385 bool reshapePrepare(const Shape& input, const int32_t* targetDims, const int32_t targetDimsSize,  in reshapePrepare()  argument
391     int32_t numInputElements = (int32_t)getNumberOfElements(input);  in reshapePrepare()
414     output->type = input.type;  in reshapePrepare()
416     output->offset = input.offset;  in reshapePrepare()
417     output->scale = input.scale;  in reshapePrepare()
422 bool depthToSpacePrepare(const Shape& input, int32_t blockSize, Shape* output) {  in depthToSpacePrepare()  argument
423     NN_OPS_CHECK(getNumberOfDimensions(input) == 4);  in depthToSpacePrepare()
426     uint32_t batches = getSizeOfDimension(input, 0);  in depthToSpacePrepare()
427     uint32_t height = getSizeOfDimension(input, 1);  in depthToSpacePrepare()
428     uint32_t width = getSizeOfDimension(input, 2);  in depthToSpacePrepare()
429     uint32_t channels = getSizeOfDimension(input, 3);  in depthToSpacePrepare()
432     output->type = input.type;  in depthToSpacePrepare()
435     output->offset = input.offset;  in depthToSpacePrepare()
436     output->scale = input.scale;  in depthToSpacePrepare()
441 bool spaceToDepthPrepare(const Shape& input, int32_t blockSize, Shape* output) {  in spaceToDepthPrepare()  argument
442     NN_OPS_CHECK(getNumberOfDimensions(input) == 4);  in spaceToDepthPrepare()
445     uint32_t batches = getSizeOfDimension(input, 0);  in spaceToDepthPrepare()
446     uint32_t height = getSizeOfDimension(input, 1);  in spaceToDepthPrepare()
447     uint32_t width = getSizeOfDimension(input, 2);  in spaceToDepthPrepare()
448     uint32_t channels = getSizeOfDimension(input, 3);  in spaceToDepthPrepare()
453     output->type = input.type;  in spaceToDepthPrepare()
456     output->offset = input.offset;  in spaceToDepthPrepare()
457     output->scale = input.scale;  in spaceToDepthPrepare()
503 bool padPrepare(const Shape& input, const int32_t* paddingsData, const Shape& paddingsShape,  in padPrepare()  argument
505     uint32_t numInputDims = getNumberOfDimensions(input);  in padPrepare()
519         outDims[i] = beforePadding + getSizeOfDimension(input, i) + afterPadding;  in padPrepare()
521     output->type = input.type;  in padPrepare()
523     output->offset = input.offset;  in padPrepare()
524     output->scale = input.scale;  in padPrepare()
529 bool batchToSpacePrepare(const Shape& input, const int32_t* blockSizeData,  in batchToSpacePrepare()  argument
532     NN_OPS_CHECK(getNumberOfDimensions(input) == 4);  in batchToSpacePrepare()
540     uint32_t batches = getSizeOfDimension(input, 0);  in batchToSpacePrepare()
541     uint32_t height = getSizeOfDimension(input, 1);  in batchToSpacePrepare()
542     uint32_t width = getSizeOfDimension(input, 2);  in batchToSpacePrepare()
543     uint32_t channels = getSizeOfDimension(input, 3);  in batchToSpacePrepare()
546     output->type = input.type;  in batchToSpacePrepare()
549     output->offset = input.offset;  in batchToSpacePrepare()
550     output->scale = input.scale;  in batchToSpacePrepare()
555 bool spaceToBatchPrepare(const Shape& input, const int32_t* blockSizeData,  in spaceToBatchPrepare()  argument
559     NN_OPS_CHECK(getNumberOfDimensions(input) == 4);  in spaceToBatchPrepare()
573     uint32_t batches = getSizeOfDimension(input, 0);  in spaceToBatchPrepare()
574     uint32_t height = getSizeOfDimension(input, 1);  in spaceToBatchPrepare()
575     uint32_t width = getSizeOfDimension(input, 2);  in spaceToBatchPrepare()
576     uint32_t channels = getSizeOfDimension(input, 3);  in spaceToBatchPrepare()
584     output->type = input.type;  in spaceToBatchPrepare()
588     output->offset = input.offset;  in spaceToBatchPrepare()
589     output->scale = input.scale;  in spaceToBatchPrepare()
594 bool meanPrepare(const Shape& input, const int32_t* axisData, const Shape& axisShape, bool keepDims,  in meanPrepare()  argument
600     int32_t numInputDims = static_cast<int32_t>(getNumberOfDimensions(input));  in meanPrepare()
617                 outDims[idx] = getSizeOfDimension(input, idx);  in meanPrepare()
654                 outDims[idx - numSkipAxis] = getSizeOfDimension(input, idx);  in meanPrepare()
664     output->type = input.type;  in meanPrepare()
665     output->offset = input.offset;  in meanPrepare()
666     output->scale = input.scale;  in meanPrepare()
671 bool argMinMaxPrepare(const Shape& input, int32_t axis, Shape* output) {  in argMinMaxPrepare()  argument
672     NN_CHECK(handleNegativeAxis(input, &axis));  in argMinMaxPrepare()
678     if (getNumberOfDimensions(input) > 1) {  in argMinMaxPrepare()
679         output->dimensions.reserve(getNumberOfDimensions(input) - 1);  in argMinMaxPrepare()
680         output->dimensions.insert(output->dimensions.end(), input.dimensions.begin(),  in argMinMaxPrepare()
681                                   input.dimensions.begin() + axis);  in argMinMaxPrepare()
682         output->dimensions.insert(output->dimensions.end(), input.dimensions.begin() + axis + 1,  in argMinMaxPrepare()
683                                   input.dimensions.end());  in argMinMaxPrepare()
691 bool splitPrepare(const Shape& input, int32_t axis, int32_t numOutputs,  in splitPrepare()  argument
693     NN_CHECK(handleNegativeAxis(input, &axis));  in splitPrepare()
695     const int32_t sizeOfAxisToSplit = input.dimensions[axis];  in splitPrepare()
700         output->at(i).type = input.type;  in splitPrepare()
701         output->at(i).dimensions = input.dimensions;  in splitPrepare()
703         output->at(i).offset = input.offset;  in splitPrepare()
704         output->at(i).scale = input.scale;  in splitPrepare()
709 bool groupedConvPrepare(const Shape& input, const Shape& filter, const Shape& bias,  in groupedConvPrepare()  argument
714         NN_OPS_CHECK(input.type == OperandType::TENSOR_QUANT8_ASYMM ||  in groupedConvPrepare()
715                      input.type == OperandType::TENSOR_QUANT8_ASYMM_SIGNED);  in groupedConvPrepare()
717         NN_OPS_CHECK(input.type == filter.type);  in groupedConvPrepare()
719     if (input.type == OperandType::TENSOR_QUANT8_ASYMM ||  in groupedConvPrepare()
720         input.type == OperandType::TENSOR_QUANT8_ASYMM_SIGNED) {  in groupedConvPrepare()
723         NN_OPS_CHECK(input.type == bias.type);  in groupedConvPrepare()
725     NN_OPS_CHECK(getNumberOfDimensions(input) == 4);  in groupedConvPrepare()
731     NN_OPS_CHECK(getSizeOfDimension(filter, 3) * numGroups == getSizeOfDimension(input, 3));  in groupedConvPrepare()
735     uint32_t width = getSizeOfDimension(input, 2);  in groupedConvPrepare()
736     uint32_t height = getSizeOfDimension(input, 1);  in groupedConvPrepare()
739     uint32_t batches = getSizeOfDimension(input, 0);  in groupedConvPrepare()
751     output->type = input.type;  in groupedConvPrepare()