• Home
  • Line#
  • Scopes#
  • Navigate#
  • Raw
  • Download
1 //
2 // Copyright © 2017 Arm Ltd and Contributors. All rights reserved.
3 // SPDX-License-Identifier: MIT
4 //
5 
6 #include "ClMaximumWorkload.hpp"
7 
8 #include "ClWorkloadUtils.hpp"
9 
10 #include <aclCommon/ArmComputeUtils.hpp>
11 #include <aclCommon/ArmComputeTensorUtils.hpp>
12 
13 #include <armnn/backends/TensorHandle.hpp>
14 
15 #include <cl/ClLayerSupport.hpp>
16 #include <cl/ClTensorHandle.hpp>
17 #include <cl/ClLayerSupport.hpp>
18 
19 namespace armnn
20 {
21 
22 using namespace armcomputetensorutils;
23 
ClMaximumWorkloadValidate(const TensorInfo & input0,const TensorInfo & input1,const TensorInfo & output)24 arm_compute::Status ClMaximumWorkloadValidate(const TensorInfo& input0,
25                                               const TensorInfo& input1,
26                                               const TensorInfo& output)
27 {
28     const arm_compute::TensorInfo aclInput0Info = BuildArmComputeTensorInfo(input0);
29     const arm_compute::TensorInfo aclInput1Info = BuildArmComputeTensorInfo(input1);
30     const arm_compute::TensorInfo aclOutputInfo = BuildArmComputeTensorInfo(output);
31 
32     const arm_compute::Status aclStatus = arm_compute::CLElementwiseMax::validate(&aclInput0Info,
33                                                                                   &aclInput1Info,
34                                                                                   &aclOutputInfo);
35 
36     return aclStatus;
37 }
38 
ClMaximumWorkload(const MaximumQueueDescriptor & descriptor,const WorkloadInfo & info,const arm_compute::CLCompileContext & clCompileContext)39 ClMaximumWorkload::ClMaximumWorkload(const MaximumQueueDescriptor& descriptor,
40                                      const WorkloadInfo& info,
41                                      const arm_compute::CLCompileContext& clCompileContext)
42     : ClBaseWorkload<MaximumQueueDescriptor>(descriptor, info)
43 {
44     m_Data.ValidateInputsOutputs("ClMaximumWorkload", 2, 1);
45 
46     arm_compute::ICLTensor& input0 = static_cast<IClTensorHandle*>(m_Data.m_Inputs[0])->GetTensor();
47     arm_compute::ICLTensor& input1 = static_cast<IClTensorHandle*>(m_Data.m_Inputs[1])->GetTensor();
48     arm_compute::ICLTensor& output = static_cast<IClTensorHandle*>(m_Data.m_Outputs[0])->GetTensor();
49 
50     {
51         ARMNN_SCOPED_PROFILING_EVENT(Compute::Undefined, "ClMaximumWorkload_configure");
52         m_MaximumLayer.configure(clCompileContext, &input0, &input1, &output);
53     }
54 }
55 
Execute() const56 void ClMaximumWorkload::Execute() const
57 {
58     ARMNN_SCOPED_PROFILING_EVENT_CL_GUID("ClMaximumWorkload_Execute", this->GetGuid());
59     RunClFunction(m_MaximumLayer, CHECK_LOCATION());
60 }
61 
62 } //namespace armnn
63