diff options
Diffstat (limited to 'src/backends/reference/workloads/RefConvertFp32ToBf16Workload.cpp')
-rw-r--r-- | src/backends/reference/workloads/RefConvertFp32ToBf16Workload.cpp | 17 |
1 files changed, 14 insertions, 3 deletions
diff --git a/src/backends/reference/workloads/RefConvertFp32ToBf16Workload.cpp b/src/backends/reference/workloads/RefConvertFp32ToBf16Workload.cpp index 181b236e83..7fe302a5ad 100644 --- a/src/backends/reference/workloads/RefConvertFp32ToBf16Workload.cpp +++ b/src/backends/reference/workloads/RefConvertFp32ToBf16Workload.cpp @@ -15,12 +15,23 @@ namespace armnn void RefConvertFp32ToBf16Workload::Execute() const { + Execute(m_Data.m_Inputs, m_Data.m_Outputs); +} + +void RefConvertFp32ToBf16Workload::ExecuteAsync(WorkingMemDescriptor &workingMemDescriptor) +{ + Execute(workingMemDescriptor.m_Inputs, workingMemDescriptor.m_Outputs); +} + +void RefConvertFp32ToBf16Workload::Execute(std::vector<ITensorHandle*> inputs, + std::vector<ITensorHandle*> outputs) const +{ ARMNN_SCOPED_PROFILING_EVENT(Compute::CpuRef, "RefConvertFp32ToBf16Workload_Execute"); - const float* const input = GetInputTensorDataFloat(0, m_Data); - BFloat16* const output = GetOutputTensorDataBFloat16(0, m_Data); + const float* const input = reinterpret_cast<const float*>(inputs[0]->Map()); + BFloat16* const output = reinterpret_cast<BFloat16*>(outputs[0]->Map()); - unsigned int numElements = GetTensorInfo(m_Data.m_Inputs[0]).GetNumElements(); + unsigned int numElements = GetTensorInfo(inputs[0]).GetNumElements(); armnnUtils::FloatingPointConverter::ConvertFloat32ToBFloat16(input, numElements, output); } |