Remove redundant implementations of Add/Sub operators
Allows only implementations where inputs/output are of the same data
type and removes legacy Computer Vision ones.
Signed-off-by: Georgios Pinitas <georgios.pinitas@arm.com>
Change-Id: Ia2b3d23a04236aab682f0c36a1110a30f7c06d1c
Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/5900
Tested-by: Arm Jenkins <bsgcomp@arm.com>
Reviewed-by: Michele Di Giorgio <michele.digiorgio@arm.com>
Comments-Addressed: Arm Jenkins <bsgcomp@arm.com>
diff --git a/src/core/gpu/cl/kernels/ClElementwiseKernel.cpp b/src/core/gpu/cl/kernels/ClElementwiseKernel.cpp
index b645353..f005e92 100644
--- a/src/core/gpu/cl/kernels/ClElementwiseKernel.cpp
+++ b/src/core/gpu/cl/kernels/ClElementwiseKernel.cpp
@@ -127,50 +127,29 @@
ARM_COMPUTE_RETURN_ERROR_ON_DATA_TYPE_CHANNEL_NOT_IN(&src1, 1, DataType::U8, DataType::QASYMM8, DataType::QASYMM8_SIGNED,
DataType::S16, DataType::QSYMM16, DataType::F16,
DataType::S32, DataType::F32);
- ARM_COMPUTE_RETURN_ERROR_ON_F16_UNSUPPORTED(&src2);
- ARM_COMPUTE_RETURN_ERROR_ON_DATA_TYPE_CHANNEL_NOT_IN(&src2, 1, DataType::U8, DataType::QASYMM8, DataType::QASYMM8_SIGNED,
- DataType::S16, DataType::QSYMM16, DataType::F16,
- DataType::S32, DataType::F32);
+ ARM_COMPUTE_RETURN_ERROR_ON_MISMATCHING_DATA_TYPES(&src1, &src2);
- const bool is_quantized = is_data_type_quantized(src1.data_type()) || is_data_type_quantized(src2.data_type());
- if(is_quantized)
+ if(is_data_type_quantized_symmetric(src1.data_type()))
{
- ARM_COMPUTE_RETURN_ERROR_ON_MISMATCHING_DATA_TYPES(&src1, &src2);
-
- if(is_data_type_quantized_symmetric(src1.data_type()))
- {
- const int32_t in1_offset = src1.quantization_info().uniform().offset;
- const int32_t in2_offset = src2.quantization_info().uniform().offset;
- ARM_COMPUTE_RETURN_ERROR_ON_MSG(in1_offset != 0, "For quantized symmetric, offset must be zero");
- ARM_COMPUTE_RETURN_ERROR_ON_MSG(in2_offset != 0, "For quantized symmetric, offset must be zero");
- }
+ const int32_t in1_offset = src1.quantization_info().uniform().offset;
+ const int32_t in2_offset = src2.quantization_info().uniform().offset;
+ ARM_COMPUTE_RETURN_ERROR_ON_MSG(in1_offset != 0, "For quantized symmetric, offset must be zero");
+ ARM_COMPUTE_RETURN_ERROR_ON_MSG(in2_offset != 0, "For quantized symmetric, offset must be zero");
}
const TensorShape out_shape = TensorShape::broadcast_shape(src1.tensor_shape(), src2.tensor_shape());
-
ARM_COMPUTE_RETURN_ERROR_ON_MSG(out_shape.total_size() == 0, "Inputs are not broadcast compatible");
// Validate in case of configured dst
if(dst.total_size() > 0)
{
- ARM_COMPUTE_RETURN_ERROR_ON_F16_UNSUPPORTED(&dst);
- ARM_COMPUTE_RETURN_ERROR_ON_DATA_TYPE_CHANNEL_NOT_IN(&dst, 1, DataType::U8, DataType::QASYMM8, DataType::QASYMM8_SIGNED,
- DataType::S16, DataType::QSYMM16, DataType::F16,
- DataType::S32, DataType::F32);
- ARM_COMPUTE_RETURN_ERROR_ON_MSG((dst.data_type() == DataType::U8) && ((src1.data_type() != DataType::U8) || (src2.data_type() != DataType::U8)),
- "dst can only be U8 if both inputs are U8");
- ARM_COMPUTE_RETURN_ERROR_ON_MSG(detail::have_different_dimensions(out_shape, dst.tensor_shape(), 0),
- "Wrong shape for dst");
+ ARM_COMPUTE_RETURN_ERROR_ON_MISMATCHING_DATA_TYPES(&src1, &dst);
+ ARM_COMPUTE_RETURN_ERROR_ON_MSG(detail::have_different_dimensions(out_shape, dst.tensor_shape(), 0), "Wrong shape for dst");
- if(is_quantized)
+ if(is_data_type_quantized_symmetric(dst.data_type()))
{
- ARM_COMPUTE_RETURN_ERROR_ON_MISMATCHING_DATA_TYPES(&src1, &dst);
-
- if(is_data_type_quantized_symmetric(dst.data_type()))
- {
- const int32_t offset = dst.quantization_info().uniform().offset;
- ARM_COMPUTE_RETURN_ERROR_ON_MSG(offset != 0, "For quantized symmetric, offset must be zero");
- }
+ const int32_t offset = dst.quantization_info().uniform().offset;
+ ARM_COMPUTE_RETURN_ERROR_ON_MSG(offset != 0, "For quantized symmetric, offset must be zero");
}
}
return Status{};
@@ -182,9 +161,7 @@
const unsigned int num_elems_processed_per_iteration = adjust_vec_size(vector_size_byte_opencl / dst.element_size(), dst.dimension(0));
- build_opts.add_option("-DDATA_TYPE_IN1=" + get_cl_type_from_data_type(src1.data_type()));
- build_opts.add_option("-DDATA_TYPE_IN2=" + get_cl_type_from_data_type(src2.data_type()));
- build_opts.add_option("-DDATA_TYPE_OUT=" + get_cl_type_from_data_type(dst.data_type()));
+ build_opts.add_option("-DDATA_TYPE=" + get_cl_type_from_data_type(src1.data_type()));
build_opts.add_option("-DVEC_SIZE_IN1=" + support::cpp11::to_string(src1.dimension(0) == 1 ? 1 : num_elems_processed_per_iteration));
build_opts.add_option("-DVEC_SIZE_IN2=" + support::cpp11::to_string(src2.dimension(0) == 1 ? 1 : num_elems_processed_per_iteration));
build_opts.add_option("-DVEC_SIZE_OUT=" + support::cpp11::to_string(num_elems_processed_per_iteration));
@@ -220,32 +197,7 @@
const std::pair<TensorShape, ValidRegion> broadcast_pair = ITensorInfo::broadcast_shape_and_valid_region(src1, src2);
const TensorShape &out_shape = broadcast_pair.first;
- set_shape_if_empty(dst, out_shape);
-
- if(src1.data_type() == DataType::S16 || src2.data_type() == DataType::S16)
- {
- set_format_if_unknown(dst, Format::S16);
- }
- else if(src1.data_type() == DataType::F16 || src2.data_type() == DataType::F16)
- {
- set_format_if_unknown(dst, Format::F16);
- }
- else if(src1.data_type() == DataType::F32 || src2.data_type() == DataType::F32)
- {
- set_format_if_unknown(dst, Format::F32);
- }
- else if(src1.data_type() == DataType::QASYMM8 || src2.data_type() == DataType::QASYMM8)
- {
- set_data_type_if_unknown(dst, DataType::QASYMM8);
- }
- else if(src1.data_type() == DataType::QASYMM8_SIGNED || src2.data_type() == DataType::QASYMM8_SIGNED)
- {
- set_data_type_if_unknown(dst, DataType::QASYMM8_SIGNED);
- }
- else if(src1.data_type() == DataType::QSYMM16 || src2.data_type() == DataType::QSYMM16)
- {
- set_data_type_if_unknown(dst, DataType::QSYMM16);
- }
+ auto_init_if_empty(dst, out_shape, 1, src1.data_type());
return configure_window_arithmetic_common(dst);
}
@@ -258,7 +210,6 @@
set_shape_if_empty(dst, out_shape);
set_data_type_if_unknown(dst, DataType::U8);
- // The arithmetic utility functions can be share
return configure_window_arithmetic_common(dst);
}
@@ -266,7 +217,9 @@
{
const std::pair<TensorShape, ValidRegion> broadcast_pair = ITensorInfo::broadcast_shape_and_valid_region(src1, src2);
const TensorShape &out_shape = broadcast_pair.first;
+
auto_init_if_empty(dst, out_shape, 1, src1.data_type());
+
return configure_window_arithmetic_common(dst);
}
} // namespace