29 auto convertFunc = [](uint8_t* dst,
const uint8_t* src,
size_t size)
31 auto input =
reinterpret_cast<const Half*
>(src);
32 auto output =
reinterpret_cast<float*
>(dst);
33 size_t numElements = size/2;
37 for (
const auto& pair : m_TensorHandlePairs)
#define ARMNN_SCOPED_PROFILING_EVENT_NEON(name)
void CopyTensorContentsGeneric(const ITensorHandle *srcTensor, ITensorHandle *dstTensor, CopyFunc copy)
NeonConvertFp16ToFp32Workload(const ConvertFp16ToFp32QueueDescriptor &descriptor, const WorkloadInfo &info)
static void ConvertFloat16To32(const void *srcFloat16Buffer, size_t numElements, float *dstFloat32Buffer)
virtual void Execute() const override
const QueueDescriptor m_Data
void GatherTensorHandlePairs(const DescriptorType &descriptor, std::vector< std::pair< SrcTensorHandleType *, DstTensorHandleType *>> &tensorHandlePairs)
void ValidateInputsOutputs(const std::string &descName, unsigned int numExpectedIn, unsigned int numExpectedOut) const