30 auto convertFunc = [](uint8_t* dst,
const uint8_t* src,
size_t size)
32 auto input =
reinterpret_cast<const float*
>(src);
33 auto output =
reinterpret_cast<BFloat16*
>(dst);
34 size_t numElements = size/2;
38 for (
const auto& pair : m_TensorHandlePairs)
virtual void Execute() const override
void ValidateInputsOutputs(const std::string &descName, unsigned int numExpectedIn, unsigned int numExpectedOut) const
Copyright (c) 2021 ARM Limited and Contributors.
NeonConvertFp32ToBf16Workload(const ConvertFp32ToBf16QueueDescriptor &descriptor, const WorkloadInfo &info)
profiling::ProfilingGuid GetGuid() const final
static void ConvertFloat32ToBFloat16(const float *srcFloat32Buffer, size_t numElements, void *dstBFloat16Buffer)
Contains information about TensorInfos of a layer.
void CopyTensorContentsGeneric(const ITensorHandle *srcTensor, ITensorHandle *dstTensor, CopyFunc copy)
void GatherTensorHandlePairs(const DescriptorType &descriptor, std::vector< std::pair< SrcTensorHandleType *, DstTensorHandleType *>> &tensorHandlePairs)
#define ARMNN_SCOPED_PROFILING_EVENT_NEON_GUID(name, guid)