1 files changed, 125 insertions, 51 deletions

diff --git a/src/core/CL/kernels/CLArithmeticSubtractionKernel.cpp b/src/core/CL/kernels/CLArithmeticSubtractionKernel.cpp
index 299ac553e9..95d201104d 100644
--- a/src/core/CL/kernels/CLArithmeticSubtractionKernel.cpp
+++ b/src/core/CL/kernels/CLArithmeticSubtractionKernel.cpp
@@ -36,45 +36,82 @@
 #include <set>
 #include <string>
 
-using namespace arm_compute;
-
+namespace arm_compute
+{
 namespace
 {
-Status validate_arguments(const ITensorInfo *input1, const ITensorInfo *input2, const ITensorInfo *output, ConvertPolicy policy)
+constexpr unsigned int num_elems_processed_per_iteration = 16;
+
+Status validate_arguments(const ITensorInfo &input1, const ITensorInfo &input2, const ITensorInfo &output, ConvertPolicy policy)
 {
     ARM_COMPUTE_UNUSED(policy);
-    ARM_COMPUTE_RETURN_ERROR_ON_F16_UNSUPPORTED(input1);
-    ARM_COMPUTE_RETURN_ERROR_ON_DATA_TYPE_CHANNEL_NOT_IN(input1, 1, DataType::U8, DataType::S16, DataType::F16, DataType::F32);
-    ARM_COMPUTE_RETURN_ERROR_ON_F16_UNSUPPORTED(input2);
-    ARM_COMPUTE_RETURN_ERROR_ON_DATA_TYPE_CHANNEL_NOT_IN(input2, 1, DataType::U8, DataType::S16, DataType::F16, DataType::F32);
-    ARM_COMPUTE_RETURN_ERROR_ON_MISMATCHING_SHAPES(input1, input2);
+    ARM_COMPUTE_RETURN_ERROR_ON_F16_UNSUPPORTED(&input1);
+    ARM_COMPUTE_RETURN_ERROR_ON_DATA_TYPE_CHANNEL_NOT_IN(&input1, 1, DataType::U8, DataType::QASYMM8, DataType::S16, DataType::F16, DataType::F32);
+    ARM_COMPUTE_RETURN_ERROR_ON_F16_UNSUPPORTED(&input2);
+    ARM_COMPUTE_RETURN_ERROR_ON_DATA_TYPE_CHANNEL_NOT_IN(&input2, 1, DataType::U8, DataType::QASYMM8, DataType::S16, DataType::F16, DataType::F32);
+    const bool is_qasymm = is_data_type_quantized_asymmetric(input1.data_type()) || is_data_type_quantized_asymmetric(input2.data_type());
+    if(is_qasymm)
+    {
+        ARM_COMPUTE_RETURN_ERROR_ON_MISMATCHING_DATA_TYPES(&input1, &input2);
+    }
+
+    const TensorShape out_shape = TensorShape::broadcast_shape(input1.tensor_shape(), input2.tensor_shape());
+
+    ARM_COMPUTE_RETURN_ERROR_ON_MSG(out_shape.total_size() == 0, "Inputs are not broadcast compatible");
 
     // Validate in case of configured output
-    if((output != nullptr) && (output->total_size() != 0))
+    if(output.total_size() > 0)
     {
-        ARM_COMPUTE_RETURN_ERROR_ON_F16_UNSUPPORTED(output);
-        ARM_COMPUTE_RETURN_ERROR_ON_DATA_TYPE_CHANNEL_NOT_IN(output, 1, DataType::U8, DataType::S16, DataType::F16, DataType::F32);
-        ARM_COMPUTE_RETURN_ERROR_ON_MSG(output->data_type() == DataType::U8 && (input1->data_type() != DataType::U8 || input2->data_type() != DataType::U8),
+        ARM_COMPUTE_RETURN_ERROR_ON_F16_UNSUPPORTED(&output);
+        ARM_COMPUTE_RETURN_ERROR_ON_DATA_TYPE_CHANNEL_NOT_IN(&output, 1, DataType::U8, DataType::QASYMM8, DataType::S16, DataType::F16, DataType::F32);
+        ARM_COMPUTE_RETURN_ERROR_ON_MSG((output.data_type() == DataType::U8) && ((input1.data_type() != DataType::U8) || (input2.data_type() != DataType::U8)),
                                         "Output can only be U8 if both inputs are U8");
-        ARM_COMPUTE_RETURN_ERROR_ON_MISMATCHING_SHAPES(input1, output);
+        ARM_COMPUTE_RETURN_ERROR_ON_MSG(detail::have_different_dimensions(out_shape, output.tensor_shape(), 0),
+                                        "Wrong shape for output");
+        if(is_qasymm)
+        {
+            ARM_COMPUTE_RETURN_ERROR_ON_MISMATCHING_DATA_TYPES(&input1, &output);
+        }
     }
 
     return Status{};
 }
 
-std::pair<Status, Window> validate_and_configure_window(ITensorInfo *input1, ITensorInfo *input2, ITensorInfo *output)
+std::pair<Status, Window> validate_and_configure_window(ITensorInfo &input1, ITensorInfo &input2, ITensorInfo &output)
 {
-    constexpr unsigned int num_elems_processed_per_iteration = 16;
+    const std::pair<TensorShape, ValidRegion> broadcast_pair = ITensorInfo::broadcast_shape_and_valid_region(input1, input2);
+    const TensorShape &out_shape    = broadcast_pair.first;
+    const ValidRegion &valid_region = broadcast_pair.second;
 
-    Window                 win = calculate_max_window(*input1, Steps(num_elems_processed_per_iteration));
-    AccessWindowHorizontal input1_access(input1, 0, num_elems_processed_per_iteration);
-    AccessWindowHorizontal input2_access(input2, 0, num_elems_processed_per_iteration);
-    AccessWindowHorizontal output_access(output, 0, num_elems_processed_per_iteration);
+    // Auto initialize output if not initialized
+    {
+        set_shape_if_empty(output, out_shape);
 
-    bool window_changed = update_window_and_padding(win, input1_access, input2_access, output_access);
+        if(input1.data_type() == DataType::S16 || input2.data_type() == DataType::S16)
+        {
+            set_format_if_unknown(output, Format::S16);
+        }
+        else if(input1.data_type() == DataType::F16 && input2.data_type() == DataType::F16)
+        {
+            set_format_if_unknown(output, Format::F16);
+        }
+        else if(input1.data_type() == DataType::F32 || input2.data_type() == DataType::F32)
+        {
+            set_format_if_unknown(output, Format::F32);
+        }
+    }
+
+    Window win        = calculate_max_window(valid_region, Steps(num_elems_processed_per_iteration));
+    Window win_input1 = win.broadcast_if_dimension_le_one(input1);
+    Window win_input2 = win.broadcast_if_dimension_le_one(input2);
 
-    ValidRegion valid_region = intersect_valid_regions(input1->valid_region(),
-                                                       input2->valid_region());
+    AccessWindowHorizontal input1_access(&input1, 0, num_elems_processed_per_iteration);
+    AccessWindowHorizontal input2_access(&input2, 0, num_elems_processed_per_iteration);
+    AccessWindowHorizontal output_access(&output, 0, num_elems_processed_per_iteration);
+
+    bool window_changed = update_window_and_padding(win_input1, input1_access)
+                          || update_window_and_padding(win_input2, input2_access)
+                          || update_window_and_padding(win, output_access);
 
     output_access.set_valid_region(win, valid_region);
 
@@ -91,22 +128,11 @@ CLArithmeticSubtractionKernel::CLArithmeticSubtractionKernel()
 void CLArithmeticSubtractionKernel::configure(const ICLTensor *input1, const ICLTensor *input2, ICLTensor *output, ConvertPolicy policy)
 {
     ARM_COMPUTE_ERROR_ON_NULLPTR(input1, input2, output);
+    ARM_COMPUTE_ERROR_THROW_ON(validate_arguments(*input1->info(), *input2->info(), *output->info(), policy));
 
-    // Auto initialize output if not initialized
-    {
-        set_shape_if_empty(*output->info(), input1->info()->tensor_shape());
-
-        if(input1->info()->data_type() == DataType::S16 || input2->info()->data_type() == DataType::S16)
-        {
-            set_format_if_unknown(*output->info(), Format::S16);
-        }
-        else if(input1->info()->data_type() == DataType::F32 || input2->info()->data_type() == DataType::F32)
-        {
-            set_format_if_unknown(*output->info(), Format::F32);
-        }
-    }
-
-    ARM_COMPUTE_ERROR_THROW_ON(validate_arguments(input1->info(), input2->info(), output->info(), policy));
+    // Configure kernel window
+    auto win_config = validate_and_configure_window(*input1->info(), *input2->info(), *output->info());
+    ARM_COMPUTE_ERROR_THROW_ON(win_config.first);
 
     _input1 = input1;
     _input2 = input2;
@@ -114,26 +140,39 @@ void CLArithmeticSubtractionKernel::configure(const ICLTensor *input1, const ICL
 
     bool has_float_out = is_data_type_float(output->info()->data_type());
 
+    // Setup kernel
+    std::string kernel_name = "arithmetic_sub";
+
     // Set kernel build options
-    std::set<std::string> build_opts;
-    build_opts.emplace((policy == ConvertPolicy::WRAP || has_float_out) ? "-DWRAP" : "-DSATURATE");
-    build_opts.emplace("-DDATA_TYPE_IN1=" + get_cl_type_from_data_type(input1->info()->data_type()));
-    build_opts.emplace("-DDATA_TYPE_IN2=" + get_cl_type_from_data_type(input2->info()->data_type()));
-    build_opts.emplace("-DDATA_TYPE_OUT=" + get_cl_type_from_data_type(output->info()->data_type()));
+    CLBuildOptions build_opts;
+    build_opts.add_option_if_else(policy == ConvertPolicy::WRAP || has_float_out, "-DWRAP", "-DSATURATE");
+    build_opts.add_option("-DDATA_TYPE_IN1=" + get_cl_type_from_data_type(input1->info()->data_type()));
+    build_opts.add_option("-DDATA_TYPE_IN2=" + get_cl_type_from_data_type(input2->info()->data_type()));
+    build_opts.add_option("-DDATA_TYPE_OUT=" + get_cl_type_from_data_type(output->info()->data_type()));
+    if(is_data_type_quantized_asymmetric(input1->info()->data_type()))
+    {
+        build_opts.add_option("-DOFFSET_IN1=" + support::cpp11::to_string(input1->info()->quantization_info().offset));
+        build_opts.add_option("-DOFFSET_IN2=" + support::cpp11::to_string(input2->info()->quantization_info().offset));
+        build_opts.add_option("-DOFFSET_OUT=" + support::cpp11::to_string(output->info()->quantization_info().offset));
+        build_opts.add_option("-DSCALE_IN1=" + support::cpp11::to_string(input1->info()->quantization_info().scale));
+        build_opts.add_option("-DSCALE_IN2=" + support::cpp11::to_string(input2->info()->quantization_info().scale));
+        build_opts.add_option("-DSCALE_OUT=" + support::cpp11::to_string(output->info()->quantization_info().scale));
+        kernel_name += "_quantized";
+    }
 
     // Create kernel
-    _kernel = static_cast<cl::Kernel>(CLKernelLibrary::get().create_kernel("arithmetic_sub", build_opts));
+    _kernel = static_cast<cl::Kernel>(CLKernelLibrary::get().create_kernel(kernel_name, build_opts.options()));
 
     // Configure kernel window
-    auto win_config = validate_and_configure_window(input1->info(), input2->info(), output->info());
-    ARM_COMPUTE_ERROR_THROW_ON(win_config.first);
     ICLKernel::configure_internal(win_config.second);
 }
 
 Status CLArithmeticSubtractionKernel::validate(const ITensorInfo *input1, const ITensorInfo *input2, const ITensorInfo *output, ConvertPolicy policy)
 {
-    ARM_COMPUTE_RETURN_ON_ERROR(validate_arguments(input1, input2, output, policy));
-    ARM_COMPUTE_RETURN_ON_ERROR(validate_and_configure_window(input1->clone().get(), input2->clone().get(), output->clone().get()).first);
+    ARM_COMPUTE_RETURN_ERROR_ON_NULLPTR(input1, input2, output);
+
+    ARM_COMPUTE_RETURN_ON_ERROR(validate_arguments(*input1, *input2, *output, policy));
+    ARM_COMPUTE_RETURN_ON_ERROR(validate_and_configure_window(*input1->clone(), *input2->clone(), *output->clone()).first);
 
     return Status{};
 }
@@ -143,16 +182,51 @@ 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 collapsed = window.collapse_if_possible(ICLKernel::window(), Window::DimZ);
-    Window slice     = collapsed.first_slice_window_3D();
+    const TensorShape &in_shape1 = _input1->info()->tensor_shape();
+    const TensorShape &in_shape2 = _input2->info()->tensor_shape();
+    const TensorShape &out_shape = _output->info()->tensor_shape();
+
+    // Collapse only if broadcast dimensions is less than 2, or in case of no broadcasting
+    bool can_collapse = true;
+    if(std::min(in_shape1.total_size(), in_shape2.total_size()) > 1)
+    {
+        can_collapse = (std::min(in_shape1.num_dimensions(), in_shape2.num_dimensions()) > Window::DimZ);
+        for(size_t d = Window::DimZ; can_collapse && (d < out_shape.num_dimensions()); d++)
+        {
+            can_collapse = (in_shape1[d] == in_shape2[d]);
+        }
+    }
+
+    bool   has_collapsed = false;
+    Window collapsed     = can_collapse ? window.collapse_if_possible(ICLKernel::window(), Window::DimZ, &has_collapsed) : window;
+
+    const TensorShape &in_shape1_collapsed = has_collapsed ? in_shape1.collapsed_from(Window::DimZ) : in_shape1;
+    const TensorShape &in_shape2_collapsed = has_collapsed ? in_shape2.collapsed_from(Window::DimZ) : in_shape2;
+
+    Window slice        = collapsed.first_slice_window_3D();
+    Window slice_input1 = slice.broadcast_if_dimension_le_one(in_shape1_collapsed);
+    Window slice_input2 = slice.broadcast_if_dimension_le_one(in_shape2_collapsed);
 
     do
     {
         unsigned int idx = 0;
-        add_3D_tensor_argument(idx, _input1, slice);
-        add_3D_tensor_argument(idx, _input2, slice);
+
+        add_3D_tensor_argument(idx, _input1, slice_input1);
+        add_3D_tensor_argument(idx, _input2, slice_input2);
         add_3D_tensor_argument(idx, _output, slice);
+
         enqueue(queue, *this, slice);
+
+        collapsed.slide_window_slice_3D(slice_input1);
+        collapsed.slide_window_slice_3D(slice_input2);
     }
     while(collapsed.slide_window_slice_3D(slice));
 }
+
+BorderSize CLArithmeticSubtractionKernel::border_size() const
+{
+    const unsigned int replicateSize = _output->info()->dimension(0) - std::min(_input1->info()->dimension(0), _input2->info()->dimension(0));
+    const unsigned int border        = std::min<unsigned int>(num_elems_processed_per_iteration - 1U, replicateSize);
+    return BorderSize(0, border, 0, 0);
+}
+} // namespace arm_compute
\ No newline at end of file