diff options
Diffstat (limited to 'src/cpu/kernels/fuse_batch_normalization/nhwc/neon/impl.h')
| -rw-r--r-- | src/cpu/kernels/fuse_batch_normalization/nhwc/neon/impl.h | 118 |
1 files changed, 116 insertions, 2 deletions
diff --git a/src/cpu/kernels/fuse_batch_normalization/nhwc/neon/impl.h b/src/cpu/kernels/fuse_batch_normalization/nhwc/neon/impl.h index 3b813132b1..6f0386276f 100644 --- a/src/cpu/kernels/fuse_batch_normalization/nhwc/neon/impl.h +++ b/src/cpu/kernels/fuse_batch_normalization/nhwc/neon/impl.h @@ -1,5 +1,5 @@ /* - * Copyright (c) 2021-2022 Arm Limited. + * Copyright (c) 2018-2023 Arm Limited. * * SPDX-License-Identifier: MIT * @@ -33,8 +33,122 @@ namespace cpu { template <typename T> void fused_batch_normalization_dwc_nhwc(const ITensor *dwc_weights, const ITensor *dwc_bias, ITensor *fused_weights, ITensor *fused_bias, - const ITensor *bn_mean, const ITensor *bn_var, const ITensor *bn_beta, const ITensor *bn_gamma, float epsilon, const Window &window); + const ITensor *bn_mean, const ITensor *bn_var, const ITensor *bn_beta, const ITensor *bn_gamma, float epsilon, const Window &window) +{ + using ScalarType = T; + const int size = 16 / dwc_weights->info()->element_size(); + using ExactTagType = typename wrapper::traits::neon_bitvector_tag_t<T, wrapper::traits::BitWidth::W128>; + + const bool run_in_place_weights = (fused_weights == nullptr) || (fused_weights == dwc_weights); + const bool run_in_place_bias = (fused_bias == nullptr) || (dwc_bias != nullptr && fused_bias == dwc_bias); + + // Set build options + Window win = window; + win.set(Window::DimX, Window::Dimension(0, 1, 1)); + + const int window_step_x = size; + const auto window_start_x = static_cast<int>(window.x().start()); + const auto window_end_x = static_cast<int>(window.x().end()); + + Iterator dwc_w_in(dwc_weights, win); + Iterator dwc_w_out(run_in_place_weights ? dwc_weights : fused_weights, win); + + const auto dwc_bias_in = (dwc_bias != nullptr ? reinterpret_cast<ScalarType *>(dwc_bias->ptr_to_element(Coordinates(0, 0))) : nullptr); + auto dwc_bias_out = (run_in_place_bias ? dwc_bias_in : reinterpret_cast<ScalarType *>(fused_bias->ptr_to_element(Coordinates(0, 0)))); + + const auto input_mean = reinterpret_cast<const ScalarType *>(bn_mean->ptr_to_element(Coordinates(0, 0))); + const auto input_var = reinterpret_cast<const ScalarType *>(bn_var->ptr_to_element(Coordinates(0, 0))); + const auto input_gamma = (bn_gamma != nullptr) ? reinterpret_cast<const ScalarType *>(bn_gamma->ptr_to_element(Coordinates(0, 0))) : nullptr; + const auto input_beta = (bn_beta != nullptr) ? reinterpret_cast<const ScalarType *>(bn_beta->ptr_to_element(Coordinates(0, 0))) : nullptr; + + auto mean_vec = wrapper::vdup_n(ScalarType(0), ExactTagType{}); + auto var_vec = wrapper::vdup_n(ScalarType(0), ExactTagType{}); + auto gamma_vec = wrapper::vdup_n(ScalarType(1), ExactTagType{}); + auto beta_vec = wrapper::vdup_n(ScalarType(0), ExactTagType{}); + auto rvar_vec = wrapper::vdup_n(ScalarType(0), ExactTagType{}); + auto dwc_bias_vec = wrapper::vdup_n(ScalarType(0), ExactTagType{}); + const auto epsilon_vec = wrapper::vdup_n(ScalarType(epsilon), ExactTagType{}); + + auto gamma = ScalarType(1.0); + auto beta = ScalarType(0.0); + auto dwc_bias_in_scalar = ScalarType(0); + + execute_window_loop(win, [&](const Coordinates & id) + { + int x = window_start_x; + for(; x <= (window_end_x - window_step_x); x += window_step_x) + { + var_vec = wrapper::vloadq(input_var + x); + if(input_gamma != nullptr) + { + gamma_vec = wrapper::vloadq(input_gamma + x); + } + + if((id[2] == 0) && (id[1] == 0)) + { + mean_vec = wrapper::vloadq(input_mean + x); + + // Construct vectors + if(input_beta != nullptr) + { + beta_vec = wrapper::vloadq(input_beta + x); + } + + if(dwc_bias_in != nullptr) + { + dwc_bias_vec = wrapper::vloadq(dwc_bias_in + x); + } + + auto dwc_bias_tmp_vec = wrapper::vmul(wrapper::vsub(dwc_bias_vec, mean_vec), wrapper::vinvsqrt(wrapper::vadd(var_vec, epsilon_vec))); + dwc_bias_tmp_vec = wrapper::vadd(wrapper::vmul(dwc_bias_tmp_vec, gamma_vec), beta_vec); + wrapper::vstore(dwc_bias_out + x, dwc_bias_tmp_vec); + } + + auto dwc_w_in_ptr = reinterpret_cast<const ScalarType *>(dwc_w_in.ptr()); + auto dwc_w_out_ptr = reinterpret_cast<ScalarType *>(dwc_w_out.ptr()); + + auto wn = wrapper::vloadq(dwc_w_in_ptr + x); + rvar_vec = wrapper::vinvsqrt(wrapper::vadd(var_vec, epsilon_vec)); + wn = wrapper::vmul(wn, rvar_vec); + wn = wrapper::vmul(wn, gamma_vec); + + // Store results + wrapper::vstore(dwc_w_out_ptr + x, wn); + } + + // Compute left-over elements + for(; x < window_end_x; ++x) + { + auto var = input_var[x]; + if(input_gamma != nullptr) + { + gamma = input_gamma[x]; + } + + if(id[2] == 0 && id[1] == 0) + { + auto mean = input_mean[x]; + if(input_beta != nullptr) + { + beta = input_beta[x]; + } + if(dwc_bias_in != nullptr) + { + dwc_bias_in_scalar = dwc_bias_in[x]; + } + + auto dwc_bias_tmp_scalar = (dwc_bias_in_scalar - mean) / std::sqrt(var + ScalarType(epsilon)); + dwc_bias_out[x] = (dwc_bias_tmp_scalar * gamma) + beta; + } + + const auto dwc_w_in_ptr = reinterpret_cast<const ScalarType *>(dwc_w_in.ptr()); + auto dwc_w_out_ptr = reinterpret_cast<ScalarType *>(dwc_w_out.ptr()); + *(dwc_w_out_ptr + x) = *(dwc_w_in_ptr + x) / std::sqrt(var + ScalarType(epsilon)) * gamma; + } + }, + dwc_w_in, dwc_w_out); +} } // namespace cpu } // namespace arm_compute #endif //SRC_CORE_NEON_KERNELS_FUSE_BATCH_NORMALIZATION_IMPL_H |