[acl] Use lws_hint() instead of default parameter (#4745)

This commit applies `lws_hint()` in the case of below operators.
TensorFlowMax, TensorFlowSum, EXP, GATHER, CAST, PAD

Signed-off-by: Jiyoung Yun <jy910.yun@samsung.com>

diff --git a/libs/ARMComputeEx/src/core/CL/kernels/CLCastKernel.cpp b/libs/ARMComputeEx/src/core/CL/kernels/CLCastKernel.cpp
index bf7ebae..ac2963f 100644 (file)
--- a/libs/ARMComputeEx/src/core/CL/kernels/CLCastKernel.cpp
+++ b/libs/ARMComputeEx/src/core/CL/kernels/CLCastKernel.cpp
@@ -97,6 +97,6 @@ void CLCastKernel::run(const Window &window, cl::CommandQueue &queue)
     unsigned int idx = 0;
     add_3D_tensor_argument(idx, _input, slice);
     add_3D_tensor_argument(idx, _output, slice);
-    enqueue(queue, *this, slice);
+    enqueue(queue, *this, slice, lws_hint());
   } while (collapsed.slide_window_slice_3D(slice));
 }

diff --git a/libs/ARMComputeEx/src/core/CL/kernels/CLExpKernel.cpp b/libs/ARMComputeEx/src/core/CL/kernels/CLExpKernel.cpp
index b1ee21b..2dc43a3 100644 (file)
--- a/libs/ARMComputeEx/src/core/CL/kernels/CLExpKernel.cpp
+++ b/libs/ARMComputeEx/src/core/CL/kernels/CLExpKernel.cpp
@@ -72,6 +72,6 @@ void CLExpKernel::run(const Window &window, cl::CommandQueue &queue)
     unsigned int idx = 0;
     add_3D_tensor_argument(idx, _input, slice);
     add_3D_tensor_argument(idx, _output, slice);
-    enqueue(queue, *this, slice);
+    enqueue(queue, *this, slice, lws_hint());
   } while (collapsed.slide_window_slice_3D(slice));
 }

diff --git a/libs/ARMComputeEx/src/core/CL/kernels/CLGatherKernel.cpp b/libs/ARMComputeEx/src/core/CL/kernels/CLGatherKernel.cpp
index 24e3dc9..f90680c 100644 (file)
--- a/libs/ARMComputeEx/src/core/CL/kernels/CLGatherKernel.cpp
+++ b/libs/ARMComputeEx/src/core/CL/kernels/CLGatherKernel.cpp
@@ -174,5 +174,5 @@ void CLGatherKernel::run(const Window &window, cl::CommandQueue &queue)
   add_4D_tensor_argument(idx, _input, window_collapsed);
   add_3D_tensor_argument(idx, _indices, window_collapsed);
   add_4D_tensor_argument(idx, _output, window_collapsed);
-  enqueue(queue, *this, window_collapsed);
+  enqueue(queue, *this, window_collapsed, lws_hint());
 }

diff --git a/libs/ARMComputeEx/src/core/CL/kernels/CLPadLayerKernel.cpp b/libs/ARMComputeEx/src/core/CL/kernels/CLPadLayerKernel.cpp
index 832b2cc..50f8a39 100644 (file)
--- a/libs/ARMComputeEx/src/core/CL/kernels/CLPadLayerKernel.cpp
+++ b/libs/ARMComputeEx/src/core/CL/kernels/CLPadLayerKernel.cpp
@@ -130,6 +130,6 @@ void CLPadLayerKernel::run(const Window &window, cl::CommandQueue &queue)
     add_4D_tensor_argument(idx, _input, slice_in);
     add_4D_tensor_argument(idx, _output, slice_out);
     _kernel.setArg<cl_int4>(idx++, paddingValues);
-    enqueue(queue, *this, slice_out);
+    enqueue(queue, *this, slice_out, lws_hint());
   } while (window.slide_window_slice_4D(slice_out) && window.slide_window_slice_4D(slice_in));
 }

diff --git a/libs/ARMComputeEx/src/core/CL/kernels/CLReduceOperationKernel.cpp b/libs/ARMComputeEx/src/core/CL/kernels/CLReduceOperationKernel.cpp
index ed53b88..24e89db 100644 (file)
--- a/libs/ARMComputeEx/src/core/CL/kernels/CLReduceOperationKernel.cpp
+++ b/libs/ARMComputeEx/src/core/CL/kernels/CLReduceOperationKernel.cpp
@@ -172,7 +172,7 @@ void CLReduceOperationKernel::run(const Window &window, cl::CommandQueue &queue)
   idx = 0;
   add_4D_tensor_argument(idx, _input, slice_in);
   add_4D_tensor_argument(idx, _output, slice_out);
-  enqueue(queue, *this, slice_out);
+  enqueue(queue, *this, slice_out, lws_hint());
 
   // Recover output's shape of output tensor
   _output->info()->set_tensor_shape(shape_out);