提交 03106471 编写于 作者: B baojun 提交者: tensor-tang

Minor ngraph update (#1933)

* remove KMP_BLOCKTIME setting

* remove duplicate

* update instruction
上级 9cd7d08b
...@@ -16,7 +16,7 @@ Only support Adam optimizer yet. ...@@ -16,7 +16,7 @@ Only support Adam optimizer yet.
Short description of aforementioned steps: Short description of aforementioned steps:
## 1. Install PaddlePaddle ## 1. Install PaddlePaddle
Follow PaddlePaddle [installation instruction](https://github.com/PaddlePaddle/models/tree/develop/fluid/PaddleCV/image_classification#installation) to install PaddlePaddle. If you build PaddlePaddle yourself, please use the following cmake arguments and ensure to set `-DWITH_NGRAPH=ON`. Follow PaddlePaddle [installation instruction](https://github.com/PaddlePaddle/models/tree/develop/fluid/PaddleCV/image_classification#installation) to install PaddlePaddle. If you [build from source](https://github.com/PaddlePaddle/FluidDoc/blob/develop/doc/fluid/beginners_guide/install/compile/compile_Ubuntu_en.md), please use the following cmake arguments and ensure to set `-DWITH_NGRAPH=ON`.
``` ```
cmake .. -DCMAKE_BUILD_TYPE=Release -DWITH_GPU=OFF -DWITH_MKL=ON -DWITH_MKLDNN=ON -DWITH_NGRAPH=ON cmake .. -DCMAKE_BUILD_TYPE=Release -DWITH_GPU=OFF -DWITH_MKL=ON -DWITH_MKLDNN=ON -DWITH_NGRAPH=ON
``` ```
...@@ -29,10 +29,9 @@ export FLAGS_use_ngraph=true ...@@ -29,10 +29,9 @@ export FLAGS_use_ngraph=true
export OMP_NUM_THREADS=<num_cpu_cores> export OMP_NUM_THREADS=<num_cpu_cores>
``` ```
Optional exports for better performance: If multiple threads are used, you may export the following for better performance:
``` ```
export KMP_AFFINITY=granularity=fine,compact,1,0 export KMP_AFFINITY=granularity=fine,compact,1,0
export KMP_BLOCKTIME=1
``` ```
## 3. How the benchmark script might be run. ## 3. How the benchmark script might be run.
......
...@@ -201,7 +201,6 @@ python train.py \ ...@@ -201,7 +201,6 @@ python train.py \
# --total_images=1281167 \ # --total_images=1281167 \
# --class_dim=1000 \ # --class_dim=1000 \
# --image_shape=3,224,224 \ # --image_shape=3,224,224 \
# --lr_strategy=none \
# --lr=0.001 \ # --lr=0.001 \
# --num_epochs=120 \ # --num_epochs=120 \
# --with_mem_opt=False \ # --with_mem_opt=False \
......
Markdown is supported
0% .
You are about to add 0 people to the discussion. Proceed with caution.
先完成此消息的编辑!
想要评论请 注册