diff options
| author | Nikhil Raj <nikhil.raj@arm.com> | 2018-10-18 14:27:50 +0100 |
|---|---|---|
| committer | Matthew Bentham <matthew.bentham@arm.com> | 2018-10-22 16:57:54 +0100 |
| commit | d134093a271b60e248942af9757e8236e8f41ac1 (patch) | |
| tree | 7b15baae3cd028dc4cb0f0bb5302ecfbc9db8225 /src/backends/cl/workloads/ClBatchNormalizationFloatWorkload.cpp | |
| parent | 33f0ae0d293f5048089f2a04985a8a8bfa1d75a6 (diff) | |
| download | armnn-d134093a271b60e248942af9757e8236e8f41ac1.tar.gz | |
IVGCVSW-2023 CL and Neon implementation of BatchNorm with NHWC
Change-Id: I962e986607e5d045cd97b9eaeaea2f5ae624db35
Diffstat (limited to 'src/backends/cl/workloads/ClBatchNormalizationFloatWorkload.cpp')
| -rw-r--r-- | src/backends/cl/workloads/ClBatchNormalizationFloatWorkload.cpp | 24 |
1 files changed, 18 insertions, 6 deletions
diff --git a/src/backends/cl/workloads/ClBatchNormalizationFloatWorkload.cpp b/src/backends/cl/workloads/ClBatchNormalizationFloatWorkload.cpp index 5bff7a63c9..24be7cddca 100644 --- a/src/backends/cl/workloads/ClBatchNormalizationFloatWorkload.cpp +++ b/src/backends/cl/workloads/ClBatchNormalizationFloatWorkload.cpp @@ -23,12 +23,20 @@ arm_compute::Status ClBatchNormalizationValidate(const TensorInfo& input, const TensorInfo& gamma, const BatchNormalizationDescriptor &desc) { - const arm_compute::TensorInfo aclInputInfo = BuildArmComputeTensorInfo(input); - const arm_compute::TensorInfo aclOutputInfo = BuildArmComputeTensorInfo(output); - const arm_compute::TensorInfo aclMeanInfo = BuildArmComputeTensorInfo(mean); - const arm_compute::TensorInfo aclVarInfo = BuildArmComputeTensorInfo(var); - const arm_compute::TensorInfo aclBetaInfo = BuildArmComputeTensorInfo(beta); - const arm_compute::TensorInfo aclGammaInfo = BuildArmComputeTensorInfo(gamma); + const DataLayout dataLayout = desc.m_DataLayout.GetDataLayout(); + + const arm_compute::TensorInfo aclInputInfo = + armcomputetensorutils::BuildArmComputeTensorInfo(input, dataLayout); + const arm_compute::TensorInfo aclOutputInfo = + armcomputetensorutils::BuildArmComputeTensorInfo(output, dataLayout); + const arm_compute::TensorInfo aclMeanInfo = + armcomputetensorutils::BuildArmComputeTensorInfo(mean, dataLayout); + const arm_compute::TensorInfo aclVarInfo = + armcomputetensorutils::BuildArmComputeTensorInfo(var, dataLayout); + const arm_compute::TensorInfo aclBetaInfo = + armcomputetensorutils::BuildArmComputeTensorInfo(beta, dataLayout); + const arm_compute::TensorInfo aclGammaInfo = + armcomputetensorutils::BuildArmComputeTensorInfo(gamma, dataLayout); return arm_compute::CLBatchNormalizationLayer::validate(&aclInputInfo, &aclOutputInfo, @@ -60,6 +68,10 @@ ClBatchNormalizationFloatWorkload::ClBatchNormalizationFloatWorkload( arm_compute::ICLTensor& input = static_cast<IClTensorHandle*>(m_Data.m_Inputs[0])->GetTensor(); arm_compute::ICLTensor& output = static_cast<IClTensorHandle*>(m_Data.m_Outputs[0])->GetTensor(); + arm_compute::DataLayout aclDataLayout = ConvertDataLayout(m_Data.m_Parameters.m_DataLayout.GetDataLayout()); + input.info()->set_data_layout(aclDataLayout); + output.info()->set_data_layout(aclDataLayout); + m_Layer.configure(&input, &output, m_Mean.get(), |