COMPMID-765: Collapse execution window in CL kernels.

Updated following kernels to collapse their execution window and reduce
number of kernel enqueues:
-CLArithmeticAddition
-CLArithmeticSubtraction
-CLPixelWiseMultiplication

Change-Id: I13d503515a20fa9be1401ead1e27e9bbc6627975
Reviewed-on: https://eu-gerrit-1.euhpc.arm.com/114878
Reviewed-by: Anthony Barbier <anthony.barbier@arm.com>
Tested-by: Anthony Barbier <anthony.barbier@arm.com>

3 files changed, 16 insertions, 12 deletions

diff --git a/src/core/CL/kernels/CLArithmeticAdditionKernel.cpp b/src/core/CL/kernels/CLArithmeticAdditionKernel.cpp
index 2789573293..75701ee011 100644
--- a/src/core/CL/kernels/CLArithmeticAdditionKernel.cpp
+++ b/src/core/CL/kernels/CLArithmeticAdditionKernel.cpp
@@ -154,14 +154,16 @@ void CLArithmeticAdditionKernel::run(const Window &window, cl::CommandQueue &que
     ARM_COMPUTE_ERROR_ON_UNCONFIGURED_KERNEL(this);
     ARM_COMPUTE_ERROR_ON_INVALID_SUBWINDOW(ICLKernel::window(), window);
 
-    Window slice = window.first_slice_window_2D();
+    Window collapsed = window.collapse_if_possible(ICLKernel::window(), Window::DimZ);
+    Window slice     = collapsed.first_slice_window_3D();
+
     do
     {
         unsigned int idx = 0;
-        add_2D_tensor_argument(idx, _input1, slice);
-        add_2D_tensor_argument(idx, _input2, slice);
-        add_2D_tensor_argument(idx, _output, slice);
+        add_3D_tensor_argument(idx, _input1, slice);
+        add_3D_tensor_argument(idx, _input2, slice);
+        add_3D_tensor_argument(idx, _output, slice);
         enqueue(queue, *this, slice);
     }
-    while(window.slide_window_slice_2D(slice));
+    while(collapsed.slide_window_slice_3D(slice));
 }
diff --git a/src/core/CL/kernels/CLArithmeticSubtractionKernel.cpp b/src/core/CL/kernels/CLArithmeticSubtractionKernel.cpp
index cc2ef1f023..8308aa0767 100644
--- a/src/core/CL/kernels/CLArithmeticSubtractionKernel.cpp
+++ b/src/core/CL/kernels/CLArithmeticSubtractionKernel.cpp
@@ -146,15 +146,16 @@ void CLArithmeticSubtractionKernel::run(const Window &window, cl::CommandQueue &
     ARM_COMPUTE_ERROR_ON_UNCONFIGURED_KERNEL(this);
     ARM_COMPUTE_ERROR_ON_INVALID_SUBWINDOW(ICLKernel::window(), window);
 
-    Window slice = window.first_slice_window_2D();
+    Window collapsed = window.collapse_if_possible(ICLKernel::window(), Window::DimZ);
+    Window slice     = collapsed.first_slice_window_3D();
 
     do
     {
         unsigned int idx = 0;
-        add_2D_tensor_argument(idx, _input1, slice);
-        add_2D_tensor_argument(idx, _input2, slice);
-        add_2D_tensor_argument(idx, _output, slice);
+        add_3D_tensor_argument(idx, _input1, slice);
+        add_3D_tensor_argument(idx, _input2, slice);
+        add_3D_tensor_argument(idx, _output, slice);
         enqueue(queue, *this, slice);
     }
-    while(window.slide_window_slice_2D(slice));
+    while(collapsed.slide_window_slice_3D(slice));
 }
diff --git a/src/core/CL/kernels/CLPixelWiseMultiplicationKernel.cpp b/src/core/CL/kernels/CLPixelWiseMultiplicationKernel.cpp
index fd5e5d5862..6dba9c0f95 100644
--- a/src/core/CL/kernels/CLPixelWiseMultiplicationKernel.cpp
+++ b/src/core/CL/kernels/CLPixelWiseMultiplicationKernel.cpp
@@ -227,7 +227,8 @@ void CLPixelWiseMultiplicationKernel::run(const Window &window, cl::CommandQueue
     ARM_COMPUTE_ERROR_ON_UNCONFIGURED_KERNEL(this);
     ARM_COMPUTE_ERROR_ON_INVALID_SUBWINDOW(ICLKernel::window(), window);
 
-    Window slice = window.first_slice_window_3D();
+    Window collapsed = window.collapse_if_possible(ICLKernel::window(), Window::DimZ);
+    Window slice     = collapsed.first_slice_window_3D();
 
     do
     {
@@ -237,5 +238,5 @@ void CLPixelWiseMultiplicationKernel::run(const Window &window, cl::CommandQueue
         add_3D_tensor_argument(idx, _output, slice);
         enqueue(queue, *this, slice);
     }
-    while(window.slide_window_slice_3D(slice));
+    while(collapsed.slide_window_slice_3D(slice));
 }