diff options
author | Felix Thomasmathibalan <felixjohnny.thomasmathibalan@arm.com> | 2023-09-27 17:46:17 +0100 |
---|---|---|
committer | felixjohnny.thomasmathibalan <felixjohnny.thomasmathibalan@arm.com> | 2023-09-28 12:08:05 +0000 |
commit | afd38f0c617d6f89b2b4532c6c44f116617e2b6f (patch) | |
tree | 03bc7d5a762099989b16a656fa8d397b490ed70e /src/cpu/operators/CpuAddMulAdd.cpp | |
parent | bdcb4c148ee2fdeaaddf4cf1e57bbb0de02bb894 (diff) | |
download | ComputeLibrary-afd38f0c617d6f89b2b4532c6c44f116617e2b6f.tar.gz |
Apply clang-format on repository
Code is formatted as per a revised clang format configuration
file(not part of this delivery). Version 14.0.6 is used.
Exclusion List:
- files with .cl extension
- files that are not strictly C/C++ (e.g. Android.bp, Sconscript ...)
And the following directories
- compute_kernel_writer/validation/
- tests/
- include/
- src/core/NEON/kernels/convolution/
- src/core/NEON/kernels/arm_gemm/
- src/core/NEON/kernels/arm_conv/
- data/
There will be a follow up for formatting of .cl files and the
files under tests/ and compute_kernel_writer/validation/.
Signed-off-by: Felix Thomasmathibalan <felixjohnny.thomasmathibalan@arm.com>
Change-Id: Ib7eb1fcf4e7537b9feaefcfc15098a804a3fde0a
Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/10391
Benchmark: Arm Jenkins <bsgcomp@arm.com>
Tested-by: Arm Jenkins <bsgcomp@arm.com>
Reviewed-by: Gunes Bayir <gunes.bayir@arm.com>
Diffstat (limited to 'src/cpu/operators/CpuAddMulAdd.cpp')
-rw-r--r-- | src/cpu/operators/CpuAddMulAdd.cpp | 85 |
1 files changed, 48 insertions, 37 deletions
diff --git a/src/cpu/operators/CpuAddMulAdd.cpp b/src/cpu/operators/CpuAddMulAdd.cpp index 590ee482ca..2f19f2f842 100644 --- a/src/cpu/operators/CpuAddMulAdd.cpp +++ b/src/cpu/operators/CpuAddMulAdd.cpp @@ -21,39 +21,49 @@ * OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE * SOFTWARE. */ +#include "src/cpu/operators/CpuAddMulAdd.h" + #include "arm_compute/core/experimental/Types.h" #include "arm_compute/runtime/NEON/NEScheduler.h" #include "src/common/utils/Log.h" #include "src/core/helpers/MemoryHelpers.h" #include "src/cpu/kernels/CpuAddMulAddKernel.h" -#include "src/cpu/operators/CpuAddMulAdd.h" #include "src/cpu/utils/CpuAuxTensorHandler.h" namespace arm_compute { namespace cpu { -void CpuAddMulAdd::configure(const ITensorInfo *input1, const ITensorInfo *input2, - const ITensorInfo *bn_mul, const ITensorInfo *bn_add, - ITensorInfo *add_output, ITensorInfo *final_output, - ConvertPolicy policy, const ActivationLayerInfo &act_info) +void CpuAddMulAdd::configure(const ITensorInfo *input1, + const ITensorInfo *input2, + const ITensorInfo *bn_mul, + const ITensorInfo *bn_add, + ITensorInfo *add_output, + ITensorInfo *final_output, + ConvertPolicy policy, + const ActivationLayerInfo &act_info) { ARM_COMPUTE_LOG_PARAMS(input1, input2, bn_mul, bn_add, add_output, final_output, policy, act_info); auto k = std::make_unique<kernels::CpuAddMulAddKernel>(); const DataType data_type = input1->data_type(); - if(is_data_type_quantized(data_type)) + if (is_data_type_quantized(data_type)) { _dequantize_bn_mul.configure(bn_mul, &_dequantized_bn_mul); _dequantize_bn_add.configure(bn_add, &_dequantized_bn_add); - k->configure(input1, input2, &_dequantized_bn_mul, &_dequantized_bn_add, add_output, final_output, policy, act_info); + k->configure(input1, input2, &_dequantized_bn_mul, &_dequantized_bn_add, add_output, final_output, policy, + act_info); // Save auxilary memory requirements after configuration - _aux_mem[DequantizedBnMul] = experimental::MemoryInfo(offset_int_vec(DequantizedBnMul), experimental::MemoryLifetime::Temporary, _dequantized_bn_mul.total_size()); - _aux_mem[DequantizedBnAdd] = experimental::MemoryInfo(offset_int_vec(DequantizedBnAdd), experimental::MemoryLifetime::Temporary, _dequantized_bn_add.total_size()); + _aux_mem[DequantizedBnMul] = + experimental::MemoryInfo(offset_int_vec(DequantizedBnMul), experimental::MemoryLifetime::Temporary, + _dequantized_bn_mul.total_size()); + _aux_mem[DequantizedBnAdd] = + experimental::MemoryInfo(offset_int_vec(DequantizedBnAdd), experimental::MemoryLifetime::Temporary, + _dequantized_bn_add.total_size()); } else { @@ -63,13 +73,17 @@ void CpuAddMulAdd::configure(const ITensorInfo *input1, const ITensorInfo *input _kernel = std::move(k); } -Status CpuAddMulAdd::validate(const ITensorInfo *input1, const ITensorInfo *input2, - const ITensorInfo *bn_mul, const ITensorInfo *bn_add, - const ITensorInfo *add_output, const ITensorInfo *final_output, - ConvertPolicy policy, const ActivationLayerInfo &act_info) +Status CpuAddMulAdd::validate(const ITensorInfo *input1, + const ITensorInfo *input2, + const ITensorInfo *bn_mul, + const ITensorInfo *bn_add, + const ITensorInfo *add_output, + const ITensorInfo *final_output, + ConvertPolicy policy, + const ActivationLayerInfo &act_info) { const DataType data_type = input1->data_type(); - if(is_data_type_quantized(data_type)) + if (is_data_type_quantized(data_type)) { TensorInfo dequantized_bn_mul = bn_mul->clone()->set_data_type(DataType::F32); TensorInfo dequantized_bn_add = bn_add->clone()->set_data_type(DataType::F32); @@ -77,11 +91,13 @@ Status CpuAddMulAdd::validate(const ITensorInfo *input1, const ITensorInfo *inpu ARM_COMPUTE_RETURN_ON_ERROR(CpuDequantize::validate(bn_mul, &dequantized_bn_mul)); ARM_COMPUTE_RETURN_ON_ERROR(CpuDequantize::validate(bn_add, &dequantized_bn_add)); - return kernels::CpuAddMulAddKernel::validate(input1, input2, &dequantized_bn_mul, &dequantized_bn_add, add_output, final_output, policy, act_info); + return kernels::CpuAddMulAddKernel::validate(input1, input2, &dequantized_bn_mul, &dequantized_bn_add, + add_output, final_output, policy, act_info); } else { - return kernels::CpuAddMulAddKernel::validate(input1, input2, bn_mul, bn_add, add_output, final_output, policy, act_info); + return kernels::CpuAddMulAddKernel::validate(input1, input2, bn_mul, bn_add, add_output, final_output, policy, + act_info); } } @@ -89,37 +105,32 @@ void CpuAddMulAdd::run(ITensorPack &tensors) { const DataType data_type = tensors.get_const_tensor(TensorType::ACL_SRC_0)->info()->data_type(); - if(is_data_type_quantized(data_type)) + if (is_data_type_quantized(data_type)) { const ITensor *bn_mul = tensors.get_const_tensor(TensorType::ACL_SRC_2); const ITensor *bn_add = tensors.get_const_tensor(TensorType::ACL_SRC_3); - CpuAuxTensorHandler dequantized_bn_mul_handler(offset_int_vec(DequantizedBnMul), _dequantized_bn_mul, tensors, true); - CpuAuxTensorHandler dequantized_bn_add_handler(offset_int_vec(DequantizedBnAdd), _dequantized_bn_add, tensors, true); + CpuAuxTensorHandler dequantized_bn_mul_handler(offset_int_vec(DequantizedBnMul), _dequantized_bn_mul, tensors, + true); + CpuAuxTensorHandler dequantized_bn_add_handler(offset_int_vec(DequantizedBnAdd), _dequantized_bn_add, tensors, + true); - ITensorPack dequantize_mul_pack = - { - { TensorType::ACL_SRC_0, bn_mul }, - { TensorType::ACL_DST_0, dequantized_bn_mul_handler.get() } - }; + ITensorPack dequantize_mul_pack = {{TensorType::ACL_SRC_0, bn_mul}, + {TensorType::ACL_DST_0, dequantized_bn_mul_handler.get()}}; - ITensorPack dequantize_add_pack = - { - { TensorType::ACL_SRC_0, bn_add }, - { TensorType::ACL_DST_0, dequantized_bn_add_handler.get() } - }; + ITensorPack dequantize_add_pack = {{TensorType::ACL_SRC_0, bn_add}, + {TensorType::ACL_DST_0, dequantized_bn_add_handler.get()}}; _dequantize_bn_mul.run(dequantize_mul_pack); _dequantize_bn_add.run(dequantize_add_pack); - ITensorPack add_mul_add_pack = - { - { TensorType::ACL_SRC_0, tensors.get_const_tensor(TensorType::ACL_SRC_0) }, - { TensorType::ACL_SRC_1, tensors.get_const_tensor(TensorType::ACL_SRC_1) }, - { TensorType::ACL_SRC_2, dequantized_bn_mul_handler.get() }, - { TensorType::ACL_SRC_3, dequantized_bn_add_handler.get() }, - { TensorType::ACL_DST_0, tensors.get_tensor(TensorType::ACL_DST_0) }, - { TensorType::ACL_DST_1, tensors.get_tensor(TensorType::ACL_DST_1) }, + ITensorPack add_mul_add_pack = { + {TensorType::ACL_SRC_0, tensors.get_const_tensor(TensorType::ACL_SRC_0)}, + {TensorType::ACL_SRC_1, tensors.get_const_tensor(TensorType::ACL_SRC_1)}, + {TensorType::ACL_SRC_2, dequantized_bn_mul_handler.get()}, + {TensorType::ACL_SRC_3, dequantized_bn_add_handler.get()}, + {TensorType::ACL_DST_0, tensors.get_tensor(TensorType::ACL_DST_0)}, + {TensorType::ACL_DST_1, tensors.get_tensor(TensorType::ACL_DST_1)}, }; NEScheduler::get().schedule_op(_kernel.get(), Window::DimY, _kernel->window(), add_mul_add_pack); |