From ad7515d231acb075a9585e52f257373b1a1b5d1f Mon Sep 17 00:00:00 2001 From: Michalis Spyrou Date: Fri, 24 Jul 2020 00:02:23 +0100 Subject: COMPMID-3385: Async support to CLArithmetic* kernels/functions Pt.1 Signed-off-by: Michalis Spyrou Change-Id: I94007565e688f8a0aead4f14c9fc30bfd9f9f7eb Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/3613 Tested-by: Arm Jenkins Reviewed-by: Georgios Pinitas --- arm_compute/runtime/CL/functions/CLLSTMLayer.h | 169 ++++++++++++------------- 1 file changed, 84 insertions(+), 85 deletions(-) (limited to 'arm_compute/runtime/CL/functions/CLLSTMLayer.h') diff --git a/arm_compute/runtime/CL/functions/CLLSTMLayer.h b/arm_compute/runtime/CL/functions/CLLSTMLayer.h index e5733cd784..abfcc3a62f 100644 --- a/arm_compute/runtime/CL/functions/CLLSTMLayer.h +++ b/arm_compute/runtime/CL/functions/CLLSTMLayer.h @@ -27,7 +27,6 @@ #include "arm_compute/runtime/IFunction.h" #include "arm_compute/core/CL/kernels/CLCopyKernel.h" -#include "arm_compute/core/CL/kernels/CLElementwiseOperationKernel.h" #include "arm_compute/core/CL/kernels/CLMemsetKernel.h" #include "arm_compute/core/CL/kernels/CLPixelWiseMultiplicationKernel.h" #include "arm_compute/core/Types.h" @@ -201,90 +200,90 @@ public: void prepare() override; private: - MemoryGroup _memory_group; - CLFullyConnectedLayer _fully_connected_input_gate; - CLArithmeticAddition _accum_input_gate1; - CLSaturatedArithmeticOperationKernel _subtract_input_gate; - CLPixelWiseMultiplicationKernel _pixelwise_mul_input_gate; - CLActivationLayer _activation_input_gate; - CLFullyConnectedLayer _fully_connected_forget_gate; - CLArithmeticAddition _accum_forget_gate1; - CLPixelWiseMultiplicationKernel _pixelwise_mul_forget_gate; - CLActivationLayer _activation_forget_gate; - CLFullyConnectedLayer _fully_connected_cell_state; - CLGEMM _gemm_cell_state1; - CLTransposeKernel _transpose_cell_state; - CLSaturatedArithmeticOperationKernel _accum_cell_state1; - CLSaturatedArithmeticOperationKernel _accum_cell_state2; - CLPixelWiseMultiplicationKernel _pixelwise_mul_cell_state1; - CLActivationLayer _activation_cell_state; - CLActivationLayer _cell_clip; - CLPixelWiseMultiplicationKernel _pixelwise_mul_cell_state2; - CLFullyConnectedLayer _fully_connected_output; - CLPixelWiseMultiplicationKernel _pixelwise_mul_output_state1; - CLArithmeticAddition _accum_output1; - CLActivationLayer _activation_output; - CLActivationLayer _activation_output_state; - CLPixelWiseMultiplicationKernel _pixelwise_mul_output_state2; - CLFullyConnectedLayer _fully_connected_output_state; - CLActivationLayer _projection_clip; - CLCopyKernel _copy_cell_state; - CLCopyKernel _copy_output; - CLConcatenateLayer _concat_scratch_buffer; - CLConcatenateLayer _concat_inputs_forget_gate; - CLConcatenateLayer _concat_weights_forget_gate; - CLConcatenateLayer _concat_weights_input_gate; - CLConcatenateLayer _concat_weights_output; - CLMemsetKernel _ones_memset_kernel; - CLMeanStdDevNormalizationLayer _mean_std_norm_input_gate; - CLPixelWiseMultiplicationKernel _pixelwise_mul_input_gate_coeff; - CLSaturatedArithmeticOperationKernel _accum_input_gate_bias; - CLMeanStdDevNormalizationLayer _mean_std_norm_forget_gate; - CLPixelWiseMultiplicationKernel _pixelwise_mul_forget_gate_coeff; - CLSaturatedArithmeticOperationKernel _accum_forget_gate_bias; - CLMeanStdDevNormalizationLayer _mean_std_norm_cell_gate; - CLPixelWiseMultiplicationKernel _pixelwise_mul_cell_gate_coeff; - CLSaturatedArithmeticOperationKernel _accum_cell_gate_bias; - CLMeanStdDevNormalizationLayer _mean_std_norm_output_gate; - CLPixelWiseMultiplicationKernel _pixelwise_mul_output_gate_coeff; - CLSaturatedArithmeticOperationKernel _accum_output_gate_bias; - CLTensor _input_gate_out1; - CLTensor _input_gate_out2; - CLTensor _input_gate_out3; - CLTensor _input_gate_out4; - CLTensor _forget_gate_out1; - CLTensor _forget_gate_out2; - CLTensor _forget_gate_out3; - CLTensor _forget_gate_out4; - CLTensor _forget_gate_out5; - CLTensor _forget_gate_out6; - CLTensor _cell_state_out1; - CLTensor _cell_state_out2; - CLTensor _cell_state_out3; - CLTensor _cell_state_out4; - CLTensor _cell_state_out5; - CLTensor _output1; - CLTensor _output2; - CLTensor _output3; - CLTensor _output4; - CLTensor _cell_state_activation; - CLTensor _output_state1; - CLTensor _ones; - CLTensor _input_layer_norm_out1; - CLTensor _input_layer_norm_out2; - CLTensor _forget_layer_norm_out1; - CLTensor _forget_layer_norm_out2; - CLTensor _cell_layer_norm_out1; - CLTensor _cell_layer_norm_out2; - CLTensor _output_layer_norm_out1; - CLTensor _output_layer_norm_out2; - bool _run_peephole_opt; - bool _run_cifg_opt; - bool _perform_cell_clipping; - bool _has_projection_weights; - bool _perform_projection_clipping; - bool _is_prepared; - bool _is_layer_norm_lstm; + MemoryGroup _memory_group; + CLFullyConnectedLayer _fully_connected_input_gate; + CLArithmeticAddition _accum_input_gate1; + CLArithmeticSubtraction _subtract_input_gate; + CLPixelWiseMultiplicationKernel _pixelwise_mul_input_gate; + CLActivationLayer _activation_input_gate; + CLFullyConnectedLayer _fully_connected_forget_gate; + CLArithmeticAddition _accum_forget_gate1; + CLPixelWiseMultiplicationKernel _pixelwise_mul_forget_gate; + CLActivationLayer _activation_forget_gate; + CLFullyConnectedLayer _fully_connected_cell_state; + CLGEMM _gemm_cell_state1; + CLTransposeKernel _transpose_cell_state; + CLArithmeticAddition _accum_cell_state1; + CLArithmeticAddition _accum_cell_state2; + CLPixelWiseMultiplicationKernel _pixelwise_mul_cell_state1; + CLActivationLayer _activation_cell_state; + CLActivationLayer _cell_clip; + CLPixelWiseMultiplicationKernel _pixelwise_mul_cell_state2; + CLFullyConnectedLayer _fully_connected_output; + CLPixelWiseMultiplicationKernel _pixelwise_mul_output_state1; + CLArithmeticAddition _accum_output1; + CLActivationLayer _activation_output; + CLActivationLayer _activation_output_state; + CLPixelWiseMultiplicationKernel _pixelwise_mul_output_state2; + CLFullyConnectedLayer _fully_connected_output_state; + CLActivationLayer _projection_clip; + CLCopyKernel _copy_cell_state; + CLCopyKernel _copy_output; + CLConcatenateLayer _concat_scratch_buffer; + CLConcatenateLayer _concat_inputs_forget_gate; + CLConcatenateLayer _concat_weights_forget_gate; + CLConcatenateLayer _concat_weights_input_gate; + CLConcatenateLayer _concat_weights_output; + CLMemsetKernel _ones_memset_kernel; + CLMeanStdDevNormalizationLayer _mean_std_norm_input_gate; + CLPixelWiseMultiplicationKernel _pixelwise_mul_input_gate_coeff; + CLArithmeticAddition _accum_input_gate_bias; + CLMeanStdDevNormalizationLayer _mean_std_norm_forget_gate; + CLPixelWiseMultiplicationKernel _pixelwise_mul_forget_gate_coeff; + CLArithmeticAddition _accum_forget_gate_bias; + CLMeanStdDevNormalizationLayer _mean_std_norm_cell_gate; + CLPixelWiseMultiplicationKernel _pixelwise_mul_cell_gate_coeff; + CLArithmeticAddition _accum_cell_gate_bias; + CLMeanStdDevNormalizationLayer _mean_std_norm_output_gate; + CLPixelWiseMultiplicationKernel _pixelwise_mul_output_gate_coeff; + CLArithmeticAddition _accum_output_gate_bias; + CLTensor _input_gate_out1; + CLTensor _input_gate_out2; + CLTensor _input_gate_out3; + CLTensor _input_gate_out4; + CLTensor _forget_gate_out1; + CLTensor _forget_gate_out2; + CLTensor _forget_gate_out3; + CLTensor _forget_gate_out4; + CLTensor _forget_gate_out5; + CLTensor _forget_gate_out6; + CLTensor _cell_state_out1; + CLTensor _cell_state_out2; + CLTensor _cell_state_out3; + CLTensor _cell_state_out4; + CLTensor _cell_state_out5; + CLTensor _output1; + CLTensor _output2; + CLTensor _output3; + CLTensor _output4; + CLTensor _cell_state_activation; + CLTensor _output_state1; + CLTensor _ones; + CLTensor _input_layer_norm_out1; + CLTensor _input_layer_norm_out2; + CLTensor _forget_layer_norm_out1; + CLTensor _forget_layer_norm_out2; + CLTensor _cell_layer_norm_out1; + CLTensor _cell_layer_norm_out2; + CLTensor _output_layer_norm_out1; + CLTensor _output_layer_norm_out2; + bool _run_peephole_opt; + bool _run_cifg_opt; + bool _perform_cell_clipping; + bool _has_projection_weights; + bool _perform_projection_clipping; + bool _is_prepared; + bool _is_layer_norm_lstm; }; } // namespace arm_compute #endif /* ARM_COMPUTE_CLLSTMLAYER_H */ -- cgit v1.2.1