Skip to content

Commit

Permalink
Merge pull request PaddlePaddle#163 from lidanqing-intel/fix-doc
Browse files Browse the repository at this point in the history
fix gru int8 doc
  • Loading branch information
juncaipeng authored Aug 31, 2021
2 parents 592e1b4 + 2449ff3 commit e90ac0a
Showing 1 changed file with 0 additions and 11 deletions.
11 changes: 0 additions & 11 deletions c++/x86_gru_int8/README.md
Original file line number Diff line number Diff line change
Expand Up @@ -65,18 +65,7 @@ config.SetModel(FLAGS_model_dir); // Load no-combined model
config.SetCpuMathLibraryNumThreads(FLAGS_threads);
config.EnableMKLDNN();
config.SwitchIrOptim();
config.EnableMemoryOptim();
```
**Note:**
- 如果在 VNNI支持的 CPU 上预测保存好的 INT8 模型,则以下无需设置,因为 **1.2** 准备的INT8 模型经过量化和fusion优化。
```
# 预测保存好的int8模型,以下3行可以删除
config.EnableMKLDNN();
config.SwitchIrOptim();
config.EnableMemoryOptim();
```
- 如果只测试FP32模型,则上述设置全部打开。


## 2 编译运行示例

Expand Down

0 comments on commit e90ac0a

Please sign in to comment.