 9x9 convolution to a tensor.
Definition at line 110 of file CLConvolutionKernel.h.
Interface for the kernel which applies a vertical pass of 9x9 convolution to a tensor.
Definition at line 137 of file CLConvolutionKernel.h.
using CLSize2DArray = CLArray<Size2D> |
using CLUInt16Array = CLArray<cl_ushort> |
using CLUInt32Array = CLArray<cl_uint> |
using CLUInt8Array = CLArray<cl_uchar> |
using Coordinates2DArray = Array<Coordinates2D> |
Array of 2D Coordinates.
using DetectionWindowArray = Array<DetectionWindow> |
using FloatArray = Array<float> |
Interface for the 1x1 direct convolution kernel.
Definition at line 86 of file GCDirectConvolutionLayerKernel.h.
Interface for the 3x3 direct convolution kernel.
Definition at line 88 of file GCDirectConvolutionLayerKernel.h.
Interface for the 5x5 direct convolution kernel.
Definition at line 90 of file GCDirectConvolutionLayerKernel.h.
OpenGL ES Image.
Definition at line 98 of file GCTensor.h.
typedef MemoryGroupBase< GCTensor > GCMemoryGroup |
Definition at line 35 of file GCMemoryGroup.h.
using GroupMappings = std::map<size_t, MemoryMappings> |
Interface for OpenCL Array of Coefficient Tables.
Definition at line 68 of file CLLKTrackerKernel.h.
using ICLCoordinates2DArray = ICLArray<Coordinates2D> |
Interface for OpenCL Array of 2D Coordinates.
Definition at line 121 of file ICLArray.h.
Interface for OpenCL Array of Detection Windows.
Definition at line 123 of file ICLArray.h.
using ICLFloatArray = ICLArray<cl_float> |
Interface for OpenCL Array of floats.
Definition at line 139 of file ICLArray.h.
Interface for OpenCL images.
Definition at line 33 of file ICLMultiImage.h.
using ICLInt16Array = ICLArray<cl_short> |
Interface for OpenCL Array of int16s.
Definition at line 135 of file ICLArray.h.
using ICLInt32Array = ICLArray<cl_int> |
Interface for OpenCL Array of int32s.
Definition at line 137 of file ICLArray.h.
using ICLKeyPointArray = ICLArray<KeyPoint> |
Interface for OpenCL Array of Key Points.
Definition at line 119 of file ICLArray.h.
Interface for OpenCL Array of Internal Key Points.
Definition at line 66 of file CLLKTrackerKernel.h.
using ICLOldValArray = ICLArray<CLOldValue> |
Interface for OpenCL Array of Old Values.
Definition at line 70 of file CLLKTrackerKernel.h.
using ICLROIArray = ICLArray<ROI> |
Interface for OpenCL Array of ROIs.
Definition at line 125 of file ICLArray.h.
using ICLSize2DArray = ICLArray<Size2D> |
Interface for OpenCL Array of 2D Sizes.
Definition at line 127 of file ICLArray.h.
using ICLUInt16Array = ICLArray<cl_ushort> |
Interface for OpenCL Array of uint16s.
Definition at line 131 of file ICLArray.h.
using ICLUInt32Array = ICLArray<cl_uint> |
Interface for OpenCL Array of uint32s.
Definition at line 133 of file ICLArray.h.
using ICLUInt8Array = ICLArray<cl_uchar> |
Interface for OpenCL Array of uint8s.
Definition at line 129 of file ICLArray.h.
using ICoordinates2DArray = IArray<Coordinates2D> |
Interface for Array of 2D Coordinates.
using IDetectionWindowArray = IArray<DetectionWindow> |
using IFloatArray = IArray<float> |
Interface for GLES Compute image.
Definition at line 111 of file IGCTensor.h.
Interface for CPP Images.
Definition at line 37 of file CPPCornerCandidatesKernel.h.
using IInt16Array = IArray<int16_t> |
using IInt32Array = IArray<int32_t> |
using IKeyPointArray = IArray<KeyPoint> |
using INEKernel = ICPPKernel |
Common interface for all kernels implemented in NEON.
Definition at line 32 of file INEKernel.h.
Interface for NEON Array of Internal Key Points.
Definition at line 49 of file NELKTrackerKernel.h.
using INESimpleKernel = ICPPSimpleKernel |
Interface for simple NEON kernels having 1 tensor input and 1 tensor output.
Definition at line 32 of file INESimpleKernel.h.
using Int16Array = Array<int16_t> |
using Int32Array = Array<int32_t> |
using InternalKeypoint = std::tuple<float, float, float> |
using ISize2DArray = IArray<Size2D> |
using IUInt16Array = IArray<uint16_t> |
using IUInt32Array = IArray<uint32_t> |
using IUInt8Array = IArray<uint8_t> |
Array of LK Internel Keypoints.
Definition at line 46 of file NEOpticalFlow.h.
typedef MemoryGroupBase< Tensor > MemoryGroup |
Memory Group.
Definition at line 34 of file MemoryGroup.h.
using MemoryMappings = std::map<void **, size_t> |
A map of (handle, index/offset), where handle is the memory handle of the object to provide the memory for and index/offset is the buffer/offset from the pool that should be used.
Interface for the accumulate weighted kernel using F16.
Definition at line 105 of file NEAccumulateKernel.h.
using NEBox3x3FP16Kernel = NEBox3x3Kernel |
NEON kernel to perform a Box 3x3 filter for FP16 datatype.
Definition at line 68 of file NEBox3x3Kernel.h.
using NEConvolution3x3Kernel = NEConvolutionKernel<3> |
Interface for the kernel which applied a 3x3 convolution to a tensor.
Definition at line 88 of file NEConvolutionKernel.h.
using NEConvolution5x5 = NEConvolutionSquare<5> |
Basic function to run 5x5 convolution.
Definition at line 102 of file NEConvolution.h.
using NEConvolution5x5Kernel = NEConvolutionKernel<5> |
Interface for the kernel which applied a 5x5 convolution to a tensor.
Definition at line 90 of file NEConvolutionKernel.h.
using NEConvolution7x7 = NEConvolutionSquare<7> |
Basic function to run 7x7 convolution.
Definition at line 104 of file NEConvolution.h.
using NEConvolution7x7Kernel = NEConvolutionKernel<7> |
Interface for the kernel which applied a 7x7 convolution to a tensor.
Definition at line 92 of file NEConvolutionKernel.h.
using NEConvolution9x9 = NEConvolutionSquare<9> |
Basic function to run 9x9 convolution.
Definition at line 106 of file NEConvolution.h.
using NEConvolution9x9Kernel = NEConvolutionKernel<9> |
Interface for the kernel which applied a 9x9 convolution to a tensor.
Definition at line 94 of file NEConvolutionKernel.h.
using NEGradientFP16Kernel = NEGradientKernel |
NEON kernel to perform Gradient computation for FP16 datatype.
Definition at line 103 of file NECannyEdgeKernel.h.
using NEHarrisScoreFP16Kernel = NEHarrisScoreKernel<block_size> |
Interface for the accumulate Weighted kernel using FP16.
Definition at line 132 of file NEHarrisCornersKernel.h.
using NEMagnitudePhaseFP16Kernel = NEMagnitudePhaseKernel<mag_type, phase_type> |
Template interface for the kernel to compute magnitude and phase.
Definition at line 170 of file NEMagnitudePhaseKernel.h.
NEON kernel to perform Non-Maxima suppression 3x3 with intermediate results in FP16 if the input data type is FP32.
Definition at line 105 of file NENonMaximaSuppression3x3Kernel.h.
using NEScheduler = Scheduler |
NEON Scheduler.
Definition at line 32 of file NEScheduler.h.
Interface for the kernel which applied a 5x1 horizontal convolution to a tensor.
Definition at line 138 of file NEConvolutionKernel.h.
Interface for the kernel which applied a 1x5 vertical convolution to a tensor.
Definition at line 198 of file NEConvolutionKernel.h.
Interface for the kernel which applied a 7x1 horizontal convolution to a tensor.
Definition at line 140 of file NEConvolutionKernel.h.
Interface for the kernel which applied a 1x7 vertical convolution to a tensor.
Definition at line 200 of file NEConvolutionKernel.h.
Interface for the kernel which applied a 9x1 horizontal convolution to a tensor.
Definition at line 142 of file NEConvolutionKernel.h.
Interface for the kernel which applied a 1x9 vertical convolution to a tensor.
Definition at line 202 of file NEConvolutionKernel.h.
using PaddingSize = BorderSize |
using PermutationVector = Strides |
using qasymm8_t = uint8_t |
using qasymm8x16_t = uint8x16_t |
using qasymm8x8_t = uint8x8_t |
using qasymm8x8x2_t = uint8x8x2_t |
using qasymm8x8x3_t = uint8x8x3_t |
using qasymm8x8x4_t = uint8x8x4_t |
using qint16_t = int16_t |
16 bit fixed point scalar value
Definition at line 30 of file FixedPoint.h.
using qint16x4_t = int16x4_t |
16 bit fixed point vector with 4 elements
Definition at line 41 of file NEFixedPoint.h.
using qint16x4x2_t = int16x4x2_t |
16 bit fixed point vector with 8 elements
Definition at line 42 of file NEFixedPoint.h.
using qint16x4x3_t = int16x4x3_t |
16 bit fixed point vector with 12 elements
Definition at line 43 of file NEFixedPoint.h.
using qint16x4x4_t = int16x4x4_t |
16 bit fixed point vector with 16 elements
Definition at line 44 of file NEFixedPoint.h.
using qint16x8_t = int16x8_t |
16 bit fixed point vector with 8 elements
Definition at line 45 of file NEFixedPoint.h.
using qint16x8x2_t = int16x8x2_t |
16 bit fixed point vector with 16 elements
Definition at line 46 of file NEFixedPoint.h.
using qint16x8x3_t = int16x8x3_t |
16 bit fixed point vector with 24 elements
Definition at line 47 of file NEFixedPoint.h.
using qint16x8x4_t = int16x8x4_t |
16 bit fixed point vector with 32 elements
Definition at line 48 of file NEFixedPoint.h.
using qint32_t = int32_t |
32 bit fixed point scalar value
Definition at line 31 of file FixedPoint.h.
using qint32x2_t = int32x2_t |
32 bit fixed point vector with 2 elements
Definition at line 49 of file NEFixedPoint.h.
using qint32x4_t = int32x4_t |
32 bit fixed point vector with 4 elements
Definition at line 50 of file NEFixedPoint.h.
using qint32x4x2_t = int32x4x2_t |
32 bit fixed point vector with 8 elements
Definition at line 51 of file NEFixedPoint.h.
using qint64_t = int64_t |
64 bit fixed point scalar value
Definition at line 32 of file FixedPoint.h.
using qint8_t = int8_t |
8 bit fixed point scalar value
Definition at line 29 of file FixedPoint.h.
using qint8x16_t = int8x16_t |
8 bit fixed point vector with 16 elements
Definition at line 37 of file NEFixedPoint.h.
using qint8x16x2_t = int8x16x2_t |
8 bit fixed point vector with 32 elements
Definition at line 38 of file NEFixedPoint.h.
using qint8x16x3_t = int8x16x3_t |
8 bit fixed point vector with 48 elements
Definition at line 39 of file NEFixedPoint.h.
using qint8x16x4_t = int8x16x4_t |
8 bit fixed point vector with 64 elements
Definition at line 40 of file NEFixedPoint.h.
using qint8x8_t = int8x8_t |
8 bit fixed point vector with 8 elements
Definition at line 33 of file NEFixedPoint.h.
using qint8x8x2_t = int8x8x2_t |
8 bit fixed point vector with 16 elements
Definition at line 34 of file NEFixedPoint.h.
using qint8x8x3_t = int8x8x3_t |
8 bit fixed point vector with 24 elements
Definition at line 35 of file NEFixedPoint.h.
using qint8x8x4_t = int8x8x4_t |
8 bit fixed point vector with 32 elements
Definition at line 36 of file NEFixedPoint.h.
using UInt16Array = Array<uint16_t> |
using UInt32Array = Array<uint32_t> |
using UInt8Array = Array<uint8_t> |
|
strong |
|
strong |
Methods available to handle borders.
Definition at line 283 of file Types.h.
|
strong |
Available channels.
Definition at line 481 of file Types.h.
|
strong |
|
strong |
|
strong |
|
strong |
CPU models - we only need to detect CPUs we have microarchitecture-specific code for.
Architecture features are detected via HWCAPs.
Enumerator | |
---|---|
GENERIC | |
A53 | |
A55r0 | |
A55r1 |
Definition at line 36 of file CPPTypes.h.
|
strong |
|
strong |
|
strong |
Available data types.
Definition at line 72 of file Types.h.
|
strong |
|
strong |
|
strong |
|
strong |
Image colour formats.
Definition at line 50 of file Types.h.
|
strong |
Available GPU Targets.
Enumerator | |
---|---|
UNKNOWN | |
GPU_ARCH_MASK | |
MIDGARD | |
BIFROST | |
T600 | |
T700 | |
T800 | |
G71 | |
G72 | |
G51 | |
G51BIG | |
G51LIT | |
TNOX | |
TTRX | |
TBOX |
Definition at line 34 of file GPUTarget.h.
|
strong |
|
strong |
|
strong |
Interpolation method.
Definition at line 388 of file Types.h.
|
strong |
|
strong |
|
strong |
|
strong |
|
strong |
The normalization type used for the normalization layer.
Enumerator | |
---|---|
IN_MAP_1D |
Normalization applied within the same map in 1D region. |
IN_MAP_2D |
Normalization applied within the same map in 2D region. |
CROSS_MAP |
Normalization applied cross maps. |
Definition at line 522 of file Types.h.
|
strong |
|
strong |
|
strong |
|
strong |
Rounding method.
Definition at line 30 of file Rounding.h.
|
strong |
|
strong |
Termination criteria.
Definition at line 410 of file Types.h.
|
strong |
|
inline |
Decrease required
in steps of step
until it's less than available
.
[in] | required | Number of required bytes. |
[in] | available | Number of available bytes. |
[in] | step | Step size used to decrease required bytes. |
available
that is a multiple of step
Definition at line 47 of file IAccessWindow.h.
References ARM_COMPUTE_ERROR_ON.
|
inline |
Adjust tensor shape size if width or height are odd for a given multi-planar format.
No modification is done for other formats.
[in,out] | shape | Tensor shape of 2D size |
[in] | format | Format of the tensor |
Definition at line 688 of file Utils.h.
References has_format_horizontal_subsampling(), has_format_vertical_subsampling(), TensorShape::set(), and U.
Referenced by error_on_tensors_not_even().
|
inline |
Increase required
in steps of step
until it's greater than available
.
[in] | required | Number of required bytes. |
[in] | available | Number of available bytes. |
[in] | step | Step size used to increase required bytes. |
available
that is a multiple of step
Definition at line 63 of file IAccessWindow.h.
References ARM_COMPUTE_ERROR_ON.
|
inline |
Allocate a workspace tensor.
[in] | workspace_size | Size to allocate. |
[out] | workspace | Tensor to allocate. |
[in] | memory_group | Tensor memory group. |
[in] | alignment | Workspace memory alignment. |
[in] | num_threads | Number of workspace threads. |
Definition at line 135 of file AssemblyHelper.h.
References TensorAllocator::allocate(), Tensor::allocator(), ARM_COMPUTE_ERROR_ON_MSG, ARM_COMPUTE_UNUSED, TensorAllocator::init(), and S8.
Referenced by setup_assembly_kernel().
bool arm_compute::arm_non_uniform_workgroup_supported | ( | const cl::Device & | device | ) |
Helper function to check whether the arm_non_uniform_work_group_size extension is supported.
[in] | device | A CL device |
|
inline |
Auto initialize the tensor info (shape, number of channels, data type and fixed point position) if the current assignment is empty.
[in,out] | info | Tensor info used to check and assign. |
[in] | shape | New shape. |
[in] | num_channels | New number of channels. |
[in] | data_type | New data type |
[in] | fixed_point_position | New fixed point position |
[in] | quantization_info | (Optional) New quantization info |
Definition at line 201 of file Helpers.inl.
References ITensorInfo::set_data_type(), ITensorInfo::set_fixed_point_position(), ITensorInfo::set_num_channels(), ITensorInfo::set_quantization_info(), ITensorInfo::set_tensor_shape(), ITensorInfo::tensor_shape(), and TensorShape::total_size().
Referenced by permute().
|
inline |
Auto initialize the tensor info using another tensor info.
Definition at line 221 of file Helpers.inl.
References ITensorInfo::data_layout(), ITensorInfo::data_type(), ITensorInfo::fixed_point_position(), ITensorInfo::num_channels(), ITensorInfo::quantization_info(), ITensorInfo::set_data_layout(), ITensorInfo::set_data_type(), ITensorInfo::set_fixed_point_position(), ITensorInfo::set_num_channels(), ITensorInfo::set_quantization_info(), ITensorInfo::set_tensor_shape(), ITensorInfo::tensor_shape(), and TensorShape::total_size().
std::string arm_compute::build_information | ( | ) |
Returns the arm_compute library build information.
Contains the version number and the build options used to build the library
Referenced by floor_to_multiple(), and main().
TensorShape arm_compute::calculate_depth_concatenate_shape | ( | const std::vector< T * > & | inputs_vector | ) |
Calculate the output shapes of the depth concatenate function.
[in] | inputs_vector | The vector that stores all the pointers to input. |
Definition at line 651 of file Utils.h.
References ARM_COMPUTE_ERROR_ON, arm_compute::test::fixed_point_arithmetic::detail::max(), TensorShape::set(), arm_compute::test::validation::shape, Dimensions< T >::x(), Dimensions< T >::y(), and Dimensions< T >::z().
|
inline |
Calculate the scale of the given square matrix.
The scale is the absolute value of the sum of all the coefficients in the matrix.
[in] | matrix | Matrix coefficients |
[in] | matrix_size | Number of elements per side of the square matrix. (Number of coefficients = matrix_size * matrix_size). |
Definition at line 637 of file Utils.h.
References arm_compute::test::fixed_point_arithmetic::detail::abs(), accumulate(), and arm_compute::test::fixed_point_arithmetic::detail::max().
Window arm_compute::calculate_max_enlarged_window | ( | const ValidRegion & | valid_region, |
const Steps & | steps = Steps() , |
||
BorderSize | border_size = BorderSize() |
||
) |
Calculate the maximum window for a given tensor shape and border setting.
The window will also includes the border.
[in] | valid_region | Valid region object defining the shape of the tensor space for which the window is created. |
[in] | steps | (Optional) Number of elements processed for each step. |
[in] | border_size | (Optional) Border size. The border region will be included in the window. |
Referenced by calculate_max_enlarged_window(), and calculate_max_window_horizontal().
|
inline |
Calculate the maximum window for a given tensor shape and border setting.
The window will also includes the border.
[in] | info | Tensor info object defining the shape of the object for which the window is created. |
[in] | steps | (Optional) Number of elements processed for each step. |
[in] | border_size | (Optional) Border size. The border region will be included in the window. |
Definition at line 457 of file Helpers.h.
References calculate_max_enlarged_window(), and ITensorInfo::valid_region().
Window arm_compute::calculate_max_window | ( | const ValidRegion & | valid_region, |
const Steps & | steps = Steps() , |
||
bool | skip_border = false , |
||
BorderSize | border_size = BorderSize() |
||
) |
Calculate the maximum window for a given tensor shape and border setting.
[in] | valid_region | Valid region object defining the shape of the tensor space for which the window is created. |
[in] | steps | (Optional) Number of elements processed for each step. |
[in] | skip_border | (Optional) If true exclude the border region from the window. |
[in] | border_size | (Optional) Border size. |
Referenced by calculate_max_window(), and update_window_and_padding().
|
inline |
Calculate the maximum window for a given tensor shape and border setting.
[in] | info | Tensor info object defining the shape of the object for which the window is created. |
[in] | steps | (Optional) Number of elements processed for each step. |
[in] | skip_border | (Optional) If true exclude the border region from the window. |
[in] | border_size | (Optional) Border size. |
Definition at line 409 of file Helpers.h.
References calculate_max_window(), calculate_max_window_horizontal(), and ITensorInfo::valid_region().
Window arm_compute::calculate_max_window_horizontal | ( | const ValidRegion & | valid_region, |
const Steps & | steps = Steps() , |
||
bool | skip_border = false , |
||
BorderSize | border_size = BorderSize() |
||
) |
Calculate the maximum window used by a horizontal kernel for a given tensor shape and border setting.
[in] | valid_region | Valid region object defining the shape of the tensor space for which the window is created. |
[in] | steps | (Optional) Number of elements processed for each step. |
[in] | skip_border | (Optional) If true exclude the border region from the window. |
[in] | border_size | (Optional) Border size. The border region will be excluded from the window. |
Referenced by calculate_max_window(), and calculate_max_window_horizontal().
|
inline |
Calculate the maximum window used by a horizontal kernel for a given tensor shape and border setting.
[in] | info | Tensor info object defining the shape of the object for which the window is created. |
[in] | steps | (Optional) Number of elements processed for each step. |
[in] | skip_border | (Optional) If true exclude the border region from the window. |
[in] | border_size | (Optional) Border size. |
Definition at line 434 of file Helpers.h.
References calculate_max_enlarged_window(), calculate_max_window_horizontal(), and ITensorInfo::valid_region().
PadStrideInfo arm_compute::calculate_same_pad | ( | TensorShape | input_shape, |
TensorShape | weights_shape, | ||
PadStrideInfo | conv_info | ||
) |
Calculate padding requirements in case of SAME padding.
[in] | input_shape | Input shape |
[in] | weights_shape | Weights shape |
[in] | conv_info | Convolution information (containing strides) |
Referenced by data_type_for_convolution_matrix().
|
inline |
Calculate subsampled shape for a given format and channel.
[in] | shape | Shape of the tensor to calculate the extracted channel. |
[in] | format | Format of the tensor. |
[in] | channel | Channel to create tensor shape to be extracted. |
Definition at line 715 of file Utils.h.
References has_format_horizontal_subsampling(), has_format_vertical_subsampling(), TensorShape::set(), U, UNKNOWN, and V.
Referenced by arm_compute::test::validation::reference::channel_extract(), and error_on_tensors_not_subsampled().
ValidRegion arm_compute::calculate_valid_region_scale | ( | const ITensorInfo & | src_info, |
const TensorShape & | dst_shape, | ||
InterpolationPolicy | interpolate_policy, | ||
SamplingPolicy | sampling_policy, | ||
bool | border_undefined | ||
) |
Helper function to calculate the Valid Region for Scale.
[in] | src_info | Input tensor info used to check. |
[in] | dst_shape | Shape of the output. |
[in] | interpolate_policy | Interpolation policy. |
[in] | sampling_policy | Sampling policy. |
[in] | border_undefined | True if the border is undefined. |
Referenced by arm_compute::test::validation::DATA_TEST_CASE(), arm_compute::test::validation::FIXTURE_DATA_TEST_CASE(), and permute().
|
inline |
Computes the smallest number larger or equal to value that is a multiple of divisor.
[in] | value | Lower bound value |
[in] | divisor | Value to compute multiple of. |
Definition at line 64 of file Utils.h.
References ARM_COMPUTE_ERROR_ON, and DIV_CEIL().
Referenced by arm_compute::test::validation::DATA_TEST_CASE(), and Window::scale().
Return the channel index of a given channel given an input format.
[in] | format | Input format |
[in] | channel | Input channel |
Definition at line 318 of file Utils.h.
References A, ARM_COMPUTE_ERROR, B, G, IYUV, NV12, NV21, R, RGB888, RGBA8888, U, UYVY422, V, Y, YUV444, and YUYV422.
Referenced by arm_compute::test::validation::reference::channel_extract().
void arm_compute::colorconvert_iyuv_to_nv12 | ( | const void *__restrict | input, |
void *__restrict | output, | ||
const Window & | win | ||
) |
Convert IYUV to NV12.
[in] | input | Input IYUV data buffer. |
[out] | output | Output NV12 buffer. |
[in] | win | Window for iterating the buffers. |
Definition at line 601 of file NEColorConvertHelper.inl.
References ARM_COMPUTE_ERROR_ON, Window::DimX, Window::DimY, Window::Dimension::end(), execute_window_loop(), Window::set(), Window::Dimension::start(), Window::Dimension::step(), Window::validate(), Window::x(), and Window::y().
void arm_compute::colorconvert_iyuv_to_rgb | ( | const void *__restrict | input, |
void *__restrict | output, | ||
const Window & | win | ||
) |
Convert IYUV to RGB.
[in] | input | Input IYUV data buffer. |
[out] | output | Output RGB buffer. |
[in] | win | Window for iterating the buffers. |
Definition at line 482 of file NEColorConvertHelper.inl.
References arm_compute::test::validation::alpha, ARM_COMPUTE_ERROR_ON, Window::DimX, Window::DimY, Window::Dimension::end(), execute_window_loop(), ITensor::info(), Window::set(), Window::Dimension::start(), Window::Dimension::step(), ITensorInfo::strides_in_bytes(), Window::validate(), Window::x(), Dimensions< T >::y(), and Window::y().
void arm_compute::colorconvert_iyuv_to_yuv4 | ( | const void *__restrict | input, |
void *__restrict | output, | ||
const Window & | win | ||
) |
Convert IYUV to YUV4.
[in] | input | Input IYUV data buffer. |
[out] | output | Output YUV4 buffer. |
[in] | win | Window for iterating the buffers. |
Definition at line 816 of file NEColorConvertHelper.inl.
References ARM_COMPUTE_ERROR_ON, Window::DimX, Window::DimY, Window::Dimension::end(), execute_window_loop(), Window::set(), Window::Dimension::start(), Window::Dimension::step(), Window::validate(), Window::x(), and Window::y().
void arm_compute::colorconvert_nv12_to_iyuv | ( | const void *__restrict | input, |
void *__restrict | output, | ||
const Window & | win | ||
) |
Convert NV12 to IYUV.
[in] | input | Input NV12 data buffer. |
[out] | output | Output IYUV buffer. |
[in] | win | Window for iterating the buffers. |
Definition at line 649 of file NEColorConvertHelper.inl.
References ARM_COMPUTE_ERROR_ON, Window::DimX, Window::DimY, Window::Dimension::end(), execute_window_loop(), Window::set(), Window::Dimension::start(), Window::Dimension::step(), Window::validate(), Window::x(), and Window::y().
void arm_compute::colorconvert_nv12_to_rgb | ( | const void *__restrict | input, |
void *__restrict | output, | ||
const Window & | win | ||
) |
Convert NV12 to RGB.
[in] | input | Input NV12 data buffer. |
[out] | output | Output RGB buffer. |
[in] | win | Window for iterating the buffers. |
Definition at line 419 of file NEColorConvertHelper.inl.
References arm_compute::test::validation::alpha, ARM_COMPUTE_ERROR_ON, Window::DimX, Window::DimY, Window::Dimension::end(), execute_window_loop(), ITensor::info(), Window::set(), Window::Dimension::start(), Window::Dimension::step(), ITensorInfo::strides_in_bytes(), Window::validate(), Window::x(), Dimensions< T >::y(), and Window::y().
void arm_compute::colorconvert_nv12_to_yuv4 | ( | const void *__restrict | input, |
void *__restrict | output, | ||
const Window & | win | ||
) |
Convert NV12 to YUV4.
[in] | input | Input NV12 data buffer. |
[out] | output | Output YUV4 buffer. |
[in] | win | Window for iterating the buffers. |
Definition at line 758 of file NEColorConvertHelper.inl.
References ARM_COMPUTE_ERROR_ON, Window::DimX, Window::DimY, Window::Dimension::end(), execute_window_loop(), Window::set(), Window::Dimension::start(), Window::Dimension::step(), Window::validate(), Window::x(), and Window::y().
void arm_compute::colorconvert_rgb_to_iyuv | ( | const void *__restrict | input, |
void *__restrict | output, | ||
const Window & | win | ||
) |
Convert RGB to IYUV.
[in] | input | Input RGB data buffer. |
[out] | output | Output IYUV buffer. |
[in] | win | Window for iterating the buffers. |
Definition at line 918 of file NEColorConvertHelper.inl.
References arm_compute::test::validation::alpha, ARM_COMPUTE_ERROR_ON, Window::DimX, Window::DimY, Window::Dimension::end(), execute_window_loop(), Iterator::ptr(), Window::set(), Window::Dimension::start(), Window::Dimension::step(), Window::validate(), Window::x(), and Window::y().
void arm_compute::colorconvert_rgb_to_nv12 | ( | const void *__restrict | input, |
void *__restrict | output, | ||
const Window & | win | ||
) |
Convert RGB to NV12.
[in] | input | Input RGB data buffer. |
[out] | output | Output NV12 buffer. |
[in] | win | Window for iterating the buffers. |
Definition at line 875 of file NEColorConvertHelper.inl.
References arm_compute::test::validation::alpha, ARM_COMPUTE_ERROR_ON, Window::DimX, Window::DimY, Window::Dimension::end(), execute_window_loop(), Iterator::ptr(), Window::set(), Window::Dimension::start(), Window::Dimension::step(), Window::validate(), Window::x(), and Window::y().
void arm_compute::colorconvert_rgb_to_rgbx | ( | const void *__restrict | input, |
void *__restrict | output, | ||
const Window & | win | ||
) |
Convert RGB to RGBX.
[in] | input | Input RGB data buffer. |
[out] | output | Output RGBX buffer. |
[in] | win | Window for iterating the buffers. |
Definition at line 312 of file NEColorConvertHelper.inl.
References ARM_COMPUTE_ERROR_ON, execute_window_loop(), and Iterator::ptr().
void arm_compute::colorconvert_rgb_to_yuv4 | ( | const void *__restrict | input, |
void *__restrict | output, | ||
const Window & | win | ||
) |
Convert RGB to YUV4.
[in] | input | Input RGB data buffer. |
[out] | output | Output YUV4 buffer. |
[in] | win | Window for iterating the buffers. |
Definition at line 962 of file NEColorConvertHelper.inl.
References arm_compute::test::validation::alpha, ARM_COMPUTE_ERROR_ON, execute_window_loop(), Iterator::ptr(), and Window::validate().
void arm_compute::colorconvert_rgbx_to_rgb | ( | const void * | input, |
void * | output, | ||
const Window & | win | ||
) |
Convert RGBX to RGB.
[in] | input | Input RGBX data buffer. |
[out] | output | Output RGB buffer. |
[in] | win | Window for iterating the buffers. |
Definition at line 343 of file NEColorConvertHelper.inl.
References ARM_COMPUTE_ERROR_ON, execute_window_loop(), and Iterator::ptr().
void arm_compute::colorconvert_yuyv_to_iyuv | ( | const void *__restrict | input, |
void *__restrict | output, | ||
const Window & | win | ||
) |
Convert YUYV to IYUV.
[in] | input | Input YUYV data buffer. |
[out] | output | Output IYUV buffer. |
[in] | win | Window for iterating the buffers. |
Definition at line 698 of file NEColorConvertHelper.inl.
References ARM_COMPUTE_ERROR_ON, Window::DimX, Window::DimY, Window::Dimension::end(), execute_window_loop(), Iterator::ptr(), Window::set(), Window::Dimension::start(), Window::Dimension::step(), Window::validate(), Window::x(), and Window::y().
void arm_compute::colorconvert_yuyv_to_nv12 | ( | const void *__restrict | input, |
void *__restrict | output, | ||
const Window & | win | ||
) |
Convert YUYV to NV12.
[in] | input | Input YUYV data buffer. |
[out] | output | Output NV12 buffer. |
[in] | win | Window for iterating the buffers. |
Definition at line 546 of file NEColorConvertHelper.inl.
References ARM_COMPUTE_ERROR_ON, Window::DimX, Window::DimY, Window::Dimension::end(), execute_window_loop(), Iterator::ptr(), Window::set(), Window::Dimension::start(), Window::Dimension::step(), Window::validate(), Window::x(), and Window::y().
void arm_compute::colorconvert_yuyv_to_rgb | ( | const void *__restrict | input, |
void *__restrict | output, | ||
const Window & | win | ||
) |
Convert YUYV to RGB.
[in] | input | Input YUYV data buffer. |
[out] | output | Output RGB buffer. |
[in] | win | Window for iterating the buffers. |
Definition at line 374 of file NEColorConvertHelper.inl.
References arm_compute::test::validation::alpha, ARM_COMPUTE_ERROR_ON, execute_window_loop(), and Iterator::ptr().
|
inline |
Create a strides object based on the provided strides and the tensor dimensions.
[in] | info | Tensor info object providing the shape of the tensor for unspecified strides. |
[in] | stride_x | Stride to be used in X dimension (in bytes). |
[in] | fixed_strides | Strides to be used in higher dimensions starting at Y (in bytes). |
Definition at line 501 of file Helpers.h.
References ITensorInfo::num_dimensions(), Dimensions< T >::set(), arm_compute::test::validation::shape, and ITensorInfo::tensor_shape().
Referenced by compute_strides().
|
inline |
Create a strides object based on the tensor dimensions.
[in] | info | Tensor info object used to compute the strides. |
Definition at line 523 of file Helpers.h.
References compute_strides(), and ITensorInfo::element_size().
|
inline |
Convert n-dimensional coordinates into a linear index.
[in] | shape | Shape of the n-dimensional tensor. |
[in] | coord | N-dimensional coordinates. |
Definition at line 322 of file Helpers.inl.
References ARM_COMPUTE_ERROR_ON_MSG, ARM_COMPUTE_UNUSED, Dimensions< T >::num_dimensions(), and TensorShape::total_size().
Referenced by arm_compute::test::validation::reference::convert_fully_connected_weights(), permute(), and arm_compute::test::validation::reference::winograd_input_transform().
std::unique_ptr<Kernel> arm_compute::create_configure_kernel | ( | T &&... | args | ) |
Helper function to create and return a unique_ptr pointed to a CL/GLES kernel object It also calls the kernel's configuration.
[in] | args | All the arguments that need pass to kernel's configuration. |
Status arm_compute::create_error | ( | ErrorCode | error_code, |
const char * | function, | ||
const char * | file, | ||
const int | line, | ||
const char * | msg, | ||
... | |||
) |
Creates an error containing the error message.
[in] | error_code | Error code |
[in] | function | Function in which the error occurred. |
[in] | file | Name of the file where the error occurred. |
[in] | line | Line on which the error occurred. |
[in] | msg | Message to display before aborting. |
[in] | ... | Variable number of arguments of the message. |
Referenced by Status::throw_if_error().
Status arm_compute::create_error_va_list | ( | ErrorCode | error_code, |
const char * | function, | ||
const char * | file, | ||
const int | line, | ||
const char * | msg, | ||
va_list | args | ||
) |
Creates an error containing the error message from variable argument list.
[in] | error_code | Error code |
[in] | function | Function in which the error occurred. |
[in] | file | Name of the file where the error occurred. |
[in] | line | Line on which the error occurred. |
[in] | msg | Message to display before aborting. |
[in] | args | Variable argument list of the message. |
Referenced by Status::throw_if_error().
std::unique_ptr<Kernel> arm_compute::create_kernel | ( | ) |
Helper function to create and return a unique_ptr pointed to a CL/GLES kernel object.
Definition at line 86 of file Helpers.h.
Referenced by GCKernelLibrary::set_context(), and CLKernelLibrary::set_device().
|
inline |
The size in bytes of the data type.
[in] | data_type | Input data type |
Definition at line 107 of file Utils.h.
References ARM_COMPUTE_ERROR, F16, F32, F64, QASYMM8, QS16, QS32, QS8, S16, S32, S64, S8, SIZET, U16, U32, U64, and U8.
Referenced by arm_compute::test::validation::DATA_TEST_CASE(), TensorInfo::element_size(), and AlexNetNetwork< ITensorType, TensorType, SubTensorType, Accessor, ActivationLayerFunction, ConvolutionLayerFunction, DirectConvolutionLayerFunction, FullyConnectedLayerFunction, NormalizationLayerFunction, PoolingLayerFunction, SoftmaxLayerFunction >::init().
|
inline |
Calculate accurary required by the horizontal and vertical convolution computations.
[in] | conv_col | Pointer to the vertical vector of the separated convolution filter |
[in] | conv_row | Pointer to the horizontal vector of the convolution filter |
[in] | size | Number of elements per vector of the separated matrix |
Definition at line 747 of file Utils.h.
References accumulate(), S16, S32, U16, and UNKNOWN.
|
inline |
Calculate the accuracy required by the squared convolution calculation.
[in] | conv | Pointer to the squared convolution matrix |
[in] | size | The total size of the convolution matrix |
Definition at line 803 of file Utils.h.
References arm_compute::test::validation::a, accumulate(), arm_compute::test::validation::b, arm_compute::test::validation::border_mode, calculate_same_pad(), arm_compute::test::validation::conv_info, deconvolution_output_dimensions(), deconvolution_output_shape(), lower_string(), S16, S32, scaled_dimensions(), string_from_activation_func(), string_from_border_mode(), string_from_channel(), string_from_data_layout(), string_from_data_type(), string_from_format(), string_from_interpolation_policy(), string_from_matrix_pattern(), string_from_non_linear_filter_function(), string_from_norm_type(), string_from_pooling_type(), U, and U16.
Return the data type used by a given single-planar pixel format.
[in] | format | Input format |
Definition at line 213 of file Utils.h.
References ARM_COMPUTE_ERROR, F16, F32, IYUV, NV12, NV21, RGB888, RGBA8888, S16, S32, U16, U32, U8, UNKNOWN, UV88, UYVY422, YUV444, and YUYV422.
Referenced by arm_compute::test::validation::DATA_TEST_CASE(), and SimpleTensor< T >::data_type().
const std::pair<unsigned int, unsigned int> arm_compute::deconvolution_output_dimensions | ( | unsigned int | in_width, |
unsigned int | in_height, | ||
unsigned int | kernel_width, | ||
unsigned int | kernel_height, | ||
unsigned int | padx, | ||
unsigned int | pady, | ||
unsigned int | inner_border_right, | ||
unsigned int | inner_border_top, | ||
unsigned int | stride_x, | ||
unsigned int | stride_y | ||
) |
Returns expected width and height of the deconvolution's output tensor.
[in] | in_width | Width of input tensor (Number of columns) |
[in] | in_height | Height of input tensor (Number of rows) |
[in] | kernel_width | Kernel width. |
[in] | kernel_height | Kernel height. |
[in] | padx | X axis padding. |
[in] | pady | Y axis padding. |
[in] | inner_border_right | The number of zeros added to right edge of the input. |
[in] | inner_border_top | The number of zeros added to top edge of the input. |
[in] | stride_x | X axis input stride. |
[in] | stride_y | Y axis input stride. |
Referenced by arm_compute::test::validation::DATA_TEST_CASE(), and data_type_for_convolution_matrix().
TensorShape arm_compute::deconvolution_output_shape | ( | const std::pair< unsigned int, unsigned int > & | out_dims, |
TensorShape | input, | ||
TensorShape | weights | ||
) |
Returns expected shape for the deconvolution output tensor.
[in] | out_dims | widht and height of the output tensor, these values can be obtained with the function deconvolution_output_dimensions. |
[in] | input | Shape of the input tensor. |
[in] | weights | Shape of the weights tensor. |
Referenced by arm_compute::test::validation::DATA_TEST_CASE(), and data_type_for_convolution_matrix().
|
inline |
Computes bilinear interpolation using the pointer to the top-left pixel and the pixel's distance between the real coordinates and the smallest following integer coordinates.
Input must be in single channel format.
[in] | pixel_ptr | Pointer to the top-left pixel value of a single channel input. |
[in] | stride | Stride to access the bottom-left and bottom-right pixel values |
[in] | dx | Pixel's distance between the X real coordinate and the smallest X following integer |
[in] | dy | Pixel's distance between the Y real coordinate and the smallest Y following integer |
Definition at line 127 of file Helpers.h.
References ARM_COMPUTE_ERROR_ON.
Referenced by pixel_bilinear_c1(), and pixel_bilinear_c1_clamp().
|
inline |
Computes linear interpolation using the pointer to the left pixel and the pixel's distance between the real coordinates and the smallest following integer coordinates.
Input must be in single channel format.
[in] | pixel_ptr | Pointer to the left pixel value of a single channel input. |
[in] | dx | Pixel's distance between the X real coordinate and the smallest X following integer |
Definition at line 184 of file Helpers.h.
References ARM_COMPUTE_ERROR_ON.
Referenced by pixel_bilinear_c1_clamp().
|
inline |
Computes linear interpolation using the pointer to the top pixel and the pixel's distance between the real coordinates and the smallest following integer coordinates.
Input must be in single channel format.
[in] | pixel_ptr | Pointer to the top pixel value of a single channel input. |
[in] | stride | Stride to access the bottom pixel value |
[in] | dy | Pixel's distance between the Y real coordinate and the smallest Y following integer |
Definition at line 159 of file Helpers.h.
References ARM_COMPUTE_ERROR_ON.
Referenced by pixel_bilinear_c1_clamp().
bool arm_compute::device_supports_extension | ( | const cl::Device & | device, |
const char * | extension_name | ||
) |
Helper function to check whether a given extension is supported.
[in] | device | A CL device |
[in] | extension_name | Name of the extension to be checked |
Referenced by CLScheduler::default_init().
constexpr auto arm_compute::DIV_CEIL | ( | S | val, |
T | m | ||
) | -> decltype((val + m - 1) / m) |
Calculate the rounded up quotient of val / m.
[in] | val | Value to divide and round up. |
[in] | m | Value to divide by. |
Definition at line 51 of file Utils.h.
Referenced by ceil_to_multiple().
|
inline |
The size in bytes of the data type.
[in] | dt | Input data type |
Definition at line 182 of file Utils.h.
References ARM_COMPUTE_ERROR, F16, F32, QASYMM8, QS16, QS32, QS8, S16, S32, S8, U16, U32, and U8.
Referenced by SimpleTensor< T >::element_size(), error_on_value_not_representable_in_fixed_point(), and arm_compute::test::validation::validate().
void arm_compute::enqueue | ( | IGCKernel & | kernel, |
const Window & | window, | ||
const gles::NDRange & | lws = gles::NDRange(1U, 1U, 1U) |
||
) |
Add the kernel to the command queue with the given window.
[in] | kernel | Kernel to enqueue |
[in] | window | Window the kernel has to process. |
[in] | lws | Local workgroup size requested, by default (1, 1, 1) |
void arm_compute::enqueue | ( | cl::CommandQueue & | queue, |
ICLKernel & | kernel, | ||
const Window & | window, | ||
const cl::NDRange & | lws_hint = CLKernelLibrary::get().default_ndrange() |
||
) |
Add the kernel to the command queue with the given window.
[in,out] | queue | OpenCL command queue. |
[in] | kernel | Kernel to enqueue |
[in] | window | Window the kernel has to process. |
[in] | lws_hint | Local workgroup size requested. Default is based on the device target. |
Referenced by IGCKernel::get_target(), and ICLKernel::get_target().
void arm_compute::error | ( | const char * | function, |
const char * | file, | ||
const int | line, | ||
const char * | msg, | ||
... | |||
) |
Print an error message then throw an std::runtime_error.
[in] | function | Function in which the error occurred. |
[in] | file | Name of the file where the error occurred. |
[in] | line | Line on which the error occurred. |
[in] | msg | Message to display before aborting. |
[in] | ... | Variable number of arguments of the message. |
Referenced by Framework::error_on_missing_assets(), main(), and Status::throw_if_error().
|
inline |
Return an error if the channel is not in channels.
[in] | function | Function in which the error occurred. |
[in] | file | Name of the file where the error occurred. |
[in] | line | Line on which the error occurred. |
[in] | cn | Input channel |
[in] | channel | First channel allowed. |
[in] | channels | (Optional) Further allowed channels. |
Definition at line 835 of file Validate.h.
References ARM_COMPUTE_RETURN_ERROR_ON_LOC, and UNKNOWN.
arm_compute::Status arm_compute::error_on_channel_not_in_known_format | ( | const char * | function, |
const char * | file, | ||
const int | line, | ||
Format | fmt, | ||
Channel | cn | ||
) |
Return an error if the channel is not in format.
[in] | function | Function in which the error occurred. |
[in] | file | Name of the file where the error occurred. |
[in] | line | Line on which the error occurred. |
[in] | fmt | Input channel |
[in] | cn | First channel allowed. |
arm_compute::Status arm_compute::error_on_coordinates_dimensions_gte | ( | const char * | function, |
const char * | file, | ||
const int | line, | ||
const Coordinates & | pos, | ||
unsigned int | max_dim | ||
) |
Return an error if the passed coordinates have too many dimensions.
The coordinates have too many dimensions if any of the dimensions greater or equal to max_dim is different from 0.
[in] | function | Function in which the error occurred. |
[in] | file | Name of the file where the error occurred. |
[in] | line | Line on which the error occurred. |
[in] | pos | Coordinates to validate |
[in] | max_dim | Maximum number of dimensions allowed. |
|
inline |
Return an error if the data type or the number of channels of the passed tensor info does not match any of the data types and number of channels provided.
[in] | function | Function in which the error occurred. |
[in] | file | Name of the file where the error occurred. |
[in] | line | Line on which the error occurred. |
[in] | tensor_info | Tensor info to validate. |
[in] | num_channels | Number of channels to check |
[in] | dt | First data type allowed. |
[in] | dts | (Optional) Further allowed data types. |
Definition at line 774 of file Validate.h.
References ARM_COMPUTE_RETURN_ERROR_ON_LOC_MSG, ARM_COMPUTE_RETURN_ON_ERROR, error_on_data_type_not_in(), and ITensorInfo::num_channels().
Referenced by error_on_data_type_channel_not_in().
|
inline |
Return an error if the data type or the number of channels of the passed tensor does not match any of the data types and number of channels provided.
[in] | function | Function in which the error occurred. |
[in] | file | Name of the file where the error occurred. |
[in] | line | Line on which the error occurred. |
[in] | tensor | Tensor to validate. |
[in] | num_channels | Number of channels to check |
[in] | dt | First data type allowed. |
[in] | dts | (Optional) Further allowed data types. |
Definition at line 795 of file Validate.h.
References ARM_COMPUTE_RETURN_ERROR_ON_LOC, ARM_COMPUTE_RETURN_ON_ERROR, error_on_data_type_channel_not_in(), and ITensor::info().
|
inline |
Return an error if the data type of the passed tensor info does not match any of the data types provided.
[in] | function | Function in which the error occurred. |
[in] | file | Name of the file where the error occurred. |
[in] | line | Line on which the error occurred. |
[in] | tensor_info | Tensor info to validate. |
[in] | dt | First data type allowed. |
[in] | dts | (Optional) Further allowed data types. |
Definition at line 721 of file Validate.h.
References ARM_COMPUTE_RETURN_ERROR_ON_LOC, ARM_COMPUTE_RETURN_ERROR_ON_LOC_MSG, ITensorInfo::data_type(), string_from_data_type(), and UNKNOWN.
Referenced by error_on_data_type_channel_not_in(), and error_on_data_type_not_in().
|
inline |
Return an error if the data type of the passed tensor does not match any of the data types provided.
[in] | function | Function in which the error occurred. |
[in] | file | Name of the file where the error occurred. |
[in] | line | Line on which the error occurred. |
[in] | tensor | Tensor to validate. |
[in] | dt | First data type allowed. |
[in] | dts | (Optional) Further allowed data types. |
Definition at line 749 of file Validate.h.
References ARM_COMPUTE_RETURN_ERROR_ON_LOC, ARM_COMPUTE_RETURN_ON_ERROR, error_on_data_type_not_in(), and ITensor::info().
void arm_compute::error_on_format_not_in | ( | const char * | function, |
const char * | file, | ||
const int | line, | ||
const T * | object, | ||
F && | format, | ||
Fs &&... | formats | ||
) |
Throw an error if the format of the passed tensor/multi-image does not match any of the formats provided.
[in] | function | Function in which the error occurred. |
[in] | file | Name of the file where the error occurred. |
[in] | line | Line on which the error occurred. |
[in] | object | Tensor/multi-image to validate. |
[in] | format | First format allowed. |
[in] | formats | (Optional) Further allowed formats. |
Definition at line 688 of file Validate.h.
References ARM_COMPUTE_ERROR_ON_LOC, ARM_COMPUTE_ERROR_ON_LOC_MSG, ARM_COMPUTE_UNUSED, string_from_format(), and UNKNOWN.
arm_compute::Status arm_compute::error_on_invalid_multi_hog | ( | const char * | function, |
const char * | file, | ||
const int | line, | ||
const IMultiHOG * | multi_hog | ||
) |
Return an error if the IMultiHOG container is invalid.
An IMultiHOG container is invalid if:
[in] | function | Function in which the error occurred. |
[in] | file | Name of the file where the error occurred. |
[in] | line | Line on which the error occurred. |
[in] | multi_hog | IMultiHOG container to validate |
arm_compute::Status arm_compute::error_on_invalid_subtensor | ( | const char * | function, |
const char * | file, | ||
const int | line, | ||
const TensorShape & | parent_shape, | ||
const Coordinates & | coords, | ||
const TensorShape & | shape | ||
) |
Return an error if if the coordinates and shape of the subtensor are within the parent tensor.
[in] | function | Function in which the error occurred. |
[in] | file | Name of the file where the error occurred. |
[in] | line | Line on which the error occurred. |
[in] | parent_shape | Parent tensor shape |
[in] | coords | Coordinates inside the parent tensor where the first element of the subtensor is |
[in] | shape | Shape of the subtensor |
arm_compute::Status arm_compute::error_on_invalid_subtensor_valid_region | ( | const char * | function, |
const char * | file, | ||
const int | line, | ||
const ValidRegion & | parent_valid_region, | ||
const ValidRegion & | valid_region | ||
) |
Return an error if the valid region of a subtensor is not inside the valid region of the parent tensor.
[in] | function | Function in which the error occurred. |
[in] | file | Name of the file where the error occurred. |
[in] | line | Line on which the error occurred. |
[in] | parent_valid_region | Parent valid region. |
[in] | valid_region | Valid region of subtensor. |
arm_compute::Status arm_compute::error_on_invalid_subwindow | ( | const char * | function, |
const char * | file, | ||
const int | line, | ||
const Window & | full, | ||
const Window & | sub | ||
) |
Return an error if the passed subwindow is invalid.
The subwindow is invalid if:
[in] | function | Function in which the error occurred. |
[in] | file | Name of the file where the error occurred. |
[in] | line | Line on which the error occurred. |
[in] | full | Full size window |
[in] | sub | Sub-window to validate. |
|
inline |
Return an error if the passed tensor infos have different data layouts.
[in] | function | Function in which the error occurred. |
[in] | file | Name of the file where the error occurred. |
[in] | line | Line on which the error occurred. |
[in] | tensor_info | The first tensor info to be compared. |
[in] | tensor_infos | (Optional) Further allowed tensor infos. |
Definition at line 457 of file Validate.h.
References ARM_COMPUTE_RETURN_ERROR_ON_LOC, ARM_COMPUTE_RETURN_ERROR_ON_LOC_MSG, ARM_COMPUTE_RETURN_ON_ERROR, ITensorInfo::data_layout(), and error_on_nullptr().
Referenced by error_on_mismatching_data_layouts().
|
inline |
Return an error if the passed tensors have different data layouts.
[in] | function | Function in which the error occurred. |
[in] | file | Name of the file where the error occurred. |
[in] | line | Line on which the error occurred. |
[in] | tensor | The first tensor to be compared. |
[in] | tensors | (Optional) Further allowed tensors. |
Definition at line 483 of file Validate.h.
References ARM_COMPUTE_RETURN_ERROR_ON_LOC, ARM_COMPUTE_RETURN_ON_ERROR, error_on_mismatching_data_layouts(), error_on_nullptr(), and ITensor::info().
|
inline |
Return an error if the passed two tensor infos have different data types.
[in] | function | Function in which the error occurred. |
[in] | file | Name of the file where the error occurred. |
[in] | line | Line on which the error occurred. |
[in] | tensor_info | The first tensor info to be compared. |
[in] | tensor_infos | (Optional) Further allowed tensor infos. |
Definition at line 508 of file Validate.h.
References ARM_COMPUTE_RETURN_ERROR_ON_LOC, ARM_COMPUTE_RETURN_ERROR_ON_LOC_MSG, ARM_COMPUTE_RETURN_ON_ERROR, ITensorInfo::data_type(), and error_on_nullptr().
Referenced by error_on_mismatching_data_types().
|
inline |
Return an error if the passed two tensors have different data types.
[in] | function | Function in which the error occurred. |
[in] | file | Name of the file where the error occurred. |
[in] | line | Line on which the error occurred. |
[in] | tensor | The first tensor to be compared. |
[in] | tensors | (Optional) Further allowed tensors. |
Definition at line 534 of file Validate.h.
References ARM_COMPUTE_RETURN_ERROR_ON_LOC, ARM_COMPUTE_RETURN_ON_ERROR, error_on_mismatching_data_types(), error_on_nullptr(), and ITensor::info().
arm_compute::Status arm_compute::error_on_mismatching_dimensions | ( | const char * | function, |
const char * | file, | ||
int | line, | ||
const Dimensions< T > & | dim1, | ||
const Dimensions< T > & | dim2, | ||
Ts &&... | dims | ||
) |
Return an error if the passed dimension objects differ.
[in] | function | Function in which the error occurred. |
[in] | file | Name of the file where the error occurred. |
[in] | line | Line on which the error occurred. |
[in] | dim1 | The first object to be compared. |
[in] | dim2 | The second object to be compared. |
[in] | dims | (Optional) Further allowed objects. |
Definition at line 280 of file Validate.h.
References ARM_COMPUTE_RETURN_ON_ERROR, and arm_compute::detail::for_each_error().
|
inline |
Return an error if the passed tensor infos have different fixed point data types or different fixed point positions.
[in] | function | Function in which the error occurred. |
[in] | file | Name of the file where the error occurred. |
[in] | line | Line on which the error occurred. |
[in] | tensor_info_1 | The first tensor info to be compared. |
[in] | tensor_info_2 | The second tensor info to be compared. |
[in] | tensor_infos | (Optional) Further allowed tensor infos. |
Definition at line 562 of file Validate.h.
References ARM_COMPUTE_RETURN_ERROR_ON_LOC_MSG, ITensorInfo::data_type(), ITensorInfo::fixed_point_position(), and is_data_type_fixed_point().
Referenced by error_on_mismatching_fixed_point().
|
inline |
Return an error if the passed tensor have different fixed point data types or different fixed point positions.
[in] | function | Function in which the error occurred. |
[in] | file | Name of the file where the error occurred. |
[in] | line | Line on which the error occurred. |
[in] | tensor_1 | The first tensor to be compared. |
[in] | tensor_2 | The second tensor to be compared. |
[in] | tensors | (Optional) Further allowed tensors. |
Definition at line 601 of file Validate.h.
References ARM_COMPUTE_RETURN_ON_ERROR, error_on_mismatching_fixed_point(), and ITensor::info().
|
inline |
Return an error if the input fixed-point positions are different.
[in] | function | Function in which the error occurred. |
[in] | file | Name of the file where the error occurred. |
[in] | line | Line on which the error occurred. |
[in] | tensor_info_1 | The first tensor info to be compared. |
[in] | tensor_info_2 | The second tensor info to be compared. |
[in] | tensor_infos | (Optional) Further allowed tensor infos. |
Definition at line 955 of file Validate.h.
References ARM_COMPUTE_RETURN_ERROR_ON_LOC_MSG, and ITensorInfo::fixed_point_position().
Referenced by error_on_mismatching_fixed_point_position().
|
inline |
Return an error if the input fixed-point positions are different.
[in] | function | Function in which the error occurred. |
[in] | file | Name of the file where the error occurred. |
[in] | line | Line on which the error occurred. |
[in] | tensor_1 | The first tensor to be compared. |
[in] | tensor_2 | The second tensor to be compared. |
[in] | tensors | (Optional) Further allowed tensors. |
Definition at line 978 of file Validate.h.
References ARM_COMPUTE_RETURN_ON_ERROR, error_on_mismatching_fixed_point_position(), and ITensor::info().
|
inline |
Return an error if the passed tensor infos have different asymmetric quantized data types or different quantization info.
[in] | function | Function in which the error occurred. |
[in] | file | Name of the file where the error occurred. |
[in] | line | Line on which the error occurred. |
[in] | tensor_info_1 | The first tensor info to be compared. |
[in] | tensor_info_2 | The second tensor info to be compared. |
[in] | tensor_infos | (Optional) Further allowed tensor infos. |
Definition at line 627 of file Validate.h.
References ARM_COMPUTE_RETURN_ERROR_ON_LOC_MSG, ITensorInfo::data_type(), is_data_type_quantized_asymmetric(), and ITensorInfo::quantization_info().
Referenced by error_on_mismatching_quantization_info().
|
inline |
Return an error if the passed tensor have different asymmetric quantized data types or different quantization info.
[in] | function | Function in which the error occurred. |
[in] | file | Name of the file where the error occurred. |
[in] | line | Line on which the error occurred. |
[in] | tensor_1 | The first tensor to be compared. |
[in] | tensor_2 | The second tensor to be compared. |
[in] | tensors | (Optional) Further allowed tensors. |
Definition at line 666 of file Validate.h.
References ARM_COMPUTE_RETURN_ON_ERROR, error_on_mismatching_quantization_info(), and ITensor::info().
|
inline |
Return an error if the passed two tensor infos have different shapes from the given dimension.
[in] | function | Function in which the error occurred. |
[in] | file | Name of the file where the error occurred. |
[in] | line | Line on which the error occurred. |
[in] | tensor_info_1 | The first tensor info to be compared. |
[in] | tensor_info_2 | The second tensor info to be compared. |
[in] | tensor_infos | (Optional) Further allowed tensor infos. |
Definition at line 368 of file Validate.h.
References U.
Referenced by error_on_mismatching_shapes().
|
inline |
Return an error if the passed two tensors have different shapes from the given dimension.
[in] | function | Function in which the error occurred. |
[in] | file | Name of the file where the error occurred. |
[in] | line | Line on which the error occurred. |
[in] | tensor_1 | The first tensor to be compared. |
[in] | tensor_2 | The second tensor to be compared. |
[in] | tensors | (Optional) Further allowed tensors. |
Definition at line 385 of file Validate.h.
References error_on_mismatching_shapes(), and U.
|
inline |
Return an error if the passed two tensors have different shapes from the given dimension.
[in] | function | Function in which the error occurred. |
[in] | file | Name of the file where the error occurred. |
[in] | line | Line on which the error occurred. |
[in] | upper_dim | The dimension from which to check. |
[in] | tensor_info_1 | The first tensor info to be compared. |
[in] | tensor_info_2 | The second tensor info to be compared. |
[in] | tensor_infos | (Optional) Further allowed tensor infos. |
Definition at line 403 of file Validate.h.
References ARM_COMPUTE_RETURN_ERROR_ON_LOC, ARM_COMPUTE_RETURN_ERROR_ON_LOC_MSG, ARM_COMPUTE_RETURN_ON_ERROR, error_on_nullptr(), and arm_compute::detail::have_different_dimensions().
|
inline |
Return an error if the passed two tensors have different shapes from the given dimension.
[in] | function | Function in which the error occurred. |
[in] | file | Name of the file where the error occurred. |
[in] | line | Line on which the error occurred. |
[in] | upper_dim | The dimension from which to check. |
[in] | tensor_1 | The first tensor to be compared. |
[in] | tensor_2 | The second tensor to be compared. |
[in] | tensors | (Optional) Further allowed tensors. |
Definition at line 431 of file Validate.h.
References ARM_COMPUTE_RETURN_ERROR_ON_LOC, ARM_COMPUTE_RETURN_ON_ERROR, error_on_mismatching_shapes(), error_on_nullptr(), and ITensor::info().
arm_compute::Status arm_compute::error_on_mismatching_windows | ( | const char * | function, |
const char * | file, | ||
const int | line, | ||
const Window & | full, | ||
const Window & | win | ||
) |
Return an error if the passed window is invalid.
The subwindow is invalid if:
[in] | function | Function in which the error occurred. |
[in] | file | Name of the file where the error occurred. |
[in] | line | Line on which the error occurred. |
[in] | full | Full size window |
[in] | win | Window to validate. |
|
inline |
Create an error if one of the pointers is a nullptr.
[in] | function | Function in which the error occurred. |
[in] | file | Name of the file where the error occurred. |
[in] | line | Line on which the error occurred. |
[in] | pointers | Pointers to check against nullptr. |
Definition at line 151 of file Validate.h.
References ARM_COMPUTE_RETURN_ERROR_ON_LOC_MSG.
Referenced by error_on_mismatching_data_layouts(), error_on_mismatching_data_types(), error_on_mismatching_shapes(), error_on_tensors_not_even(), and error_on_tensors_not_subsampled().
arm_compute::Status arm_compute::error_on_tensor_not_2d | ( | const char * | function, |
const char * | file, | ||
const int | line, | ||
const ITensor * | tensor | ||
) |
arm_compute::Status arm_compute::error_on_tensors_not_even | ( | const char * | function, |
const char * | file, | ||
int | line, | ||
const Format & | format, | ||
const ITensor * | tensor1, | ||
Ts... | tensors | ||
) |
Return an error if the passed tensor objects are not even.
[in] | function | Function in which the error occurred. |
[in] | file | Name of the file where the error occurred. |
[in] | line | Line on which the error occurred. |
[in] | format | Format to check if odd shape is allowed |
[in] | tensor1 | The first object to be compared for odd shape. |
[in] | tensors | (Optional) Further allowed objects. |
Definition at line 303 of file Validate.h.
References adjust_odd_shape(), ARM_COMPUTE_RETURN_ERROR_ON_LOC, ARM_COMPUTE_RETURN_ERROR_ON_LOC_MSG, ARM_COMPUTE_RETURN_ON_ERROR, error_on_nullptr(), and arm_compute::detail::have_different_dimensions().
arm_compute::Status arm_compute::error_on_tensors_not_subsampled | ( | const char * | function, |
const char * | file, | ||
int | line, | ||
const Format & | format, | ||
const TensorShape & | shape, | ||
const ITensor * | tensor1, | ||
Ts... | tensors | ||
) |
Return an error if the passed tensor objects are not sub-sampled.
[in] | function | Function in which the error occurred. |
[in] | file | Name of the file where the error occurred. |
[in] | line | Line on which the error occurred. |
[in] | format | Format to check if sub-sampling allowed. |
[in] | shape | The tensor shape to calculate sub-sampling from. |
[in] | tensor1 | The first object to be compared. |
[in] | tensors | (Optional) Further allowed objects. |
Definition at line 336 of file Validate.h.
References ARM_COMPUTE_RETURN_ERROR_ON_LOC, ARM_COMPUTE_RETURN_ERROR_ON_LOC_MSG, ARM_COMPUTE_RETURN_ON_ERROR, calculate_subsampled_shape(), error_on_nullptr(), and arm_compute::detail::have_different_dimensions().
arm_compute::Status arm_compute::error_on_unconfigured_kernel | ( | const char * | function, |
const char * | file, | ||
const int | line, | ||
const IKernel * | kernel | ||
) |
|
inline |
Return an error if the fixed-point value is not representable in the specified Q format.
[in] | function | Function in which the error occurred. |
[in] | file | Name of the file where the error occurred. |
[in] | line | Line on which the error occurred. |
[in] | value | The floating point value to be checked. |
[in] | tensor_info | Input tensor info that has information on data type and fixed-point position. |
Definition at line 1000 of file Validate.h.
References ARM_COMPUTE_RETURN_ERROR_ON_LOC_MSG, ITensorInfo::data_type(), element_size_from_data_type(), ITensorInfo::fixed_point_position(), and string_from_data_type().
Referenced by error_on_value_not_representable_in_fixed_point().
|
inline |
Return an error an error if the fixed-point value is not representable in the specified Q format.
[in] | function | Function in which the error occurred. |
[in] | file | Name of the file where the error occurred. |
[in] | line | Line on which the error occurred. |
[in] | value | The floating point value to be checked. |
[in] | tensor | Input tensor that has information on data type and fixed-point position. |
Definition at line 1022 of file Validate.h.
References ARM_COMPUTE_RETURN_ERROR_ON_LOC, ARM_COMPUTE_RETURN_ON_ERROR, error_on_value_not_representable_in_fixed_point(), and ITensor::info().
arm_compute::Status arm_compute::error_on_window_dimensions_gte | ( | const char * | function, |
const char * | file, | ||
const int | line, | ||
const Window & | win, | ||
unsigned int | max_dim | ||
) |
Return an error if the passed window has too many dimensions.
The window has too many dimensions if any of the dimension greater or equal to max_dim is different from 0.
[in] | function | Function in which the error occurred. |
[in] | file | Name of the file where the error occurred. |
[in] | line | Line on which the error occurred. |
[in] | win | Window to validate |
[in] | max_dim | Maximum number of dimensions allowed. |
arm_compute::Status arm_compute::error_on_window_not_collapsable_at_dimension | ( | const char * | function, |
const char * | file, | ||
const int | line, | ||
const Window & | full, | ||
const Window & | window, | ||
const int | dim | ||
) |
Return an error if the window can't be collapsed at the given dimension.
The window cannot be collapsed if the given dimension not equal to the full window's dimension or not start from 0.
[in] | function | Function in which the error occurred. |
[in] | file | Name of the file where the error occurred. |
[in] | line | Line on which the error occurred. |
[in] | full | Full size window |
[in] | window | Window to be collapsed. |
[in] | dim | Dimension need to be checked. |
|
inline |
Iterate through the passed window, automatically adjusting the iterators and calling the lambda_functino for each element.
It passes the x and y positions to the lambda_function for each iteration
[in] | w | Window to iterate through. |
[in] | lambda_function | The function of type void(function)( const Coordinates & id ) to call at each iteration. Where id represents the absolute coordinates of the item to process. |
[in,out] | iterators | Tensor iterators which will be updated by this function before calling lambda_function. |
Definition at line 122 of file Helpers.inl.
References ARM_COMPUTE_ERROR_ON, Dimensions< int >::num_max_dimensions, and Window::validate().
Referenced by NumPyBinLoader::access_tensor(), colorconvert_iyuv_to_nv12(), colorconvert_iyuv_to_rgb(), colorconvert_iyuv_to_yuv4(), colorconvert_nv12_to_iyuv(), colorconvert_nv12_to_rgb(), colorconvert_nv12_to_yuv4(), colorconvert_rgb_to_iyuv(), colorconvert_rgb_to_nv12(), colorconvert_rgb_to_rgbx(), colorconvert_rgb_to_yuv4(), colorconvert_rgbx_to_rgb(), colorconvert_yuyv_to_iyuv(), colorconvert_yuyv_to_nv12(), colorconvert_yuyv_to_rgb(), arm_compute::utils::compare_tensor(), AssetsLibrary::fill_borders_with_garbage(), PPMLoader::fill_image(), AssetsLibrary::fill_layer_data(), PPMLoader::fill_planar_tensor(), arm_compute::utils::fill_random_tensor(), NPYLoader::fill_tensor(), LeNet5Network< TensorType, Accessor, ActivationLayerFunction, ConvolutionLayerFunction, FullyConnectedLayerFunction, PoolingLayerFunction, SoftmaxLayerFunction >::get_classifications(), MobileNetNetwork< TensorType, Accessor, ActivationLayerFunction, ConvolutionLayerFunction, DirectConvolutionLayerFunction, DepthwiseConvolutionLayerFunction, ReshapeFunction, PoolingLayerFunction >::get_classifications(), MobileNetV1Network< TensorType, Accessor, ActivationLayerFunction, BatchNormalizationLayerFunction, ConvolutionLayerFunction, DirectConvolutionLayerFunction, DepthwiseConvolutionFunction, ReshapeFunction, PoolingLayerFunction, SoftmaxLayerFunction >::get_classifications(), AlexNetNetwork< ITensorType, TensorType, SubTensorType, Accessor, ActivationLayerFunction, ConvolutionLayerFunction, DirectConvolutionLayerFunction, FullyConnectedLayerFunction, NormalizationLayerFunction, PoolingLayerFunction, SoftmaxLayerFunction >::get_classifications(), arm_compute::utils::load_trained_data(), CaffePreproccessor::preprocess(), TFPreproccessor::preprocess(), RandomAccessor::RandomAccessor(), arm_compute::utils::save_to_npy(), and arm_compute::utils::save_to_ppm().
uint8x16_t arm_compute::finalize_quantization | ( | int32x4x4_t & | in_s32, |
int | result_fixedpoint_multiplier, | ||
int32_t | result_shift, | ||
int32x4_t | result_offset_after_shift_s32, | ||
uint8x16_t | min_u8, | ||
uint8x16_t | max_u8 | ||
) |
Performs final quantization step on 16 elements.
is_bounded_relu | Specified if a fused bounded relu should be applied |
in_s32 | Input to be quantized. |
result_fixedpoint_multiplier | Result multiplier parameter |
result_shift | Result shift parameter |
result_offset_after_shift_s32 | Result offset parameter |
min_u8 | Relu lower bound |
max_u8 | Relu upper bound |
Definition at line 74 of file NEAsymm.h.
References rounding_divide_by_pow2().
|
inline |
Create a string with the float in full precision.
val | Floating point value |
|
inline |
Computes the largest number smaller or equal to value that is a multiple of divisor.
[in] | value | Upper bound value |
[in] | divisor | Value to compute multiple of. |
Definition at line 78 of file Utils.h.
References ARM_COMPUTE_ERROR_ON, build_information(), and read_file().
bool arm_compute::fp16_supported | ( | const cl::Device & | device | ) |
Helper function to check whether the cl_khr_fp16 extension is supported.
[in] | device | A CL device |
Helper function to get the GPU arch.
[in] | target | GPU target |
std::string arm_compute::get_cl_type_from_data_type | ( | const DataType & | dt | ) |
Translates a tensor data type to the appropriate OpenCL type.
[in] | dt | DataType to be translated to OpenCL type. |
CLVersion arm_compute::get_cl_version | ( | const cl::Device & | device | ) |
Helper function to get the highest OpenCL version supported.
[in] | device | A CL device |
void arm_compute::get_cpu_configuration | ( | CPUInfo & | cpuinfo | ) |
This function will try to detect the CPU configuration on the system and will fill the cpuinfo object accordingly to reflect this.
[out] | cpuinfo | CPUInfo to be used to hold the system's cpu configuration. |
|
inline |
Get the index of the given dimension.
[in] | data_layout | The data layout. |
[in] | data_layout_dimension | The dimension which this index is requested for. |
Definition at line 340 of file Helpers.inl.
References ARM_COMPUTE_ERROR, ARM_COMPUTE_ERROR_ON_MSG, BATCHES, CHANNEL, HEIGHT, NCHW, UNKNOWN, and WIDTH.
Referenced by arm_compute::misc::shape_calculator::compute_deep_convolution_shape(), arm_compute::misc::shape_calculator::compute_depthwise_convolution_shape(), arm_compute::misc::shape_calculator::compute_im2col_conv_shape(), arm_compute::misc::shape_calculator::compute_pool_shape(), arm_compute::misc::shape_calculator::compute_winograd_filter_transform_shape(), arm_compute::misc::shape_calculator::compute_winograd_input_transform_shape(), arm_compute::misc::shape_calculator::compute_winograd_output_transform_shape(), arm_compute::test::validation::DATA_TEST_CASE(), SubTensorInfo::dimension(), TensorInfo::dimension(), PPMLoader::fill_planar_tensor(), permute(), and arm_compute::graph::backends::detail::validate_depthwise_convolution_layer().
std::string arm_compute::get_data_size_from_data_type | ( | const DataType & | dt | ) |
Get the size of a data type in number of bits.
[in] | dt | DataType. |
Return the promoted data type of a given data type.
[in] | dt | Data type to get the promoted type of. |
Definition at line 517 of file Utils.h.
References ARM_COMPUTE_ERROR, F16, F32, QASYMM8, QS16, QS32, QS8, S16, S32, S8, U16, U32, U8, and UNKNOWN.
GPUTarget arm_compute::get_target_from_device | ( | ) |
Helper function to get the GPU target from GLES using GL_RENDERER enum.
GPUTarget arm_compute::get_target_from_device | ( | cl::Device & | device | ) |
Helper function to get the GPU target from CL device.
[in] | device | A CL device |
Referenced by CLScheduler::init().
GPUTarget arm_compute::get_target_from_name | ( | const std::string & | device_name | ) |
Helper function to get the GPU target from a device name.
[in] | device_name | A device name |
Referenced by arm_compute::test::validation::TEST_CASE().
unsigned int arm_compute::get_threads_hint | ( | ) |
Some systems have both big and small cores, this fuction computes the minimum number of cores that are exactly the same on the system.
To maximize performance the library attempts to process workloads concurrently using as many threads as big cores are available on the system.
std::string arm_compute::get_underlying_cl_type_from_data_type | ( | const DataType & | dt | ) |
Translates fixed point tensor data type to the underlying OpenCL type.
[in] | dt | DataType to be translated to OpenCL type. |
Helper function to check whether a gpu target is equal to the provided targets.
[in] | target_to_check | gpu target to check |
[in] | target | First target to compare against |
[in] | targets | (Optional) Additional targets to compare with |
Definition at line 92 of file GPUTarget.h.
Referenced by arm_compute::test::validation::TEST_CASE().
Variant of gpu_target_is_in for comparing two targets.
Definition at line 98 of file GPUTarget.h.
|
inline |
Return true if the given format has horizontal subsampling.
[in] | format | Format to determine subsampling. |
Definition at line 552 of file Utils.h.
References IYUV, NV12, NV21, UV88, UYVY422, and YUYV422.
Referenced by adjust_odd_shape(), and calculate_subsampled_shape().
|
inline |
Return true if the given format has vertical subsampling.
[in] | format | Format to determine subsampling. |
Definition at line 563 of file Utils.h.
References IYUV, NV12, NV21, and UV88.
Referenced by adjust_odd_shape(), and calculate_subsampled_shape().
|
inline |
|
inline |
Convert a linear index into n-dimensional coordinates.
[in] | shape | Shape of the n-dimensional tensor. |
[in] | index | Linear index specifying the i-th element. |
Definition at line 303 of file Helpers.inl.
References ARM_COMPUTE_ERROR_ON_MSG, Dimensions< T >::num_dimensions(), TensorShape::set(), and TensorShape::total_size().
Referenced by arm_compute::test::validation::reference::convert_fully_connected_weights(), and permute().
ValidRegion arm_compute::intersect_valid_regions | ( | const Ts &... | regions | ) |
Intersect multiple valid regions.
[in] | regions | Valid regions. |
Definition at line 469 of file Helpers.h.
References ValidRegion::anchor, arm_compute::utility::foldl(), arm_compute::test::fixed_point_arithmetic::detail::max(), arm_compute::test::fixed_point_arithmetic::detail::min(), Dimensions< T >::num_dimensions(), Dimensions< T >::set(), TensorShape::set(), and ValidRegion::shape.
|
inline |
Check if a given data type is of fixed point type.
[in] | dt | Input data type. |
Definition at line 1037 of file Utils.h.
References QS16, QS32, and QS8.
Referenced by arm_compute::test::validation::DATA_TEST_CASE(), error_on_mismatching_fixed_point(), and arm_compute::test::validation::FIXTURE_DATA_TEST_CASE().
|
inline |
|
inline |
|
inline |
Check if a given data type is of asymmetric quantized type.
[in] | dt | Input data type. |
Definition at line 1056 of file Utils.h.
References QASYMM8.
Referenced by arm_compute::test::validation::DATA_TEST_CASE(), error_on_mismatching_quantization_info(), arm_compute::test::validation::reference::im2col_nchw(), arm_compute::test::validation::reference::im2col_nhwc(), set_quantization_info_if_empty(), and arm_compute::graph::backends::detail::validate_convolution_layer().
std::string arm_compute::lower_string | ( | const std::string & | val | ) |
Lower a given string.
[in] | val | Given string to lower. |
Referenced by data_type_for_convolution_matrix().
int arm_compute::max_consecutive_elements_display_width | ( | std::ostream & | s, |
DataType | dt, | ||
const uint8_t * | ptr, | ||
unsigned int | n | ||
) |
Identify the maximum width of n consecutive elements.
[in] | s | Output stream to print the elements to. |
[in] | dt | Data type of the elements |
[in] | ptr | Pointer to print the elements from. |
[in] | n | Number of elements to print. |
Referenced by max_consecutive_elements_display_width_impl().
int arm_compute::max_consecutive_elements_display_width_impl | ( | std::ostream & | s, |
const T * | ptr, | ||
unsigned int | n | ||
) |
Identify the maximum width of n consecutive elements.
[in] | s | The output stream which will be used to print the elements. Used to extract the stream format. |
[in] | ptr | Pointer to the elements. |
[in] | n | Number of elements. |
Definition at line 1123 of file Utils.h.
References max_consecutive_elements_display_width(), and print_consecutive_elements().
|
inline |
Return the number of channels for a given single-planar pixel format.
[in] | format | Input format |
Definition at line 476 of file Utils.h.
References F16, F32, IYUV, NV12, NV21, RGB888, RGBA8888, S16, S32, U16, U32, U8, UV88, UYVY422, YUV444, and YUYV422.
|
inline |
Return the number of planes for a given format.
[in] | format | Input format |
Definition at line 442 of file Utils.h.
References ARM_COMPUTE_ERROR, F16, F32, IYUV, NV12, NV21, RGB888, RGBA8888, S16, S32, U16, U32, U8, UYVY422, YUV444, and YUYV422.
bool arm_compute::opencl_is_available | ( | ) |
Check if OpenCL is available.
Referenced by main(), and arm_compute::test::sync_if_necessary().
bool arm_compute::opengles31_is_available | ( | ) |
Check if the OpenGL ES 3.1 API is available at runtime.
Referenced by NDRange::get(), and arm_compute::test::sync_tensor_if_necessary().
|
inline |
Check that given dimensions are not equal.
[in] | lhs | Left-hand side Dimensions. |
[in] | rhs | Right-hand side Dimensions. |
Definition at line 246 of file Dimensions.h.
inline ::std::ostream& arm_compute::operator<< | ( | ::std::ostream & | os, |
const Dimensions< T > & | dimensions | ||
) |
Formatted output of the Dimensions type.
[out] | os | Output stream. |
[in] | dimensions | Type to output. |
Definition at line 53 of file TypePrinter.h.
inline ::std::ostream& arm_compute::operator<< | ( | ::std::ostream & | os, |
const NonLinearFilterFunction & | function | ||
) |
Formatted output of the NonLinearFilterFunction type.
[out] | os | Output stream. |
[in] | function | Type to output. |
Definition at line 75 of file TypePrinter.h.
References ARM_COMPUTE_ERROR, MAX, MEDIAN, and MIN.
inline ::std::ostream& arm_compute::operator<< | ( | ::std::ostream & | os, |
const MatrixPattern & | pattern | ||
) |
Formatted output of the MatrixPattern type.
[out] | os | Output stream. |
[in] | pattern | Type to output. |
Definition at line 115 of file TypePrinter.h.
References ARM_COMPUTE_ERROR, BOX, CROSS, DISK, and OTHER.
inline ::std::ostream& arm_compute::operator<< | ( | ::std::ostream & | os, |
const RoundingPolicy & | rounding_policy | ||
) |
Formatted output of the RoundingPolicy type.
[out] | os | Output stream. |
[in] | rounding_policy | Type to output. |
Definition at line 158 of file TypePrinter.h.
References ARM_COMPUTE_ERROR, TO_NEAREST_EVEN, TO_NEAREST_UP, and TO_ZERO.
inline ::std::ostream& arm_compute::operator<< | ( | ::std::ostream & | os, |
const WeightsInfo & | weights_info | ||
) |
Formatted output of the WeightsInfo type.
[out] | os | Output stream. |
[in] | weights_info | Type to output. |
Definition at line 185 of file TypePrinter.h.
References WeightsInfo::are_reshaped(), WeightsInfo::kernel_size(), and WeightsInfo::num_kernels().
inline ::std::ostream& arm_compute::operator<< | ( | ::std::ostream & | os, |
const ROIPoolingLayerInfo & | pool_info | ||
) |
Formatted output of the ROIPoolingInfo type.
[out] | os | Output stream. |
[in] | pool_info | Type to output. |
Definition at line 200 of file TypePrinter.h.
References ROIPoolingLayerInfo::pooled_height(), ROIPoolingLayerInfo::pooled_width(), and ROIPoolingLayerInfo::spatial_scale().
inline ::std::ostream& arm_compute::operator<< | ( | ::std::ostream & | os, |
const QuantizationInfo & | quantization_info | ||
) |
Formatted output of the QuantizationInfo type.
[out] | os | Output stream. |
[in] | quantization_info | Type to output. |
Definition at line 213 of file TypePrinter.h.
References QuantizationInfo::offset, and QuantizationInfo::scale.
inline ::std::ostream& arm_compute::operator<< | ( | ::std::ostream & | os, |
const FixedPointOp & | op | ||
) |
Formatted output of the FixedPointOp type.
[out] | os | Output stream. |
[in] | op | Type to output. |
Definition at line 240 of file TypePrinter.h.
References ADD, ARM_COMPUTE_ERROR, EXP, INV_SQRT, LOG, MUL, RECIPROCAL, and SUB.
inline ::std::ostream& arm_compute::operator<< | ( | ::std::ostream & | os, |
const ActivationLayerInfo::ActivationFunction & | act_function | ||
) |
Formatted output of the activation function type.
[out] | os | Output stream. |
[in] | act_function | Type to output. |
Definition at line 292 of file TypePrinter.h.
References ActivationLayerInfo::ABS, ARM_COMPUTE_ERROR, ActivationLayerInfo::BOUNDED_RELU, ActivationLayerInfo::LEAKY_RELU, ActivationLayerInfo::LINEAR, ActivationLayerInfo::LOGISTIC, ActivationLayerInfo::LU_BOUNDED_RELU, ActivationLayerInfo::RELU, ActivationLayerInfo::SOFT_RELU, ActivationLayerInfo::SQRT, ActivationLayerInfo::SQUARE, and ActivationLayerInfo::TANH.
inline ::std::ostream& arm_compute::operator<< | ( | ::std::ostream & | os, |
const NormType & | norm_type | ||
) |
Formatted output of the NormType type.
[out] | os | Output stream. |
[in] | norm_type | Type to output. |
Definition at line 372 of file TypePrinter.h.
References ARM_COMPUTE_ERROR, CROSS_MAP, IN_MAP_1D, and IN_MAP_2D.
inline ::std::ostream& arm_compute::operator<< | ( | ::std::ostream & | os, |
const NormalizationLayerInfo & | info | ||
) |
Formatted output of NormalizationLayerInfo.
[out] | os | Output stream. |
[in] | info | Type to output. |
Definition at line 412 of file TypePrinter.h.
References NormalizationLayerInfo::norm_size(), and NormalizationLayerInfo::type().
inline ::std::ostream& arm_compute::operator<< | ( | ::std::ostream & | os, |
const PoolingType & | pool_type | ||
) |
Formatted output of the PoolingType type.
[out] | os | Output stream. |
[in] | pool_type | Type to output. |
Definition at line 425 of file TypePrinter.h.
References ARM_COMPUTE_ERROR, AVG, L2, and MAX.
inline ::std::ostream& arm_compute::operator<< | ( | ::std::ostream & | os, |
const PoolingLayerInfo & | info | ||
) |
Formatted output of PoolingLayerInfo.
[out] | os | Output stream. |
[in] | info | Type to output. |
Definition at line 452 of file TypePrinter.h.
References PoolingLayerInfo::pool_type().
inline ::std::ostream& arm_compute::operator<< | ( | ::std::ostream & | os, |
const DataLayout & | data_layout | ||
) |
Formatted output of the DataLayout type.
[out] | os | Output stream. |
[in] | data_layout | Type to output. |
Definition at line 479 of file TypePrinter.h.
References ARM_COMPUTE_ERROR, NCHW, NHWC, and UNKNOWN.
inline ::std::ostream& arm_compute::operator<< | ( | ::std::ostream & | os, |
const DataType & | data_type | ||
) |
Formatted output of the DataType type.
[out] | os | Output stream. |
[in] | data_type | Type to output. |
Definition at line 519 of file TypePrinter.h.
References ARM_COMPUTE_ERROR, F16, F32, F64, QASYMM8, QS16, QS8, S16, S32, S64, S8, SIZET, U16, U32, U64, U8, and UNKNOWN.
inline ::std::ostream& arm_compute::operator<< | ( | ::std::ostream & | os, |
const Format & | format | ||
) |
Formatted output of the Format type.
[out] | os | Output stream. |
[in] | format | Type to output. |
Definition at line 598 of file TypePrinter.h.
References ARM_COMPUTE_ERROR, F16, F32, IYUV, NV12, NV21, RGB888, RGBA8888, S16, S32, U16, U32, U8, UNKNOWN, UV88, UYVY422, YUV444, and YUYV422.
inline ::std::ostream& arm_compute::operator<< | ( | ::std::ostream & | os, |
const Channel & | channel | ||
) |
Formatted output of the Channel type.
[out] | os | Output stream. |
[in] | channel | Type to output. |
Definition at line 680 of file TypePrinter.h.
References A, ARM_COMPUTE_ERROR, B, C0, C1, C2, C3, G, R, U, UNKNOWN, V, and Y.
inline ::std::ostream& arm_compute::operator<< | ( | ::std::ostream & | os, |
const BorderMode & | mode | ||
) |
Formatted output of the BorderMode type.
[out] | os | Output stream. |
[in] | mode | Type to output. |
Definition at line 747 of file TypePrinter.h.
References ARM_COMPUTE_ERROR, CONSTANT, REPLICATE, and UNDEFINED.
inline ::std::ostream& arm_compute::operator<< | ( | ::std::ostream & | os, |
const BorderSize & | border | ||
) |
Formatted output of the BorderSize type.
[out] | os | Output stream. |
[in] | border | Type to output. |
Definition at line 774 of file TypePrinter.h.
References BorderSize::bottom, BorderSize::left, BorderSize::right, and BorderSize::top.
inline ::std::ostream& arm_compute::operator<< | ( | ::std::ostream & | os, |
const InterpolationPolicy & | policy | ||
) |
Formatted output of the InterpolationPolicy type.
[out] | os | Output stream. |
[in] | policy | Type to output. |
Definition at line 791 of file TypePrinter.h.
References AREA, ARM_COMPUTE_ERROR, BILINEAR, and NEAREST_NEIGHBOR.
inline ::std::ostream& arm_compute::operator<< | ( | ::std::ostream & | os, |
const SamplingPolicy & | policy | ||
) |
Formatted output of the SamplingPolicy type.
[out] | os | Output stream. |
[in] | policy | Type to output. |
Definition at line 818 of file TypePrinter.h.
References ARM_COMPUTE_ERROR, CENTER, and TOP_LEFT.
inline ::std::ostream& arm_compute::operator<< | ( | ::std::ostream & | os, |
const Rectangle & | rect | ||
) |
Formatted output of the Rectangle type.
[out] | os | Output stream. |
[in] | rect | Type to output. |
Definition at line 911 of file TypePrinter.h.
References Rectangle::height, Rectangle::width, Rectangle::x, and Rectangle::y.
inline ::std::ostream& arm_compute::operator<< | ( | ::std::ostream & | os, |
const PadStrideInfo & | pad_stride_info | ||
) |
Formatted output of the PadStrideInfo type.
[out] | os | Output stream. |
[in] | pad_stride_info | Type to output. |
Definition at line 926 of file TypePrinter.h.
References PadStrideInfo::pad_bottom(), PadStrideInfo::pad_left(), PadStrideInfo::pad_right(), PadStrideInfo::pad_top(), and PadStrideInfo::stride().
inline ::std::ostream& arm_compute::operator<< | ( | ::std::ostream & | os, |
const ConvertPolicy & | policy | ||
) |
Formatted output of the ConvertPolicy type.
[out] | os | Output stream. |
[in] | policy | Type to output. |
Definition at line 1008 of file TypePrinter.h.
References ARM_COMPUTE_ERROR, SATURATE, and WRAP.
inline ::std::ostream& arm_compute::operator<< | ( | ::std::ostream & | os, |
const ReductionOperation & | op | ||
) |
Formatted output of the Reduction Operations.
[out] | os | Output stream. |
[in] | op | Type to output. |
Definition at line 1039 of file TypePrinter.h.
References ARM_COMPUTE_ERROR, and SUM_SQUARE.
inline ::std::ostream& arm_compute::operator<< | ( | ::std::ostream & | os, |
const KeyPoint & | point | ||
) |
Formatted output of the KeyPoint type.
[out] | os | Output stream |
[in] | point | Type to output. |
Definition at line 1120 of file TypePrinter.h.
References KeyPoint::error, KeyPoint::orientation, KeyPoint::scale, KeyPoint::strength, KeyPoint::tracking_status, KeyPoint::x, and KeyPoint::y.
inline ::std::ostream& arm_compute::operator<< | ( | ::std::ostream & | os, |
const PhaseType & | phase_type | ||
) |
Formatted output of the PhaseType type.
[out] | os | Output stream |
[in] | phase_type | Type to output. |
Definition at line 1140 of file TypePrinter.h.
References ARM_COMPUTE_ERROR, SIGNED, and UNSIGNED.
inline ::std::ostream& arm_compute::operator<< | ( | ::std::ostream & | os, |
const MagnitudeType & | magnitude_type | ||
) |
Formatted output of the MagnitudeType type.
[out] | os | Output stream |
[in] | magnitude_type | Type to output. |
Definition at line 1177 of file TypePrinter.h.
References ARM_COMPUTE_ERROR, L1NORM, and L2NORM.
inline ::std::ostream& arm_compute::operator<< | ( | ::std::ostream & | os, |
const GradientDimension & | dim | ||
) |
Formatted output of the GradientDimension type.
[out] | os | Output stream |
[in] | dim | Type to output |
Definition at line 1214 of file TypePrinter.h.
References ARM_COMPUTE_ERROR, and GRAD_XY.
inline ::std::ostream& arm_compute::operator<< | ( | ::std::ostream & | os, |
const HOGNormType & | norm_type | ||
) |
Formatted output of the HOGNormType type.
[out] | os | Output stream |
[in] | norm_type | Type to output |
Definition at line 1254 of file TypePrinter.h.
References ARM_COMPUTE_ERROR, L1_NORM, L2_NORM, and L2HYS_NORM.
inline ::std::ostream& arm_compute::operator<< | ( | ::std::ostream & | os, |
const Size2D & | size | ||
) |
Formatted output of the Size2D type.
[out] | os | Output stream |
[in] | size | Type to output |
Definition at line 1294 of file TypePrinter.h.
References Size2D::height, and Size2D::width.
inline ::std::ostream& arm_compute::operator<< | ( | ::std::ostream & | os, |
const HOGInfo & | hog_info | ||
) |
Formatted output of the HOGInfo type.
[out] | os | Output stream |
[in] | hog_info | Type to output |
Definition at line 1321 of file TypePrinter.h.
References HOGInfo::block_size(), HOGInfo::block_stride(), HOGInfo::cell_size(), HOGInfo::detection_window_size(), HOGInfo::l2_hyst_threshold(), HOGInfo::normalization_type(), HOGInfo::num_bins(), and HOGInfo::phase_type().
inline ::std::ostream& arm_compute::operator<< | ( | ::std::ostream & | os, |
const ConvolutionMethod & | conv_method | ||
) |
Formatted output of the ConvolutionMethod type.
[out] | os | Output stream |
[in] | conv_method | Type to output |
Definition at line 1355 of file TypePrinter.h.
References ARM_COMPUTE_ERROR, DIRECT, GEMM, and WINOGRAD.
inline ::std::ostream& arm_compute::operator<< | ( | ::std::ostream & | os, |
const GPUTarget & | gpu_target | ||
) |
Formatted output of the GPUTarget type.
[out] | os | Output stream |
[in] | gpu_target | Type to output |
Definition at line 1395 of file TypePrinter.h.
References ARM_COMPUTE_ERROR, BIFROST, G51, G51BIG, G51LIT, G71, G72, GPU_ARCH_MASK, MIDGARD, T600, T700, T800, TBOX, TNOX, and TTRX.
inline ::std::ostream& arm_compute::operator<< | ( | ::std::ostream & | os, |
const DetectionWindow & | detection_window | ||
) |
Formatted output of the DetectionWindow type.
[out] | os | Output stream |
[in] | detection_window | Type to output |
Definition at line 1468 of file TypePrinter.h.
References DetectionWindow::height, DetectionWindow::idx_class, DetectionWindow::score, DetectionWindow::width, DetectionWindow::x, and DetectionWindow::y.
inline ::std::ostream& arm_compute::operator<< | ( | ::std::ostream & | os, |
const Termination & | termination | ||
) |
Formatted output of the Termination type.
[out] | os | Output stream |
[in] | termination | Type to output |
Definition at line 1500 of file TypePrinter.h.
References ARM_COMPUTE_ERROR, TERM_CRITERIA_BOTH, TERM_CRITERIA_EPSILON, and TERM_CRITERIA_ITERATIONS.
inline ::std::ostream& arm_compute::operator<< | ( | ::std::ostream & | os, |
const WinogradInfo & | info | ||
) |
Formatted output of the WinogradInfo type.
Definition at line 1534 of file TypePrinter.h.
References WinogradInfo::convolution_info, WinogradInfo::kernel_size, WinogradInfo::output_data_layout, and WinogradInfo::output_tile_size.
|
inline |
Check that given dimensions are equal.
[in] | lhs | Left-hand side Dimensions. |
[in] | rhs | Right-hand side Dimensions. |
Definition at line 234 of file Dimensions.h.
References Dimensions< T >::cbegin(), Dimensions< T >::cend(), and Dimensions< T >::num_dimensions().
inline ::std::istream& arm_compute::operator>> | ( | ::std::istream & | is, |
BorderMode & | mode | ||
) |
Formatted input of the BorderMode type.
[out] | is | Input stream. |
[in] | mode | Border mode. |
Definition at line 42 of file TypeReader.h.
References arm_compute::test::validation::c, CONSTANT, REPLICATE, and UNDEFINED.
|
inline |
Permutes given Dimensions according to a permutation vector.
[in,out] | dimensions | Dimensions to permute |
[in] | perm | Permutation vector |
Definition at line 536 of file Helpers.h.
References Dimensions< T >::begin(), Dimensions< T >::end(), Dimensions< T >::num_dimensions(), and Dimensions< T >::set().
Referenced by NumPyBinLoader::access_tensor(), arm_compute::misc::shape_calculator::compute_permutation_output_shape(), arm_compute::test::validation::DATA_TEST_CASE(), AssetsLibrary::fill(), arm_compute::test::validation::validate(), and arm_compute::test::validation::validate_wrap().
|
inline |
Permutes given TensorShape according to a permutation vector.
[in,out] | shape | Shape to permute |
[in] | perm | Permutation vector |
Definition at line 553 of file Helpers.h.
References auto_init_if_empty(), calculate_valid_region_scale(), coords2index(), arm_compute::test::validation::data_type, get_data_layout_dimension_index(), index2coords(), Dimensions< T >::num_dimensions(), TensorShape::set(), set_data_layout_if_unknown(), set_data_type_if_unknown(), set_fixed_point_position_if_zero(), set_format_if_unknown(), set_quantization_info_if_empty(), set_shape_if_empty(), and arm_compute::test::validation::shape.
|
inline |
Return the pixel at (x,y) using area interpolation by clamping when out of borders.
The image must be single channel U8
[in] | first_pixel_ptr | Pointer to the first pixel of a single channel U8 image. |
[in] | stride | Stride in bytes of the image |
[in] | width | Width of the image |
[in] | height | Height of the image |
[in] | wr | Width ratio among the input image width and output image width. |
[in] | hr | Height ratio among the input image height and output image height. |
[in] | x | X position of the wanted pixel |
[in] | y | Y position of the wanted pixel |
Definition at line 32 of file Helpers.inl.
References accumulate(), ARM_COMPUTE_ERROR_ON, arm_compute::utility::for_each(), arm_compute::test::fixed_point_arithmetic::detail::max(), arm_compute::test::fixed_point_arithmetic::detail::min(), Window::set(), and sum().
Referenced by pixel_bilinear_c1_clamp().
|
inline |
Return the pixel at (x,y) using bilinear interpolation.
[in] | first_pixel_ptr | Pointer to the first pixel of a single channel input. |
[in] | stride | Stride in bytes of the image; |
[in] | x | X position of the wanted pixel |
[in] | y | Y position of the wanted pixel |
Definition at line 210 of file Helpers.h.
References ARM_COMPUTE_ERROR_ON, and delta_bilinear_c1().
|
inline |
Return the pixel at (x,y) using bilinear interpolation by clamping when out of borders.
The image must be single channel input
[in] | first_pixel_ptr | Pointer to the first pixel of a single channel image. |
[in] | stride | Stride in bytes of the image |
[in] | width | Width of the image |
[in] | height | Height of the image |
[in] | x | X position of the wanted pixel |
[in] | y | Y position of the wanted pixel |
Definition at line 237 of file Helpers.h.
References ARM_COMPUTE_ERROR_ON, delta_bilinear_c1(), delta_linear_c1_x(), delta_linear_c1_y(), arm_compute::test::fixed_point_arithmetic::detail::max(), arm_compute::test::fixed_point_arithmetic::detail::min(), and pixel_area_c1u8_clamp().
|
inline |
The size in bytes of the pixel format.
[in] | format | Input format |
Definition at line 144 of file Utils.h.
References ARM_COMPUTE_ERROR, F16, F32, IYUV, NV12, NV21, RGB888, RGBA8888, S16, S32, U16, U32, U8, UV88, UYVY422, YUV444, and YUYV422.
Return the plane index of a given channel given an input format.
[in] | format | Input format |
[in] | channel | Input channel |
Definition at line 254 of file Utils.h.
References ARM_COMPUTE_ERROR, F16, F32, IYUV, NV12, NV21, RGB888, RGBA8888, S16, S32, U, U16, U32, U8, UV88, UYVY422, V, Y, YUV444, and YUYV422.
Referenced by arm_compute::test::validation::reference::channel_extract().
void arm_compute::print_consecutive_elements | ( | std::ostream & | s, |
DataType | dt, | ||
const uint8_t * | ptr, | ||
unsigned int | n, | ||
int | stream_width, | ||
const std::string & | element_delim = " " |
||
) |
Print consecutive elements to an output stream.
[out] | s | Output stream to print the elements to. |
[in] | dt | Data type of the elements |
[in] | ptr | Pointer to print the elements from. |
[in] | n | Number of elements to print. |
[in] | stream_width | (Optional) Width of the stream. If set to 0 the element's width is used. Defaults to 0. |
[in] | element_delim | (Optional) Delimeter among the consecutive elements. Defaults to space delimeter |
Referenced by max_consecutive_elements_display_width_impl().
void arm_compute::print_consecutive_elements_impl | ( | std::ostream & | s, |
const T * | ptr, | ||
unsigned int | n, | ||
int | stream_width = 0 , |
||
const std::string & | element_delim = " " |
||
) |
Print consecutive elements to an output stream.
[out] | s | Output stream to print the elements to. |
[in] | ptr | Pointer to print the elements from. |
[in] | n | Number of elements to print. |
[in] | stream_width | (Optional) Width of the stream. If set to 0 the element's width is used. Defaults to 0. |
[in] | element_delim | (Optional) Delimeter among the consecutive elements. Defaults to space delimeter |
std::string arm_compute::read_file | ( | const std::string & | filename, |
bool | binary | ||
) |
Load an entire file in memory.
[in] | filename | Name of the file to read. |
[in] | binary | Is it a binary file ? |
Referenced by floor_to_multiple().
int arm_compute::round | ( | float | x, |
RoundingPolicy | rounding_policy | ||
) |
Return a rounded value of x.
Rounding is done according to the rounding_policy.
[in] | x | Float value to be rounded. |
[in] | rounding_policy | Policy determining how rounding is done. |
Referenced by DATA_TEST_CASE(), finalize(), lktracker_stage0(), lktracker_stage1(), pooling_layer_MxN_quantized_nchw(), pooling_layer_MxN_quantized_nhwc(), and roi_pooling_layer().
|
inline |
Round to the nearest division by a power-of-two using exponent.
[in] | x | Vector of 4 elements |
[in] | exponent | Integer value used to round to nearest division by a power-of-two |
Definition at line 26 of file NEAsymm.inl.
Referenced by finalize_quantization().
16 bit fixed point scalar absolute value
[in] | a | 16 bit fixed point input |
Definition at line 67 of file FixedPoint.inl.
References arm_compute::test::validation::a, arm_compute::test::fixed_point_arithmetic::detail::max(), and arm_compute::test::fixed_point_arithmetic::detail::min().
Referenced by sqexp_qs16().
8 bit fixed point scalar absolute value
[in] | a | 8 bit fixed point input |
Definition at line 62 of file FixedPoint.inl.
References arm_compute::test::validation::a, arm_compute::test::fixed_point_arithmetic::detail::max(), and arm_compute::test::fixed_point_arithmetic::detail::min().
Referenced by sqexp_qs8().
16 bit fixed point scalar add
[in] | a | First 16 bit fixed point input |
[in] | b | Second 16 bit fixed point input |
Definition at line 77 of file FixedPoint.inl.
References arm_compute::test::validation::b.
Referenced by slog_qs16().
8 bit fixed point scalar add
[in] | a | First 8 bit fixed point input |
[in] | b | Second 8 bit fixed point input |
Definition at line 72 of file FixedPoint.inl.
References arm_compute::test::validation::b.
Referenced by slog_qs8().
const std::pair<unsigned int, unsigned int> arm_compute::scaled_dimensions | ( | unsigned int | width, |
unsigned int | height, | ||
unsigned int | kernel_width, | ||
unsigned int | kernel_height, | ||
const PadStrideInfo & | pad_stride_info, | ||
const Size2D & | dilation = Size2D(1U, 1U) |
||
) |
Returns expected width and height of output scaled tensor depending on dimensions rounding mode.
[in] | width | Width of input tensor (Number of columns) |
[in] | height | Height of input tensor (Number of rows) |
[in] | kernel_width | Kernel width. |
[in] | kernel_height | Kernel height. |
[in] | pad_stride_info | Pad and stride information. |
[in] | dilation | (Optional) Dilation, in elements, across x and y. Defaults to (1, 1). |
Referenced by arm_compute::misc::shape_calculator::compute_deep_convolution_shape(), arm_compute::misc::shape_calculator::compute_depthwise_convolution_shape(), arm_compute::misc::shape_calculator::compute_im2col_conv_shape(), arm_compute::misc::shape_calculator::compute_pool_shape(), arm_compute::misc::shape_calculator::compute_winograd_output_transform_shape(), arm_compute::test::validation::reference::convolution_layer_nchw(), data_type_for_convolution_matrix(), and arm_compute::test::validation::reference::locally_connected().
|
inline |
Convert a 16 bit fixed point to float.
[in] | a | Input to convert |
[in] | fixed_point_position | Fixed point position that expresses the number of bits for the fractional part of the number |
Definition at line 384 of file FixedPoint.inl.
References arm_compute::test::validation::a.
|
inline |
Convert an 8 bit fixed point to float.
[in] | a | Input to convert |
[in] | fixed_point_position | Fixed point position that expresses the number of bits for the fractional part of the number |
Definition at line 373 of file FixedPoint.inl.
References arm_compute::test::validation::a.
16 bit fixed point scalar division
[in] | a | First 16 bit fixed point input |
[in] | b | Second 16 bit fixed point input |
[in] | fixed_point_position | Fixed point position that expresses the number of bits for the fractional part of the number |
Definition at line 255 of file FixedPoint.inl.
References arm_compute::test::validation::b.
Referenced by slog_qs16().
8 bit fixed point scalar division
[in] | a | First 8 bit fixed point input |
[in] | b | Second 8 bit fixed point input |
[in] | fixed_point_position | Fixed point position that expresses the number of bits for the fractional part of the number |
Definition at line 249 of file FixedPoint.inl.
References arm_compute::test::validation::b.
Referenced by slog_qs8().
|
inline |
Separate a 2D convolution into two 1D convolutions.
[in] | conv | 2D convolution |
[out] | conv_col | 1D vertical convolution |
[out] | conv_row | 1D horizontal convolution |
[in] | size | Size of the 2D convolution |
Definition at line 577 of file Utils.h.
References arm_compute::test::fixed_point_arithmetic::detail::abs().
|
inline |
Set the data layout to the specified value if the current data layout is unknown.
[in,out] | info | Tensor info used to check and assign. |
[in] | data_layout | New data layout. |
Definition at line 270 of file Helpers.inl.
References ITensorInfo::data_layout(), ITensorInfo::set_data_layout(), and UNKNOWN.
Referenced by permute().
|
inline |
Set the data type and number of channels to the specified value if the current data type is unknown.
[in,out] | info | Tensor info used to check and assign. |
[in] | data_type | New data type. |
Definition at line 259 of file Helpers.inl.
References ITensorInfo::data_type(), ITensorInfo::set_data_type(), and UNKNOWN.
Referenced by permute().
|
inline |
Set the fixed point position to the specified value if the current fixed point position is 0 and the data type is QS8 or QS16.
[in,out] | info | Tensor info used to check and assign. |
[in] | fixed_point_position | New fixed point position |
Definition at line 281 of file Helpers.inl.
References ITensorInfo::data_type(), ITensorInfo::fixed_point_position(), QS16, QS8, and ITensorInfo::set_fixed_point_position().
Referenced by permute().
|
inline |
Set the format, data type and number of channels to the specified value if the current data type is unknown.
[in,out] | info | Tensor info used to check and assign. |
[in] | format | New format. |
Definition at line 248 of file Helpers.inl.
References ITensorInfo::data_type(), ITensorInfo::set_format(), and UNKNOWN.
Referenced by permute().
|
inline |
Set the quantization info to the specified value if the current quantization info is empty and the data type of asymmetric quantized type.
[in,out] | info | Tensor info used to check and assign. |
[in] | quantization_info | Quantization info |
Definition at line 292 of file Helpers.inl.
References ITensorInfo::data_type(), QuantizationInfo::empty(), is_data_type_quantized_asymmetric(), ITensorInfo::quantization_info(), and ITensorInfo::set_quantization_info().
Referenced by permute().
|
inline |
Set the shape to the specified value if the current assignment is empty.
[in,out] | info | Tensor info used to check and assign. |
[in] | shape | New shape. |
Definition at line 237 of file Helpers.inl.
References ITensorInfo::set_tensor_shape(), ITensorInfo::tensor_shape(), and TensorShape::total_size().
Referenced by permute().
|
inline |
Create a wrapper kernel.
[in] | a | Input tensor A. |
[in] | b | Input tensor B. |
[out] | d | Output tensor. |
[in] | alpha | Alpha value. |
[in] | beta | Beta value. |
[in] | pretranspose_hint | Pre-transpose hint in case matrix b should be pre-transposed |
[out] | workspace | Workspace tensor |
[out] | B_pretranspose | Tensor to hold the pre-transposed B |
[in] | memory_group | Tensor memory group. |
[out] | asm_glue | Assembly glue kernel. |
Definition at line 159 of file AssemblyHelper.h.
References arm_compute::test::validation::a, allocate_workspace(), ARM_COMPUTE_ERROR_ON_NULLPTR, arm_compute::test::validation::b, Tensor::buffer(), IScheduler::cpu_info(), Scheduler::get(), ITensor::info(), IScheduler::num_threads(), ITensorInfo::tensor_shape(), TensorShape::total_size_upper(), Dimensions< T >::x(), Dimensions< T >::y(), and Dimensions< T >::z().
16 bit fixed point scalar exponential
[in] | a | 16 bit fixed point input |
[in] | fixed_point_position | Fixed point position that expresses the number of bits for the fractional part of the number |
16 bit fixed point scalar inverse square root
[in] | a | 16 bit fixed point input |
[in] | fixed_point_position | Fixed point position that expresses the number of bits for the fractional part of the number |
Definition at line 229 of file FixedPoint.inl.
References smul_qs16(), and ssub_qs16().
8 bit fixed point scalar inverse square root
[in] | a | 8 bit fixed point input |
[in] | fixed_point_position | Fixed point position that expresses the number of bits for the fractional part of the number |
Definition at line 209 of file FixedPoint.inl.
References smul_qs8(), and ssub_qs8().
16 bit fixed point scalar logarithm
[in] | a | 16 bit fixed point input |
[in] | fixed_point_position | Fixed point position that expresses the number of bits for the fractional part of the number |
Definition at line 340 of file FixedPoint.inl.
References A, B, sadd_qs16(), sdiv_qs16(), smul_qs16(), sqadd_qs16(), sqmul_qs16(), ssub_qs16(), and sum().
8 bit fixed point scalar logarithm
[in] | a | 8 bit fixed point input |
[in] | fixed_point_position | Fixed point position that expresses the number of bits for the fractional part of the number |
Definition at line 307 of file FixedPoint.inl.
References A, B, sadd_qs8(), sdiv_qs8(), smul_qs8(), sqadd_qs8(), sqmul_qs8(), ssub_qs8(), and sum().
16 bit fixed point scalar multiply
[in] | a | First 16 bit fixed point input |
[in] | b | Second 16 bit fixed point input |
[in] | fixed_point_position | Fixed point position that expresses the number of bits for the fractional part of the number |
Definition at line 149 of file FixedPoint.inl.
References arm_compute::test::validation::a.
Referenced by sinvsqrt_qs16(), and slog_qs16().
8 bit fixed point scalar multiply
[in] | a | First 8 bit fixed point input |
[in] | b | Second 8 bit fixed point input |
[in] | fixed_point_position | Fixed point position that expresses the number of bits for the fractional part of the number |
Definition at line 137 of file FixedPoint.inl.
References arm_compute::test::validation::a.
Referenced by sinvsqrt_qs8(), and slog_qs8().
16 bit fixed point scalar saturating add
[in] | a | First 16 bit fixed point input |
[in] | b | Second 16 bit fixed point input |
Definition at line 91 of file FixedPoint.inl.
References arm_compute::test::validation::a, and arm_compute::utility::saturate_cast().
Referenced by slog_qs16(), sqexp_qs16(), and sshr_qs16().
32 bit fixed point scalar saturating add
[in] | a | First 32 bit fixed point input |
[in] | b | Second 32 bit fixed point input |
Definition at line 100 of file FixedPoint.inl.
References arm_compute::test::validation::a, and arm_compute::utility::saturate_cast().
8 bit fixed point scalar saturating add
[in] | a | First 8 bit fixed point input |
[in] | b | Second 8 bit fixed point input |
Definition at line 82 of file FixedPoint.inl.
References arm_compute::test::validation::a, and arm_compute::utility::saturate_cast().
Referenced by slog_qs8(), sqexp_qs8(), and sshr_qs8().
|
inline |
Convert a float to 16 bit fixed point.
[in] | a | Input to convert |
[in] | fixed_point_position | Fixed point position that expresses the number of bits for the fractional part of the number |
Definition at line 389 of file FixedPoint.inl.
References arm_compute::utility::saturate_cast().
|
inline |
Convert a float to 8 bit fixed point.
[in] | a | Input to convert |
[in] | fixed_point_position | Fixed point position that expresses the number of bits for the fractional part of the number |
Definition at line 378 of file FixedPoint.inl.
References arm_compute::utility::saturate_cast().
16 bit fixed point scalar exponential
[in] | a | 16 bit fixed point input |
[in] | fixed_point_position | Fixed point position that expresses the number of bits for the fractional part of the number |
Definition at line 284 of file FixedPoint.inl.
References A, arm_compute::test::validation::alpha, B, sabs_qs16(), sqadd_qs16(), sqmul_qs16(), sqshl_qs16(), sqsub_qs16(), and sum().
8 bit fixed point scalar exponential
[in] | a | 8 bit fixed point input |
[in] | fixed_point_position | Fixed point position that expresses the number of bits for the fractional part of the number |
Definition at line 261 of file FixedPoint.inl.
References A, arm_compute::test::validation::alpha, B, sabs_qs8(), sqadd_qs8(), sqmul_qs8(), sqshl_qs8(), sqsub_qs8(), and sum().
Scalar saturating move and narrow.
[in] | a | Input to convert to 8 bit fixed point |
Definition at line 395 of file FixedPoint.inl.
References arm_compute::test::validation::a, and arm_compute::utility::saturate_cast().
Scalar saturating move and narrow.
[in] | a | Input to convert to 16 bit fixed point |
Definition at line 401 of file FixedPoint.inl.
References arm_compute::test::validation::a, and arm_compute::utility::saturate_cast().
16 bit fixed point scalar saturating multiply
[in] | a | First 16 bit fixed point input |
[in] | b | Second 16 bit fixed point input |
[in] | fixed_point_position | Fixed point position that expresses the number of bits for the fractional part of the number |
Definition at line 173 of file FixedPoint.inl.
References arm_compute::test::validation::a, and arm_compute::utility::saturate_cast().
Referenced by slog_qs16(), and sqexp_qs16().
8 bit fixed point scalar saturating multiply
[in] | a | First 8 bit fixed point input |
[in] | b | Second 8 bit fixed point input |
[in] | fixed_point_position | Fixed point position that expresses the number of bits for the fractional part of the number |
Definition at line 161 of file FixedPoint.inl.
References arm_compute::test::validation::a, and arm_compute::utility::saturate_cast().
Referenced by slog_qs8(), and sqexp_qs8().
16 bit fixed point scalar multiply long
[in] | a | First 16 bit fixed point input |
[in] | b | Second 16 bit fixed point input |
[in] | fixed_point_position | Fixed point position that expresses the number of bits for the fractional part of the number |
Definition at line 197 of file FixedPoint.inl.
References arm_compute::test::validation::a.
8 bit fixed point scalar multiply long
[in] | a | First 8 bit fixed point input |
[in] | b | Second 8 bit fixed point input |
[in] | fixed_point_position | Fixed point position that expresses the number of bits for the fractional part of the number |
Definition at line 185 of file FixedPoint.inl.
References arm_compute::test::validation::a.
16 bit fixed point scalar saturating shift left
[in] | a | First 16 bit fixed point input |
[in] | shift | Shift amount (positive only values) |
Definition at line 40 of file FixedPoint.inl.
References arm_compute::test::validation::a, and arm_compute::utility::saturate_cast().
Referenced by sqexp_qs16().
8 bit fixed point scalar saturating shift left
[in] | a | First 8 bit fixed point input |
[in] | shift | Shift amount (positive only values) |
Definition at line 32 of file FixedPoint.inl.
References arm_compute::test::validation::a, and arm_compute::utility::saturate_cast().
Referenced by sqexp_qs8().
16 bit fixed point scalar saturating subtraction
[in] | a | First 16 bit fixed point input |
[in] | b | Second 16 bit fixed point input |
Definition at line 128 of file FixedPoint.inl.
References arm_compute::test::validation::a, and arm_compute::utility::saturate_cast().
Referenced by sqexp_qs16().
8 bit fixed point scalar saturating subtraction
[in] | a | First 8 bit fixed point input |
[in] | b | Second 8 bit fixed point input |
Definition at line 119 of file FixedPoint.inl.
References arm_compute::test::validation::a, and arm_compute::utility::saturate_cast().
Referenced by sqexp_qs8().
16 bit fixed point scalar shift right
[in] | a | First 16 bit fixed point input |
[in] | shift | Shift amount (positive only values) |
Definition at line 55 of file FixedPoint.inl.
References ARM_COMPUTE_ERROR_ON_MSG, and sqadd_qs16().
8 bit fixed point scalar shift right
[in] | a | First 8 bit fixed point input |
[in] | shift | Shift amount (positive only values) |
Definition at line 48 of file FixedPoint.inl.
References ARM_COMPUTE_ERROR_ON_MSG, and sqadd_qs8().
16 bit fixed point scalar subtraction
[in] | a | First 16 bit fixed point input |
[in] | b | Second 16 bit fixed point input |
Definition at line 114 of file FixedPoint.inl.
References arm_compute::test::validation::b.
Referenced by sinvsqrt_qs16(), and slog_qs16().
8 bit fixed point scalar subtraction
[in] | a | First 8 bit fixed point input |
[in] | b | Second 8 bit fixed point input |
Definition at line 109 of file FixedPoint.inl.
References arm_compute::test::validation::b.
Referenced by sinvsqrt_qs8(), and slog_qs8().
const std::string& arm_compute::string_from_activation_func | ( | ActivationLayerInfo::ActivationFunction | act | ) |
Translates a given activation function to a string.
[in] | act | ActivationLayerInfo::ActivationFunction to be translated to string. |
Referenced by data_type_for_convolution_matrix().
const std::string& arm_compute::string_from_border_mode | ( | BorderMode | border_mode | ) |
Translates a given border mode policy to a string.
[in] | border_mode | BorderMode to be translated to string. |
Referenced by data_type_for_convolution_matrix().
const std::string& arm_compute::string_from_channel | ( | Channel | channel | ) |
Convert a channel identity into a string.
[in] | channel | Channel to be translated to string. |
Referenced by data_type_for_convolution_matrix().
const std::string& arm_compute::string_from_data_layout | ( | DataLayout | dl | ) |
Convert a data layout identity into a string.
[in] | dl | DataLayout to be translated to string. |
Referenced by data_type_for_convolution_matrix().
const std::string& arm_compute::string_from_data_type | ( | DataType | dt | ) |
Convert a data type identity into a string.
[in] | dt | DataType to be translated to string. |
Referenced by data_type_for_convolution_matrix(), error_on_data_type_not_in(), and error_on_value_not_representable_in_fixed_point().
const std::string& arm_compute::string_from_format | ( | Format | format | ) |
Convert a tensor format into a string.
[in] | format | Format to be translated to string. |
Referenced by data_type_for_convolution_matrix(), and error_on_format_not_in().
const std::string& arm_compute::string_from_interpolation_policy | ( | InterpolationPolicy | policy | ) |
Translates a given interpolation policy to a string.
[in] | policy | InterpolationPolicy to be translated to string. |
Referenced by data_type_for_convolution_matrix().
const std::string& arm_compute::string_from_matrix_pattern | ( | MatrixPattern | pattern | ) |
Convert a matrix pattern into a string.
[in] | pattern | MatrixPattern to be translated to string. |
Referenced by data_type_for_convolution_matrix().
const std::string& arm_compute::string_from_non_linear_filter_function | ( | NonLinearFilterFunction | function | ) |
Translates a given non linear function to a string.
[in] | function | NonLinearFilterFunction to be translated to string. |
Referenced by data_type_for_convolution_matrix().
const std::string& arm_compute::string_from_norm_type | ( | NormType | type | ) |
Translates a given normalization type to a string.
[in] | type | NormType to be translated to string. |
Referenced by data_type_for_convolution_matrix().
const std::string& arm_compute::string_from_pooling_type | ( | PoolingType | type | ) |
Translates a given pooling type to a string.
[in] | type | PoolingType to be translated to string. |
Referenced by data_type_for_convolution_matrix().
const std::string& arm_compute::string_from_scheduler_type | ( | Scheduler::Type | t | ) |
Convert a Scheduler::Type into a string.
[in] | t | Scheduler::Type to be translated to string. |
const std::string& arm_compute::string_from_target | ( | GPUTarget | target | ) |
Translates a given gpu device target to string.
[in] | target | Given gpu target. |
|
inline |
Formatted output of the NonLinearFilterFunction type.
[in] | function | Type to output. |
Definition at line 101 of file TypePrinter.h.
Referenced by arm_compute::test::validation::DATA_TEST_CASE(), and main().
|
inline |
Formatted output of the MatrixPattern type.
[in] | pattern | Type to output. |
Definition at line 144 of file TypePrinter.h.
|
inline |
Formatted output of the QuantizationInfo type.
[in] | quantization_info | Type to output. |
Definition at line 226 of file TypePrinter.h.
|
inline |
Formatted output of the FixedPointOp type.
[in] | op | Type to output. |
Definition at line 278 of file TypePrinter.h.
|
inline |
Formatted output of the activation function info type.
[in] | info | Type to output. |
Definition at line 342 of file TypePrinter.h.
References ActivationLayerInfo::activation(), and ActivationLayerInfo::enabled().
|
inline |
Formatted output of the activation function type.
[in] | function | Type to output. |
Definition at line 358 of file TypePrinter.h.
|
inline |
Formatted output of NormalizationLayerInfo.
[in] | info | Type to output. |
Definition at line 398 of file TypePrinter.h.
References NormalizationLayerInfo::norm_size(), and NormalizationLayerInfo::type().
|
inline |
Formatted output of RoundingPolicy.
[in] | rounding_policy | Type to output. |
Definition at line 465 of file TypePrinter.h.
References arm_compute::test::validation::rounding_policy.
|
inline |
Formatted output of the DataLayout type.
[in] | data_layout | Type to output. |
Definition at line 505 of file TypePrinter.h.
|
inline |
Formatted output of the DataType type.
[in] | data_type | Type to output. |
Definition at line 584 of file TypePrinter.h.
References arm_compute::test::validation::data_type.
|
inline |
Formatted output of the Format type.
[in] | format | Type to output. |
Definition at line 666 of file TypePrinter.h.
|
inline |
Formatted output of the Channel type.
[in] | channel | Type to output. |
Definition at line 733 of file TypePrinter.h.
|
inline |
Formatted output of the TensorInfo type.
[in] | info | Type to output. |
Definition at line 841 of file TypePrinter.h.
References TensorInfo::data_type(), TensorInfo::fixed_point_position(), TensorInfo::num_channels(), and TensorInfo::tensor_shape().
|
inline |
Formatted output of the Dimensions type.
[in] | dimensions | Type to output. |
Definition at line 858 of file TypePrinter.h.
|
inline |
Formatted output of the Strides type.
[in] | stride | Type to output. |
Definition at line 871 of file TypePrinter.h.
|
inline |
Formatted output of the TensorShape type.
[in] | shape | Type to output. |
Definition at line 884 of file TypePrinter.h.
References arm_compute::test::validation::shape.
|
inline |
Formatted output of the Coordinates type.
[in] | coord | Type to output. |
Definition at line 897 of file TypePrinter.h.
|
inline |
Formatted output of the PadStrideInfo type.
[in] | pad_stride_info | Type to output. |
Definition at line 942 of file TypePrinter.h.
|
inline |
Formatted output of the BorderMode type.
[in] | mode | Type to output. |
Definition at line 955 of file TypePrinter.h.
|
inline |
Formatted output of the BorderSize type.
[in] | border | Type to output. |
Definition at line 968 of file TypePrinter.h.
|
inline |
Formatted output of the InterpolationPolicy type.
[in] | policy | Type to output. |
Definition at line 981 of file TypePrinter.h.
|
inline |
Formatted output of the SamplingPolicy type.
[in] | policy | Type to output. |
Definition at line 994 of file TypePrinter.h.
|
inline |
Definition at line 1025 of file TypePrinter.h.
|
inline |
Formatted output of the Reduction Operations.
[in] | op | Type to output. |
Definition at line 1059 of file TypePrinter.h.
|
inline |
Formatted output of the Norm Type.
[in] | type | Type to output. |
Definition at line 1072 of file TypePrinter.h.
|
inline |
Formatted output of the Pooling Type.
[in] | type | Type to output. |
Definition at line 1085 of file TypePrinter.h.
|
inline |
Formatted output of the Pooling Layer Info.
[in] | info | Type to output. |
Definition at line 1098 of file TypePrinter.h.
References Size2D::height, PoolingLayerInfo::is_global_pooling(), PoolingLayerInfo::pad_stride_info(), PoolingLayerInfo::pool_size(), PoolingLayerInfo::pool_type(), and Size2D::width.
|
inline |
Formatted output of the PhaseType type.
[in] | type | Type to output. |
Definition at line 1163 of file TypePrinter.h.
|
inline |
Formatted output of the MagnitudeType type.
[in] | type | Type to output. |
Definition at line 1200 of file TypePrinter.h.
|
inline |
Formatted output of the GradientDimension type.
[in] | type | Type to output |
Definition at line 1240 of file TypePrinter.h.
|
inline |
Formatted output of the HOGNormType type.
[in] | type | Type to output |
Definition at line 1280 of file TypePrinter.h.
|
inline |
Formatted output of the Size2D type.
[in] | type | Type to output |
Definition at line 1307 of file TypePrinter.h.
|
inline |
Formatted output of the HOGInfo type.
[in] | type | Type to output |
Definition at line 1341 of file TypePrinter.h.
|
inline |
Formatted output of the ConvolutionMethod type.
[in] | conv_method | Type to output |
Definition at line 1381 of file TypePrinter.h.
|
inline |
Formatted output of the GPUTarget type.
[in] | gpu_target | Type to output |
Definition at line 1454 of file TypePrinter.h.
|
inline |
Formatted output of the DetectionWindow type.
[in] | detection_window | Type to output |
Definition at line 1486 of file TypePrinter.h.
|
inline |
Formatted output of the Termination type.
[in] | termination | Type to output |
Definition at line 1526 of file TypePrinter.h.
|
inline |
Definition at line 1544 of file TypePrinter.h.
bool arm_compute::update_window_and_padding | ( | Window & | win, |
Ts &&... | patterns | ||
) |
Update window and padding size for each of the access patterns.
First the window size is reduced based on all access patterns that are not allowed to modify the padding of the underlying tensor. Then the padding of the remaining tensors is increased to match the window.
[in] | win | Window that is used by the kernel. |
[in] | patterns | Access patterns used to calculate the final window and padding. |
Definition at line 368 of file Helpers.h.
References calculate_max_window(), arm_compute::utility::for_each(), IAccessWindow::update_padding_if_needed(), and IAccessWindow::update_window_if_needed().
qint16x4_t arm_compute::vabs_qs16 | ( | qint16x4_t | a | ) |
Absolute value of 16 bit fixed point vector (4 elements)
[in] | a | 16 bit fixed point input vector |
Absolute value of 8 bit fixed point vector (8 elements)
[in] | a | 8 bit fixed point input vector |
qint16x8_t arm_compute::vabsq_qs16 | ( | qint16x8_t | a | ) |
Absolute value of 16 bit fixed point vector (8 elements)
[in] | a | 16 bit fixed point input vector |
qint8x16_t arm_compute::vabsq_qs8 | ( | qint8x16_t | a | ) |
Absolute value of 8 bit fixed point vector (16 elements)
[in] | a | 8 bit fixed point input vector |
qint16x4_t arm_compute::vadd_qs16 | ( | qint16x4_t | a, |
qint16x4_t | b | ||
) |
16 bit fixed point vector add (4 elements)
[in] | a | First 16 bit fixed point input vector |
[in] | b | Second 16 bit fixed point input vector |
8 bit fixed point vector add (8 elements)
[in] | a | First 8 bit fixed point input vector |
[in] | b | Second 8 bit fixed point input vector |
qint16x8_t arm_compute::vaddq_qs16 | ( | qint16x8_t | a, |
qint16x8_t | b | ||
) |
16 bit fixed point vector add (8 elements)
[in] | a | First 16 bit fixed point input vector |
[in] | b | Second 16 bit fixed point input vector |
qint8x16_t arm_compute::vaddq_qs8 | ( | qint8x16_t | a, |
qint8x16_t | b | ||
) |
8 bit fixed point vector add (16 elements)
[in] | a | First 8 bit fixed point input vector |
[in] | b | Second 8 bit fixed point input vector |
float32x4_t arm_compute::vcvt_f32_qs16 | ( | qint16x4_t | a, |
int | fixed_point_position | ||
) |
Convert a 16 bit fixed point vector with 4 elements to a float vector with 4 elements.
[in] | a | 16 bit fixed point input vector |
[in] | fixed_point_position | Fixed point position that expresses the number of bits for the fractional part of the number |
float32x4x2_t arm_compute::vcvt_f32_qs8 | ( | qint8x8_t | a, |
int | fixed_point_position | ||
) |
Convert a 8 bit fixed point vector with 8 elements to a float vector with 4x2 elements.
[in] | a | 8 bit fixed point input vector |
[in] | fixed_point_position | Fixed point position that expresses the number of bits for the fractional part of the number |
float32x4x2_t arm_compute::vcvtq_qs16_f32 | ( | qint16x8_t | a, |
int | fixed_point_position | ||
) |
Convert a 16 bit fixed point vector with 8 elements to a float vector with 4x2 elements.
[in] | a | 16 bit fixed point input vector |
[in] | fixed_point_position | Fixed point position that expresses the number of bits for the fractional part of the number |
float32x4x4_t arm_compute::vcvtq_qs8_f32 | ( | qint8x16_t | a, |
int | fixed_point_position | ||
) |
Convert a 8 bit fixed point vector with 16 elements to a float vector with 4x4 elements.
[in] | a | 8 bit fixed point input vector |
[in] | fixed_point_position | Fixed point position that expresses the number of bits for the fractional part of the number |
qint16x4_t arm_compute::vdiv_qs16 | ( | qint16x4_t | a, |
qint16x4_t | b, | ||
int | fixed_point_position | ||
) |
Division fixed point 16 bit (4 elements)
[in] | a | First 16 bit fixed point input vector |
[in] | b | Second 16 bit fixed point input vector |
[in] | fixed_point_position | Fixed point position that expresses the number of bits for the fractional part of the number |
Division fixed point 8bit (8 elements)
[in] | a | First 8bit fixed point input vector |
[in] | b | Second 8bit fixed point input vector |
[in] | fixed_point_position | Fixed point position that expresses the number of bits for the fractional part of the number |
qint16x8_t arm_compute::vdivq_qs16 | ( | qint16x8_t | a, |
qint16x8_t | b, | ||
int | fixed_point_position | ||
) |
Division fixed point 16 bit (8 elements)
[in] | a | First 16 bit fixed point input vector |
[in] | b | Second 16 bit fixed point input vector |
[in] | fixed_point_position | Fixed point position that expresses the number of bits for the fractional part of the number |
qint8x16_t arm_compute::vdivq_qs8 | ( | qint8x16_t | a, |
qint8x16_t | b, | ||
int | fixed_point_position | ||
) |
Division fixed point 8bit (16 elements)
[in] | a | First 8bit fixed point input vector |
[in] | b | Second 8bit fixed point input vector |
[in] | fixed_point_position | Fixed point position that expresses the number of bits for the fractional part of the number |
qint16x4_t arm_compute::vdup_n_qs16 | ( | qint16_t | a | ) |
16 bit fixed point vector duplicate (4 elements)
[in] | a | 16 bit fixed point to duplicate |
8 bit fixed point vector duplicate (8 elements)
[in] | a | 8 bit fixed point to duplicate |
qint16x8_t arm_compute::vdupq_n_qs16 | ( | qint16x8_t | a | ) |
16 bit fixed point vector duplicate (8 elements)
[in] | a | 16 bit fixed point to duplicate |
qint16x8_t arm_compute::vdupq_n_qs16_f32 | ( | float | a, |
int | fixed_point_position | ||
) |
Duplicate a float and convert it to 16 bit fixed point vector (8 elements)
[in] | a | floating point value to convert and duplicate |
[in] | fixed_point_position | Fixed point position that expresses the number of bits for the fractional part of the number |
qint8x16_t arm_compute::vdupq_n_qs8 | ( | qint8_t | a | ) |
8 bit fixed point vector duplicate (16 elements)
[in] | a | 8 bit fixed point to duplicate |
qint8x16_t arm_compute::vdupq_n_qs8_f32 | ( | float | a, |
int | fixed_point_position | ||
) |
Duplicate a float and convert it to 8 bit fixed point vector (16 elements)
[in] | a | floating point value to convert and duplicate |
[in] | fixed_point_position | Fixed point position that expresses the number of bits for the fractional part of the number |
float32x4_t arm_compute::vexpq_f32 | ( | float32x4_t | x | ) |
Calculate exponential.
[in] | x | Input vector value in F32 format. |
float32x4_t arm_compute::vfloorq_f32 | ( | float32x4_t | val | ) |
Calculate floor of a vector.
[in] | val | Input vector value in F32 format. |
qint16x4_t arm_compute::vget_high_qs16 | ( | qint16x8_t | a | ) |
Get the higher half of a 16 elements vector.
[in] | a | vector of 8 elements |
qint8x8_t arm_compute::vget_high_qs8 | ( | qint8x16_t | a | ) |
Get the higher half of a 16 elements vector.
[in] | a | vector of 16 elements |
qint16x4_t arm_compute::vget_low_qs16 | ( | qint16x8_t | a | ) |
Get the lower half of a 16 elements vector.
[in] | a | vector of 8 elements |
qint8x8_t arm_compute::vget_low_qs8 | ( | qint8x16_t | a | ) |
Get the lower half of a 16 elements vector.
[in] | a | vector of 16 elements |
float32x2_t arm_compute::vinv_f32 | ( | float32x2_t | x | ) |
Calculate reciprocal.
[in] | x | Input value. |
float32x4_t arm_compute::vinvq_f32 | ( | float32x4_t | x | ) |
Calculate reciprocal.
[in] | x | Input value. |
float32x2_t arm_compute::vinvsqrt_f32 | ( | float32x2_t | x | ) |
Calculate inverse square root.
[in] | x | Input value. |
qint16x4_t arm_compute::vinvsqrt_qs16 | ( | qint16x4_t | a, |
int | fixed_point_position | ||
) |
Calculate inverse square root for fixed point 16 bit using Newton-Raphosn method (4 elements)
[in] | a | 16 bit fixed point input vector |
[in] | fixed_point_position | Fixed point position that expresses the number of bits for the fractional part of the number |
Calculate inverse square root for fixed point 8bit using Newton-Raphosn method (8 elements)
[in] | a | 8bit fixed point input vector |
[in] | fixed_point_position | Fixed point position that expresses the number of bits for the fractional part of the number |
float32x4_t arm_compute::vinvsqrtq_f32 | ( | float32x4_t | x | ) |
Calculate inverse square root.
[in] | x | Input value. |
qint16x8_t arm_compute::vinvsqrtq_qs16 | ( | qint16x8_t | a, |
int | fixed_point_position | ||
) |
Calculate inverse square root for fixed point 8bit using Newton-Raphosn method (8 elements)
[in] | a | 16 bit fixed point input vector |
[in] | fixed_point_position | Fixed point position that expresses the number of bits for the fractional part of the number |
qint8x16_t arm_compute::vinvsqrtq_qs8 | ( | qint8x16_t | a, |
int | fixed_point_position | ||
) |
Calculate inverse square root for fixed point 8bit using Newton-Raphosn method (16 elements)
[in] | a | 8bit fixed point input vector |
[in] | fixed_point_position | Fixed point position that expresses the number of bits for the fractional part of the number |
qint16x4_t arm_compute::vld1_dup_qs16 | ( | const qint16_t * | addr | ) |
Load all lanes of 16 bit fixed point vector with same value from memory (4 elements)
[in] | addr | Memory address of the 16 bit fixed point scalar value to load |
Load all lanes of 8 bit fixed point vector with same value from memory (8 elements)
[in] | addr | Memory address of the 8 bit fixed point scalar value to load |
Referenced by arm_compute::detail::load_matrix_row().
qint16x4_t arm_compute::vld1_qs16 | ( | const qint16_t * | addr | ) |
Load a single 16 bit fixed point vector from memory (4 elements)
[in] | addr | Memory address of the 16 bit fixed point vector to load |
Load a single 8 bit fixed point vector from memory (8 elements)
[in] | addr | Memory address of the 8 bit fixed point vector to load |
Referenced by arm_compute::detail::convolve_3x3< 1 >().
qint16x8_t arm_compute::vld1q_dup_qs16 | ( | const qint16_t * | addr | ) |
Load all lanes of 16 bit fixed point vector with same value from memory (8 elements)
[in] | addr | Memory address of the 16 bit fixed point scalar value to load |
qint8x16_t arm_compute::vld1q_dup_qs8 | ( | const qint8_t * | addr | ) |
Load all lanes of 8 bit fixed point vector with same value from memory (16 elements)
[in] | addr | Memory address of the 8 bit fixed point scalar value to load |
qint16x8_t arm_compute::vld1q_qs16 | ( | const qint16_t * | addr | ) |
Load a single 16 bit fixed point vector from memory (8 elements)
[in] | addr | Memory address of the 16 bit fixed point vector to load |
qint8x16_t arm_compute::vld1q_qs8 | ( | const qint8_t * | addr | ) |
Load a single 8 bit fixed point vector from memory (16 elements)
[in] | addr | Memory address of the 8 bit fixed point vector to load |
qint16x8x2_t arm_compute::vld2q_qs16 | ( | qint16_t * | addr | ) |
Load two 16 bit fixed point vectors from memory (8x2 elements)
[in] | addr | Memory address of the 16 bit fixed point vectors to load |
qint16x4_t arm_compute::vlog_qs16 | ( | qint16x4_t | a, |
int | fixed_point_position | ||
) |
Calculate logarithm fixed point 16 bit (4 elements)
[in] | a | 16 bit fixed point input vector |
[in] | fixed_point_position | Fixed point position that expresses the number of bits for the fractional part of the number |
Calculate logarithm fixed point 8 bit (8 elements)
[in] | a | 8bit fixed point input vector |
[in] | fixed_point_position | Fixed point position that expresses the number of bits for the fractional part of the number |
float32x4_t arm_compute::vlogq_f32 | ( | float32x4_t | x | ) |
Calculate logarithm.
[in] | x | Input vector value in F32 format. |
qint16x8_t arm_compute::vlogq_qs16 | ( | qint16x8_t | a, |
int | fixed_point_position | ||
) |
Calculate logarithm fixed point 16 bit (8 elements)
[in] | a | 16 bit fixed point input vector |
[in] | fixed_point_position | Fixed point position that expresses the number of bits for the fractional part of the number |
qint8x16_t arm_compute::vlogq_qs8 | ( | qint8x16_t | a, |
int | fixed_point_position | ||
) |
Calculate logarithm fixed point 16bit (16 elements)
[in] | a | 8bit fixed point input vector |
[in] | fixed_point_position | Fixed point position that expresses the number of bits for the fractional part of the number |
float32x4x2_t arm_compute::vmax2q_f32 | ( | float32x4x2_t | a, |
float32x4x2_t | b | ||
) |
Compute lane-by-lane maximum between elements of a float vector with 4x2 elements.
[in] | a | Float input vector |
[in] | b | Float input vector |
qint16x4_t arm_compute::vmax_qs16 | ( | qint16x4_t | a, |
qint16x4_t | b | ||
) |
16 bit fixed point vector max (4 elements)
[in] | a | First 16 bit fixed point input vector |
[in] | b | Second 16 bit fixed point input vector |
8 bit fixed point vector max (8 elements)
[in] | a | First 8 bit fixed point input vector |
[in] | b | Second 8 bit fixed point input vector |
qint16x8_t arm_compute::vmaxq_qs16 | ( | qint16x8_t | a, |
qint16x8_t | b | ||
) |
16 bit fixed point vector max (8 elements)
[in] | a | First 16 bit fixed point input vector |
[in] | b | Second 16 bit fixed point input vector |
qint8x16_t arm_compute::vmaxq_qs8 | ( | qint8x16_t | a, |
qint8x16_t | b | ||
) |
8 bit fixed point vector max (16 elements)
[in] | a | First 8 bit fixed point input vector |
[in] | b | Second 8 bit fixed point input vector |
qint16x4_t arm_compute::vmin_qs16 | ( | qint16x4_t | a, |
qint16x4_t | b | ||
) |
16 bit fixed point vector min (4 elements)
[in] | a | First 16 bit fixed point input vector |
[in] | b | Second 16 bit fixed point input vector |
8 bit fixed point vector min (8 elements)
[in] | a | First 8 bit fixed point input vector |
[in] | b | Second 8 bit fixed point input vector |
qint16x8_t arm_compute::vminq_qs16 | ( | qint16x8_t | a, |
qint16x8_t | b | ||
) |
16 bit fixed point vector min (8 elements)
[in] | a | First 16 bit fixed point input vector |
[in] | b | Second 16 bit fixed point input vector |
qint8x16_t arm_compute::vminq_qs8 | ( | qint8x16_t | a, |
qint8x16_t | b | ||
) |
8 bit fixed point vector min (16 elements)
[in] | a | First 8 bit fixed point input vector |
[in] | b | Second 8 bit fixed point input vector |
qint16x4_t arm_compute::vmla_qs16 | ( | qint16x4_t | a, |
qint16x4_t | b, | ||
qint16x4_t | c, | ||
int | fixed_point_position | ||
) |
16 bit fixed point vector multiply-accumulate (4 elements).
This operation performs the product between b
and c
and add the result to a
(a + b * c).
[in] | a | First 16 bit fixed point input vector where the result of multiplication must be added to |
[in] | b | Second 16 bit fixed point input vector |
[in] | c | Third 16 bit fixed point input vector |
[in] | fixed_point_position | Fixed point position that expresses the number of bits for the fractional part of the number |
8 bit fixed point vector multiply-accumulate (8 elements).
This operation performs the product between b
and c
and add the result to a
(a + b * c).
[in] | a | First 8 bit fixed point input vector where the result of multiplication must be added to |
[in] | b | Second 8 bit fixed point input vector |
[in] | c | Third 8 bit fixed point input vector |
[in] | fixed_point_position | Fixed point position that expresses the number of bits for the fractional part of the number |
qint32x4_t arm_compute::vmlal_qs16 | ( | qint32x4_t | a, |
qint16x4_t | b, | ||
qint16x4_t | c, | ||
int | fixed_point_position | ||
) |
16 bit fixed point vector multiply-accumulate long (4 elements).
This operation performs the product between b
and c
and add the result to the 32 bit fixed point vector a
(a + b * c). 4 elements
[in] | a | First 32 bit fixed point input vector where the result of multiplication must be added to |
[in] | b | Second 16 bit fixed point input vector |
[in] | c | Third 16 bit fixed point input vector |
[in] | fixed_point_position | Fixed point position that expresses the number of bits for the fractional part of the number |
qint16x8_t arm_compute::vmlal_qs8 | ( | qint16x8_t | a, |
qint8x8_t | b, | ||
qint8x8_t | c, | ||
int | fixed_point_position | ||
) |
8 bit fixed point vector multiply-accumulate long (8 elements).
This operation performs the product between b
and c
and add the result to the 16 bit fixed point vector a
(a + b * c). 8 elements
[in] | a | First 16 bit fixed point input vector where the result of multiplication must be added to |
[in] | b | Second 8 bit fixed point input vector |
[in] | c | Third 8 bit fixed point input vector |
[in] | fixed_point_position | Fixed point position that expresses the number of bits for the fractional part of the number |
|
inline |
Perform a multiply-accumulate on all 16 components of a QASYMM8 vector.
vd*vs + vo
[in] | vd | Input vector value in QASYMM8 format |
[in] | vs | Vector multiplier in F32 format. The multiplier value must be duplicated across all four lanes. |
[in] | vo | Vector addend in F32 format. The addend value must be duplicated across all four lanes. |
Definition at line 34 of file NEAsymm.inl.
qint16x8_t arm_compute::vmlaq_qs16 | ( | qint16x8_t | a, |
qint16x8_t | b, | ||
qint16x8_t | c, | ||
int | fixed_point_position | ||
) |
16 bit fixed point vector multiply-accumulate (16 elements).
This operation performs the product between b
and c
and add the result to a
(a + b * c).
[in] | a | First 16 bit fixed point input vector where the result of multiplication must be added to |
[in] | b | Second 16 bit fixed point input vector |
[in] | c | Third 16 bit fixed point input vector |
[in] | fixed_point_position | Fixed point position that expresses the number of bits for the fractional part of the number |
qint8x16_t arm_compute::vmlaq_qs8 | ( | qint8x16_t | a, |
qint8x16_t | b, | ||
qint8x16_t | c, | ||
int | fixed_point_position | ||
) |
8 bit fixed point vector multiply-accumulate (16 elements).
This operation performs the product between b
and c
and add the result to a
(a + b * c).
[in] | a | First 8 bit fixed point input vector where the result of multiplication must be added to |
[in] | b | Second 8 bit fixed point input vector |
[in] | c | Third 8 bit fixed point input vector |
[in] | fixed_point_position | Fixed point position that expresses the number of bits for the fractional part of the number |
qint16x4_t arm_compute::vmul_qs16 | ( | qint16x4_t | a, |
qint16x4_t | b, | ||
int | fixed_point_position | ||
) |
16 bit fixed point vector multiply (4 elements)
[in] | a | First 16 bit fixed point input vector |
[in] | b | Second 16 bit fixed point input vector |
[in] | fixed_point_position | Fixed point position that expresses the number of bits for the fractional part of the number |
8 bit fixed point vector multiply (8 elements)
[in] | a | First 8 bit fixed point input vector |
[in] | b | Second 8 bit fixed point input vector |
[in] | fixed_point_position | Fixed point position that expresses the number of bits for the fractional part of the number |
qint32x4_t arm_compute::vmull_qs16 | ( | qint16x4_t | a, |
qint16x4_t | b, | ||
int | fixed_point_position | ||
) |
16 bit fixed point vector long multiply (4 elements)
[in] | a | First 16 bit fixed point input vector |
[in] | b | Second 16 bit fixed point input vector |
[in] | fixed_point_position | Fixed point position that expresses the number of bits for the fractional part of the number |
qint16x8_t arm_compute::vmull_qs8 | ( | qint8x8_t | a, |
qint8x8_t | b, | ||
int | fixed_point_position | ||
) |
8 bit fixed point vector long multiply (8 elements)
[in] | a | First 8 bit fixed point input vector |
[in] | b | Second 8 bit fixed point input vector |
[in] | fixed_point_position | Fixed point position that expresses the number of bits for the fractional part of the number |
Referenced by arm_compute::detail::convolve_3x3< 1 >().
qint16x8_t arm_compute::vmulq_qs16 | ( | qint16x8_t | a, |
qint16x8_t | b, | ||
int | fixed_point_position | ||
) |
16 bit fixed point vector multiply (8 elements)
[in] | a | First 16 bit fixed point input vector |
[in] | b | Second 16 bit fixed point input vector |
[in] | fixed_point_position | Fixed point position that expresses the number of bits for the fractional part of the number |
qint8x16_t arm_compute::vmulq_qs8 | ( | qint8x16_t | a, |
qint8x16_t | b, | ||
int | fixed_point_position | ||
) |
8 bit fixed point vector multiply (16 elements)
[in] | a | First 8 bit fixed point input vector |
[in] | b | Second 8 bit fixed point input vector |
[in] | fixed_point_position | Fixed point position that expresses the number of bits for the fractional part of the number |
int16x4_t arm_compute::vpaddl_qs8 | ( | qint8x8_t | a | ) |
8 bit fixed point vector saturating pairwise add (8 elements)
[in] | a | 8 bit fixed point input vector |
qint16x4_t arm_compute::vpmax_qs16 | ( | qint16x4_t | a, |
qint16x4_t | b | ||
) |
16 bit fixed point vector pairwise max (4 elements)
[in] | a | First 16 bit fixed point input vector |
[in] | b | Second 16 bit fixed point input vector |
8 bit fixed point vector pairwise max (8 elements)
[in] | a | First 8 bit fixed point input vector |
[in] | b | Second 8 bit fixed point input vector |
qint16x4_t arm_compute::vpmin_qs16 | ( | qint16x4_t | a, |
qint16x4_t | b | ||
) |
16 bit fixed point vector pairwise min (4 elements)
[in] | a | First 16 bit fixed point input vector |
[in] | b | Second 16 bit fixed point input vector |
8 bit fixed point vector pairwise min (8 elements)
[in] | a | First 8 bit fixed point input vector |
[in] | b | Second 8 bit fixed point input vector |
float32x4_t arm_compute::vpowq_f32 | ( | float32x4_t | val, |
float32x4_t | n | ||
) |
Calculate n power of a number.
pow(x,n) = e^(n*log(x))
[in] | val | Input vector value in F32 format. |
[in] | n | Powers to raise the input to. |
qint16x4_t arm_compute::vqabs_qs16 | ( | qint16x4_t | a | ) |
Saturating absolute value of 16 bit fixed point vector (4 elements)
[in] | a | 4 bit fixed point input vector |
Saturating absolute value of 8 bit fixed point vector (8 elements)
[in] | a | 8 bit fixed point input vector |
qint16x8_t arm_compute::vqabsq_qs16 | ( | qint16x8_t | a | ) |
Saturating absolute value of 16 bit fixed point vector (8 elements)
[in] | a | 16 bit fixed point input vector |
qint8x16_t arm_compute::vqabsq_qs8 | ( | qint8x16_t | a | ) |
Saturating absolute value of 8 bit fixed point vector (16 elements)
[in] | a | 8 bit fixed point input vector |
qint16x4_t arm_compute::vqadd_qs16 | ( | qint16x4_t | a, |
qint16x4_t | b | ||
) |
16 bit fixed point vector saturating add (4 elements)
[in] | a | First 16 bit fixed point input vector |
[in] | b | Second 16 bit fixed point input vector |
8 bit fixed point vector saturating add (8 elements)
[in] | a | First 8 bit fixed point input vector |
[in] | b | Second 8 bit fixed point input vector |
qint16x8_t arm_compute::vqaddq_qs16 | ( | qint16x8_t | a, |
qint16x8_t | b | ||
) |
16 bit fixed point vector saturating add (8 elements)
[in] | a | First 16 bit fixed point input vector |
[in] | b | Second 16 bit fixed point input vector |
qint8x16_t arm_compute::vqaddq_qs8 | ( | qint8x16_t | a, |
qint8x16_t | b | ||
) |
8 bit fixed point vector saturating add (16 elements)
[in] | a | First 8 bit fixed point input vector |
[in] | b | Second 8 bit fixed point input vector |
qint16x4_t arm_compute::vqcvt_qs16_f32 | ( | const float32x4_t | a, |
int | fixed_point_position | ||
) |
Convert a float vector with 4 elements to 16 bit fixed point vector with 4 elements.
[in] | a | Float input vector |
[in] | fixed_point_position | Fixed point position that expresses the number of bits for the fractional part of the number |
qint8x8_t arm_compute::vqcvt_qs8_f32 | ( | const float32x4x2_t | a, |
int | fixed_point_position | ||
) |
Convert a float vector with 4x2 elements to 8 bit fixed point vector with 8 elements.
[in] | a | Float input vector |
[in] | fixed_point_position | Fixed point position that expresses the number of bits for the fractional part of the number |
qint16x8_t arm_compute::vqcvtq_qs16_f32 | ( | const float32x4x2_t & | a, |
int | fixed_point_position | ||
) |
Convert a float vector with 4x2 elements to 16 bit fixed point vector with 8 elements.
[in] | a | Float input vector |
[in] | fixed_point_position | Fixed point position that expresses the number of bits for the fractional part of the number |
qint8x16_t arm_compute::vqcvtq_qs8_f32 | ( | const float32x4x4_t & | a, |
int | fixed_point_position | ||
) |
Convert a float vector with 4x4 elements to 8 bit fixed point vector with 16 elements.
[in] | a | Float input vector |
[in] | fixed_point_position | Fixed point position that expresses the number of bits for the fractional part of the number |
qint16x4_t arm_compute::vqexp_qs16 | ( | qint16x4_t | a, |
int | fixed_point_position | ||
) |
Calculate saturating exponential fixed point 16 bit (4 elements)
[in] | a | 8bit fixed point input vector |
[in] | fixed_point_position | Fixed point position that expresses the number of bits for the fractional part of the number |
Calculate saturating exponential fixed point 8bit (8 elements)
[in] | a | 8bit fixed point input vector |
[in] | fixed_point_position | Fixed point position that expresses the number of bits for the fractional part of the number |
qint16x8_t arm_compute::vqexpq_qs16 | ( | qint16x8_t | a, |
int | fixed_point_position | ||
) |
Calculate saturating exponential fixed point 16 bit (8 elements)
[in] | a | 16 bit fixed point input vector |
[in] | fixed_point_position | Fixed point position that expresses the number of bits for the fractional part of the number |
qint8x16_t arm_compute::vqexpq_qs8 | ( | qint8x16_t | a, |
int | fixed_point_position | ||
) |
Calculate saturating exponential fixed point 8bit (16 elements)
[in] | a | 8bit fixed point input vector |
[in] | fixed_point_position | Fixed point position that expresses the number of bits for the fractional part of the number |
qint16x4_t arm_compute::vqinvsqrt_qs16 | ( | qint16x4_t | a, |
int | fixed_point_position | ||
) |
Calculate saturating inverse square root for fixed point 16 bit using Newton-Raphosn method (4 elements)
[in] | a | 16 bit fixed point input vector |
[in] | fixed_point_position | Fixed point position that expresses the number of bits for the fractional part of the number |
Calculate saturating inverse square root for fixed point 8bit using Newton-Raphosn method (8 elements)
[in] | a | 8bit fixed point input vector |
[in] | fixed_point_position | Fixed point position that expresses the number of bits for the fractional part of the number |
qint16x8_t arm_compute::vqinvsqrtq_qs16 | ( | qint16x8_t | a, |
int | fixed_point_position | ||
) |
Calculate saturating inverse square root for fixed point 16 bit using Newton-Raphosn method (8 elements)
[in] | a | 16 bit fixed point input vector |
[in] | fixed_point_position | Fixed point position that expresses the number of bits for the fractional part of the number |
qint8x16_t arm_compute::vqinvsqrtq_qs8 | ( | qint8x16_t | a, |
int | fixed_point_position | ||
) |
Calculate saturating inverse square root for fixed point 8bit using Newton-Raphosn method (16 elements)
[in] | a | 8bit fixed point input vector |
[in] | fixed_point_position | Fixed point position that expresses the number of bits for the fractional part of the number |
qint16x4_t arm_compute::vqmla_qs16 | ( | qint16x4_t | a, |
qint16x4_t | b, | ||
qint16x4_t | c, | ||
int | fixed_point_position | ||
) |
16 bit fixed point vector saturating multiply-accumulate (4 elements).
This operation performs the product between b
and c
and add the result to a
(a + b * c).
[in] | a | First 16 bit fixed point input vector where the result of multiplication must be added to |
[in] | b | Second 16 bit fixed point input vector |
[in] | c | Third 16 bit fixed point input vector |
[in] | fixed_point_position | Fixed point position that expresses the number of bits for the fractional part of the number |
qint8x8_t arm_compute::vqmla_qs8 | ( | qint8x8_t | a, |
qint8x8_t | b, | ||
qint8x8_t | c, | ||
int | fixed_point_position | ||
) |
8 bit fixed point vector saturating multiply-accumulate (8 elements).
This operation performs the product between b
and c
and add the result to a
(a + b * c).
[in] | a | First 8 bit fixed point input vector where the result of multiplication must be added to |
[in] | b | Second 8 bit fixed point input vector |
[in] | c | Third 8 bit fixed point input vector |
[in] | fixed_point_position | Fixed point position that expresses the number of bits for the fractional part of the number |
qint32x4_t arm_compute::vqmlal_qs16 | ( | qint32x4_t | a, |
qint16x4_t | b, | ||
qint16x4_t | c, | ||
int | fixed_point_position | ||
) |
16 bit fixed point vector saturating multiply-accumulate long (4 elements).
The saturation is performed on the 16 bit fixed point output vector. This operation performs the product between b
and c
and add the result to the 32 bit fixed point vector a
(a + b * c). 4 elements
[in] | a | First 32 bit fixed point input vector where the result of multiplication must be added to |
[in] | b | Second 16 bit fixed point input vector |
[in] | c | Third 16 bit fixed point input vector |
[in] | fixed_point_position | Fixed point position that expresses the number of bits for the fractional part of the number |
qint16x8_t arm_compute::vqmlal_qs8 | ( | qint16x8_t | a, |
qint8x8_t | b, | ||
qint8x8_t | c, | ||
int | fixed_point_position | ||
) |
8 bit fixed point vector saturating multiply-accumulate long (8 elements).
The saturation is performed on the 16 bit fixed point output vector. This operation performs the product between b
and c
and add the result to the 16 bit fixed point vector a
(a + b * c). 8 elements
[in] | a | First 16 bit fixed point input vector where the result of multiplication must be added to |
[in] | b | Second 8 bit fixed point input vector |
[in] | c | Third 8 bit fixed point input vector |
[in] | fixed_point_position | Fixed point position that expresses the number of bits for the fractional part of the number |
Referenced by arm_compute::detail::convolve_3x3< 1 >().
qint16x8_t arm_compute::vqmlaq_qs16 | ( | qint16x8_t | a, |
qint16x8_t | b, | ||
qint16x8_t | c, | ||
int | fixed_point_position | ||
) |
16 bit fixed point vector saturating multiply-accumulate (8 elements).
This operation performs the product between b
and c
and add the result to a
(a + b * c).
[in] | a | First 16 bit fixed point input vector where the result of multiplication must be added to |
[in] | b | Second 16 bit fixed point input vector |
[in] | c | Third 16 bit fixed point input vector |
[in] | fixed_point_position | Fixed point position that expresses the number of bits for the fractional part of the number |
qint8x16_t arm_compute::vqmlaq_qs8 | ( | qint8x16_t | a, |
qint8x16_t | b, | ||
qint8x16_t | c, | ||
int | fixed_point_position | ||
) |
8 bit fixed point vector saturating multiply-accumulate (16 elements).
This operation performs the product between b
and c
and add the result to a
(a + b * c).
[in] | a | First 8 bit fixed point input vector where the result of multiplication must be added to |
[in] | b | Second 8 bit fixed point input vector |
[in] | c | Third 8 bit fixed point input vector |
[in] | fixed_point_position | Fixed point position that expresses the number of bits for the fractional part of the number |
qint8x8_t arm_compute::vqmovn_q16 | ( | qint16x8_t | a | ) |
16 bit fixed point vector saturating narrow (8 elements)
[in] | a | 16 bit fixed point vector to convert |
qint16x4_t arm_compute::vqmovn_q32 | ( | qint32x4_t | a | ) |
32 bit fixed point vector saturating narrow (4 elements)
[in] | a | 32 bit fixed point vector to convert |
qint16x4_t arm_compute::vqmul_qs16 | ( | qint16x4_t | a, |
qint16x4_t | b, | ||
int | fixed_point_position | ||
) |
16 bit fixed point vector saturating multiply (4 elements)
[in] | a | First 16 bit fixed point input vector |
[in] | b | Second 16 bit fixed point input vector |
[in] | fixed_point_position | Fixed point position that expresses the number of bits for the fractional part of the number |
8 bit fixed point vector saturating multiply (8 elements)
[in] | a | First 8 bit fixed point input vector |
[in] | b | Second 8 bit fixed point input vector |
[in] | fixed_point_position | Fixed point position that expresses the number of bits for the fractional part of the number |
qint16x8_t arm_compute::vqmulq_qs16 | ( | qint16x8_t | a, |
qint16x8_t | b, | ||
int | fixed_point_position | ||
) |
16 bit fixed point vector saturating multiply (8 elements)
[in] | a | First 16 bit fixed point input vector |
[in] | b | Second 16 bit fixed point input vector |
[in] | fixed_point_position | Fixed point position that expresses the number of bits for the fractional part of the number |
qint8x16_t arm_compute::vqmulq_qs8 | ( | qint8x16_t | a, |
qint8x16_t | b, | ||
int | fixed_point_position | ||
) |
8 bit fixed point vector saturating multiply (16 elements)
[in] | a | First 8 bit fixed point input vector |
[in] | b | Second 8 bit fixed point input vector |
[in] | fixed_point_position | Fixed point position that expresses the number of bits for the fractional part of the number |
qint16x8_t arm_compute::vqpowq_qs16 | ( | qint16x8_t | a, |
qint16x8_t | b, | ||
int | fixed_point_position | ||
) |
Calculate saturating n power for fixed point 16bit (8 elements).
pow(a,b) = e^(b*log(a))
[in] | a | 16bit fixed point input vector |
[in] | b | 16bit fixed point power vector |
[in] | fixed_point_position | Fixed point position that expresses the number of bits for the fractional part of the number |
qint8x16_t arm_compute::vqpowq_qs8 | ( | qint8x16_t | a, |
qint8x16_t | b, | ||
int | fixed_point_position | ||
) |
Calculate saturating n power for fixed point 8bit (16 elements).
pow(a,b) = e^(b*log(a))
[in] | a | 8bit fixed point input vector |
[in] | b | 8bit fixed point power vector |
[in] | fixed_point_position | Fixed point position that expresses the number of bits for the fractional part of the number |
qint16x4_t arm_compute::vqsub_qs16 | ( | qint16x4_t | a, |
qint16x4_t | b | ||
) |
16 bit fixed point vector saturating subtraction (4 elements)
[in] | a | First 16 bit fixed point input vector |
[in] | b | Second 16 bit fixed point input vector |
8 bit fixed point vector saturating subtraction (8 elements)
[in] | a | First 8 bit fixed point input vector |
[in] | b | Second 8 bit fixed point input vector |
qint16x8_t arm_compute::vqsubq_qs16 | ( | qint16x8_t | a, |
qint16x8_t | b | ||
) |
16 bit fixed point vector saturating subtraction (8 elements)
[in] | a | First 16 bit fixed point input vector |
[in] | b | Second 16 bit fixed point input vector |
qint8x16_t arm_compute::vqsubq_qs8 | ( | qint8x16_t | a, |
qint8x16_t | b | ||
) |
8 bit fixed point vector saturating subtraction (16 elements)
[in] | a | First 8 bit fixed point input vector |
[in] | b | Second 8 bit fixed point input vector |
qint16x4_t arm_compute::vqtanh_qs16 | ( | qint16x4_t | a, |
int | fixed_point_position | ||
) |
Calculate hyperbolic tangent for fixed point 16 bit (4 elements)
[in] | a | 16 bit fixed point input vector |
[in] | fixed_point_position | Fixed point position that expresses the number of bits for the fractional part of the number |
Calculate hyperbolic tangent for fixed point 8bit (8 elements)
[in] | a | 8bit fixed point input vector |
[in] | fixed_point_position | Fixed point position that expresses the number of bits for the fractional part of the number |
qint16x8_t arm_compute::vqtanhq_qs16 | ( | qint16x8_t | a, |
int | fixed_point_position | ||
) |
Calculate hyperbolic tangent for fixed point 16bit (8 elements)
[in] | a | 16 bit fixed point input vector |
[in] | fixed_point_position | Fixed point position that expresses the number of bits for the fractional part of the number |
qint8x16_t arm_compute::vqtanhq_qs8 | ( | qint8x16_t | a, |
int | fixed_point_position | ||
) |
Calculate hyperbolic tangent for fixed point 8bit (16 elements)
[in] | a | 8bit fixed point input vector |
[in] | fixed_point_position | Fixed point position that expresses the number of bits for the fractional part of the number |
qint16x4_t arm_compute::vrecip_qs16 | ( | qint16x4_t | a, |
int | fixed_point_position | ||
) |
Calculate reciprocal of a fixed point 8bit number using the Newton-Raphson method.
(4 elements)
[in] | a | 16 bit fixed point input vector |
[in] | fixed_point_position | Fixed point position that expresses the number of bits for the fractional part of the number |
Calculate reciprocal of a fixed point 8bit number using the Newton-Raphson method.
(8 elements)
[in] | a | 8bit fixed point input vector |
[in] | fixed_point_position | Fixed point position that expresses the number of bits for the fractional part of the number |
qint16x8_t arm_compute::vrecipq_qs16 | ( | qint16x8_t | a, |
int | fixed_point_position | ||
) |
Calculate reciprocal of a fixed point 8bit number using the Newton-Raphson method.
(8 elements)
[in] | a | 16 bit fixed point input vector |
[in] | fixed_point_position | Fixed point position that expresses the number of bits for the fractional part of the number |
qint8x16_t arm_compute::vrecipq_qs8 | ( | qint8x16_t | a, |
int | fixed_point_position | ||
) |
Calculate reciprocal of a fixed point 8bit number using the Newton-Raphson method.
(16 elements)
[in] | a | 8bit fixed point input vector |
[in] | fixed_point_position | Fixed point position that expresses the number of bits for the fractional part of the number |
void arm_compute::vst1_qs16 | ( | qint16_t * | addr, |
qint16x4_t | b | ||
) |
Store a single 16 bit fixed point vector to memory (4 elements)
[in] | addr | Memory address where the 16 bit fixed point vector should be stored |
[in] | b | 16 bit fixed point vector to store |
Referenced by arm_compute::detail::store_results< 3 >().
Store a single 8 bit fixed point vector to memory (8 elements)
[in] | addr | Memory address where the 8 bit fixed point vector should be stored |
[in] | b | 8 bit fixed point vector to store |
void arm_compute::vst1q_qs16 | ( | qint16_t * | addr, |
qint16x8_t | b | ||
) |
Store a single 16 bit fixed point vector to memory (8 elements)
[in] | addr | Memory address where the 16 bit fixed point vector should be stored |
[in] | b | 16 bit fixed point vector to store |
Referenced by arm_compute::detail::store_results< 1 >(), and arm_compute::detail::store_results< 2 >().
void arm_compute::vst1q_qs8 | ( | qint8_t * | addr, |
qint8x16_t | b | ||
) |
Store a single 8 bit fixed point vector to memory (16 elements)
[in] | addr | Memory address where the 8 bit fixed point vector should be stored |
[in] | b | 8 bit fixed point vector to store |
void arm_compute::vst2q_qs16 | ( | qint16_t * | addr, |
qint16x8x2_t | b | ||
) |
Store two 16 bit fixed point vector to memory (8x2 elements)
[in] | addr | Memory address where the 16 bit fixed point vectors should be stored |
[in] | b | 16 bit fixed point vectors to store |
qint16x4_t arm_compute::vsub_qs16 | ( | qint16x4_t | a, |
qint16x4_t | b | ||
) |
16 bit fixed point vector subtraction (4 elements)
[in] | a | First 16 bit fixed point input vector |
[in] | b | Second 16 bit fixed point input vector |
8 bit fixed point vector subtraction (8 elements)
[in] | a | First 8 bit fixed point input vector |
[in] | b | Second 8 bit fixed point input vector |
qint16x8_t arm_compute::vsubq_qs16 | ( | qint16x8_t | a, |
qint16x8_t | b | ||
) |
16 bit fixed point vector subtraction (8 elements)
[in] | a | First 16 bit fixed point input vector |
[in] | b | Second 16 bit fixed point input vector |
qint8x16_t arm_compute::vsubq_qs8 | ( | qint8x16_t | a, |
qint8x16_t | b | ||
) |
8 bit fixed point vector subtraction (16 elements)
[in] | a | First 8 bit fixed point input vector |
[in] | b | Second 8 bit fixed point input vector |
float32x4_t arm_compute::vtanhq_f32 | ( | float32x4_t | val | ) |
Calculate hyperbolic tangent.
tanh(x) = (e^2x - 1)/(e^2x + 1)
[in] | val | Input vector value in F32 format. |
qint16x4_t arm_compute::vtaylor_poly_qs16 | ( | qint16x4_t | a, |
int | fixed_point_position | ||
) |
Perform a 4th degree polynomial approximation.
(4 elements)
[in] | a | 16 bit fixed point input vector |
[in] | fixed_point_position | Fixed point position that expresses the number of bits for the fractional part of the number |
Perform a 4th degree polynomial approximation.
(8 elements)
[in] | a | 8bit fixed point input vector |
[in] | fixed_point_position | Fixed point position that expresses the number of bits for the fractional part of the number |
float32x4_t arm_compute::vtaylor_polyq_f32 | ( | float32x4_t | x, |
const std::array< float32x4_t, 8 > & | coeffs | ||
) |
Perform a 7th degree polynomial approximation using Estrin's method.
[in] | x | Input vector value in F32 format. |
[in] | coeffs | Polynomial coefficients table. |
qint16x8_t arm_compute::vtaylor_polyq_qs16 | ( | qint16x8_t | a, |
int | fixed_point_position | ||
) |
Perform a 4th degree polynomial approximation.
(8 elements)
[in] | a | 16 bit fixed point input vector |
[in] | fixed_point_position | Fixed point position that expresses the number of bits for the fractional part of the number |
qint8x16_t arm_compute::vtaylor_polyq_qs8 | ( | qint8x16_t | a, |
int | fixed_point_position | ||
) |
Perform a 4th degree polynomial approximation.
(16 elements)
[in] | a | 8bit fixed point input vector |
[in] | fixed_point_position | Fixed point position that expresses the number of bits for the fractional part of the number |
constexpr uint8_t CONSTANT_BORDER_VALUE = 199 |
Constant value of the border pixels when using BorderMode::CONSTANT.
const std::array<float32x4_t, 8> exp_tab |
Exponent polynomial coefficients.
Definition at line 28 of file NEMath.inl.
const std::array<float32x4_t, 8> log_tab |
Logarithm polynomial coefficients.
Definition at line 43 of file NEMath.inl.
constexpr size_t MAX_DIMS = 6 |
Constant value used to indicate maximum dimensions of a Window, TensorShape and Coordinates.
Definition at line 37 of file Dimensions.h.
constexpr float SCALE_PYRAMID_HALF = 0.5f |
Constant value used to indicate a half-scale pyramid.
Definition at line 104 of file Types.h.
Referenced by arm_compute::test::validation::DATA_TEST_CASE(), arm_compute::test::validation::reference::gaussian_pyramid_half(), and arm_compute::test::validation::reference::optical_flow().