IVGCVSW-5328-5329 Fuse Activation

 * Added Fused Activation Optimization to both CL and Neon backends.
 * Added Fused Activation support to all the CL and Neon workloads
   that support it.
 * Changed ProfilingTest network to be a Convolution layer
   followed by an Abs layer rather than an Activation layer.
 * Added IBackendInternal::OptimizeSubgraphView function that can accept a
   ModelOptions.
 * Network will now call OptimizeSubgraphView passing in the ModelOptions.

Signed-off-by: Keith Davis <keith.davis@arm.com>
Signed-off-by: Mike Kelly <mike.kelly@arm.com>
Signed-off-by: Teresa Charlin <teresa.charlinreyes@arm.com>
Change-Id: Ib536ac3cbafc7d9b35c139ad9a65b7735262cd9d

1 files changed, 12 insertions, 3 deletions

diff --git a/src/backends/cl/workloads/ClAdditionWorkload.cpp b/src/backends/cl/workloads/ClAdditionWorkload.cpp
index 18e2400ccd..7e75a04110 100644
--- a/src/backends/cl/workloads/ClAdditionWorkload.cpp
+++ b/src/backends/cl/workloads/ClAdditionWorkload.cpp
@@ -8,6 +8,7 @@
 #include <cl/ClTensorHandle.hpp>
 #include <backendsCommon/CpuTensorHandle.hpp>
 #include <aclCommon/ArmComputeTensorUtils.hpp>
+#include <aclCommon/ArmComputeUtils.hpp>
 
 #include "ClWorkloadUtils.hpp"
 
@@ -26,7 +27,10 @@ ClAdditionWorkload::ClAdditionWorkload(const AdditionQueueDescriptor& descriptor
     arm_compute::ICLTensor& input0 = static_cast<IClTensorHandle*>(this->m_Data.m_Inputs[0])->GetTensor();
     arm_compute::ICLTensor& input1 = static_cast<IClTensorHandle*>(this->m_Data.m_Inputs[1])->GetTensor();
     arm_compute::ICLTensor& output = static_cast<IClTensorHandle*>(this->m_Data.m_Outputs[0])->GetTensor();
-    m_Layer.configure(&input0, &input1, &output, g_AclConvertPolicy);
+
+    const arm_compute::ActivationLayerInfo activationInfo = ConvertAdditionalInfoToAclActivationLayerInfo(descriptor);
+
+    m_Layer.configure(&input0, &input1, &output, g_AclConvertPolicy, activationInfo);
 }
 
 void ClAdditionWorkload::Execute() const
@@ -37,16 +41,21 @@ void ClAdditionWorkload::Execute() const
 
 arm_compute::Status ClAdditionValidate(const TensorInfo& input0,
                                        const TensorInfo& input1,
-                                       const TensorInfo& output)
+                                       const TensorInfo& output,
+                                       const ActivationDescriptor* activationDescriptor)
 {
     const arm_compute::TensorInfo aclInput0Info = BuildArmComputeTensorInfo(input0);
     const arm_compute::TensorInfo aclInput1Info = BuildArmComputeTensorInfo(input1);
     const arm_compute::TensorInfo aclOutputInfo = BuildArmComputeTensorInfo(output);
 
+    const arm_compute::ActivationLayerInfo activationInfo = ConvertActivationDescriptorToAclActivationLayerInfo(
+            activationDescriptor);
+
     const arm_compute::Status aclStatus = arm_compute::CLArithmeticAddition::validate(&aclInput0Info,
                                                                                       &aclInput1Info,
                                                                                       &aclOutputInfo,
-                                                                                      g_AclConvertPolicy);
+                                                                                      g_AclConvertPolicy,
+                                                                                      activationInfo);
 
     return aclStatus;
 }