From bcc90123f0bba8111a913254a805c8466f5fb688 Mon Sep 17 00:00:00 2001 From: luotao1 Date: Thu, 29 Nov 2018 21:13:02 +0800 Subject: [PATCH] speedup box_coder_op for multi-threads test=develop --- paddle/fluid/operators/detection/box_coder_op.h | 6 ++++++ 1 file changed, 6 insertions(+) diff --git a/paddle/fluid/operators/detection/box_coder_op.h b/paddle/fluid/operators/detection/box_coder_op.h index 5ed8520acdd..b2a2bcdce93 100644 --- a/paddle/fluid/operators/detection/box_coder_op.h +++ b/paddle/fluid/operators/detection/box_coder_op.h @@ -43,6 +43,9 @@ class BoxCoderKernel : public framework::OpKernel { const T* prior_box_var_data = nullptr; if (prior_box_var) prior_box_var_data = prior_box_var->data(); +#ifdef PADDLE_WITH_MKLML +#pragma omp parallel for collapse(2) +#endif for (int64_t i = 0; i < row; ++i) { for (int64_t j = 0; j < col; ++j) { T prior_box_width = prior_box_data[j * len + 2] - @@ -96,6 +99,9 @@ class BoxCoderKernel : public framework::OpKernel { const T* prior_box_var_data = nullptr; if (prior_box_var) prior_box_var_data = prior_box_var->data(); +#ifdef PADDLE_WITH_MKLML +#pragma omp parallel for collapse(2) +#endif for (int64_t i = 0; i < row; ++i) { for (int64_t j = 0; j < col; ++j) { size_t offset = i * col * len + j * len; -- GitLab