diff options
| author | Georgios Pinitas <georgios.pinitas@arm.com> | 2017-07-03 18:17:28 +0100 |
|---|---|---|
| committer | Anthony Barbier <anthony.barbier@arm.com> | 2018-09-17 14:15:39 +0100 |
| commit | f0dea703ce3e2b465e79298bca95c4952a60f608 (patch) | |
| tree | 487c9ab15ea8cef9d57aa3256599cde616310eeb /src/core/CL/kernels/CLArithmeticAdditionKernel.cpp | |
| parent | 1dad50ea7b662e7c07a0a06bbbd175fd1bcf9bfc (diff) | |
| download | ComputeLibrary-f0dea703ce3e2b465e79298bca95c4952a60f608.tar.gz | |
COMPMID-417: Auto configuration for Add/Sub/Mul Neon/CL.
Change-Id: I3580de76bc53d342b53443d1077b1407d75a672a
Reviewed-on: http://mpd-gerrit.cambridge.arm.com/79570
Tested-by: Kaizen <jeremy.johnson+kaizengerrit@arm.com>
Reviewed-by: Michele DiGiorgio <michele.digiorgio@arm.com>
Reviewed-by: Moritz Pflanzer <moritz.pflanzer@arm.com>
Diffstat (limited to 'src/core/CL/kernels/CLArithmeticAdditionKernel.cpp')
| -rw-r--r-- | src/core/CL/kernels/CLArithmeticAdditionKernel.cpp | 29 |
1 files changed, 23 insertions, 6 deletions
diff --git a/src/core/CL/kernels/CLArithmeticAdditionKernel.cpp b/src/core/CL/kernels/CLArithmeticAdditionKernel.cpp index aaa62d0268..0cb0847784 100644 --- a/src/core/CL/kernels/CLArithmeticAdditionKernel.cpp +++ b/src/core/CL/kernels/CLArithmeticAdditionKernel.cpp @@ -48,9 +48,32 @@ CLArithmeticAdditionKernel::CLArithmeticAdditionKernel() void CLArithmeticAdditionKernel::configure(const ICLTensor *input1, const ICLTensor *input2, ICLTensor *output, ConvertPolicy policy) { + ARM_COMPUTE_ERROR_ON_NULLPTR(input1, input2, output); + + // 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); + } + else if(input1->info()->data_type() == DataType::F16 && input2->info()->data_type() == DataType::F16) + { + set_format_if_unknown(*output->info(), Format::F16); + } + } + + ARM_COMPUTE_ERROR_ON_MISMATCHING_SHAPES(input1, input2, output); ARM_COMPUTE_ERROR_ON_DATA_TYPE_CHANNEL_NOT_IN(input1, 1, DataType::U8, DataType::S16, DataType::F16, DataType::F32); ARM_COMPUTE_ERROR_ON_DATA_TYPE_CHANNEL_NOT_IN(input2, 1, DataType::U8, DataType::S16, DataType::F16, DataType::F32); ARM_COMPUTE_ERROR_ON_DATA_TYPE_CHANNEL_NOT_IN(output, 1, DataType::U8, DataType::S16, DataType::F16, DataType::F32); + ARM_COMPUTE_ERROR_ON_MSG(output->info()->data_type() == DataType::U8 && (input1->info()->data_type() != DataType::U8 || input2->info()->data_type() != DataType::U8), + "Output can only be U8 if both inputs are U8"); _input1 = input1; _input2 = input2; @@ -58,12 +81,6 @@ void CLArithmeticAdditionKernel::configure(const ICLTensor *input1, const ICLTen const bool has_float_out = is_data_type_float(output->info()->data_type()); - // Check for invalid combination - if(output->info()->data_type() == DataType::U8 && (input1->info()->data_type() != DataType::U8 || input2->info()->data_type() != DataType::U8)) - { - ARM_COMPUTE_ERROR("You called with the wrong data types."); - } - // Set kernel build options std::set<std::string> build_opts; build_opts.emplace((policy == ConvertPolicy::WRAP || has_float_out) ? "-DWRAP" : "-DSATURATE"); |