diff --git a/PyTorch/contrib/cv/semantic_segmentation/FFHQ_ID2978_for_Pytorch/readme.md b/PyTorch/contrib/cv/semantic_segmentation/FFHQ_ID2978_for_Pytorch/readme.md
index 1b5653c87677baa033be72cfedeea61d26248d40..0475466d671a7e138f162ca545ed4b4b8ed7816a 100644
--- a/PyTorch/contrib/cv/semantic_segmentation/FFHQ_ID2978_for_Pytorch/readme.md
+++ b/PyTorch/contrib/cv/semantic_segmentation/FFHQ_ID2978_for_Pytorch/readme.md
@@ -41,3 +41,10 @@ python3.7 run_deeplab.py
 
 - 原始图像、GPU分割、NPU分割效果对比如下：
 <br> 调用deeplab模型后，原始图片被分割为：背景、皮肤、鼻子、眼睛等部分。
+
+### 性能达标
+
+| Name   | FPS  |
+| ------ | ---- |
+| GPU-1p | 20.9 |
+| NPU-1p | 21.5 |
diff --git a/PyTorch/contrib/cv/semantic_segmentation/FFHQ_ID2978_for_Pytorch/run_deeplab.py b/PyTorch/contrib/cv/semantic_segmentation/FFHQ_ID2978_for_Pytorch/run_deeplab.py
index 41de24d34f976d9844f0f08275fd5349a0bc63d7..a40089791e6093a70c2b57682431e410842e8404 100644
--- a/PyTorch/contrib/cv/semantic_segmentation/FFHQ_ID2978_for_Pytorch/run_deeplab.py
+++ b/PyTorch/contrib/cv/semantic_segmentation/FFHQ_ID2978_for_Pytorch/run_deeplab.py
@@ -21,6 +21,7 @@ import torch
 import torch.nn as nn
 from pdb import set_trace as st
 from PIL import Image
+from tqdm import tqdm
 from torchvision import transforms
 import deeplab
 from data_loader import CelebASegmentation
@@ -35,6 +36,12 @@ dataset_root = "ffhq_aging128x128"
 assert torch.npu.is_available()
 assert os.path.isdir(dataset_root)
 
+option = {}
+option["ACL_OP_COMPILER_CACHE_MODE"] = "enable" # cache功能启用
+option["ACL_OP_COMPILER_CACHE_DIR"] = "./my_kernel_meta" # cache所在文件夹
+print("option:",option)
+torch.npu.set_option(option)
+
 dataset = CelebASegmentation(dataset_root, crop_size=513)
 print("len(dataset)", len(dataset))
 print("dataset.CLASSES", dataset.CLASSES)
@@ -58,7 +65,7 @@ device = "npu"
 model = model.to(device)
 model.eval()
 
-for i in range(len(dataset)):
+for i in tqdm(range(len(dataset))):
     inputs = dataset[i]
     # inputs = inputs.npu()
     inputs = inputs.unsqueeze(0).to(device)