diff --git a/PyTorch/built-in/cv/classification/ResNet50_for_PyTorch/requirements.txt b/PyTorch/built-in/cv/classification/ResNet50_for_PyTorch/requirements.txt
index fbefb37521789025484a96ad28dc194561590cfe..f5f431ac5b9c83bc32048524a218dd6d86225423 100644
--- a/PyTorch/built-in/cv/classification/ResNet50_for_PyTorch/requirements.txt
+++ b/PyTorch/built-in/cv/classification/ResNet50_for_PyTorch/requirements.txt
@@ -1,3 +1,5 @@
 apex
 torchvision
-opencv-python
\ No newline at end of file
+opencv-python
+
+
diff --git a/PyTorch/contrib/cv/classification/InceptionV3_ID1596_for_PyTorch/test/train_full_8p.sh b/PyTorch/contrib/cv/classification/InceptionV3_ID1596_for_PyTorch/test/train_full_8p.sh
index e80116c3416e1f49037188137f67d33f194ae069..1c846c057643290ac5ade7122cba33b0c1ed6d92 100644
--- a/PyTorch/contrib/cv/classification/InceptionV3_ID1596_for_PyTorch/test/train_full_8p.sh
+++ b/PyTorch/contrib/cv/classification/InceptionV3_ID1596_for_PyTorch/test/train_full_8p.sh
@@ -66,29 +66,60 @@ if [ x"${etp_flag}" != x"true" ];then
     source ${test_path_dir}/env_npu.sh
 fi
 
-python3 ./main-8p.py \
-    -a inception_v3 \
-    --amp \
-    --loss-scale 128 \
-    --data ${data_path} \
-    --addr=$(hostname -I |awk '{print $1}') \
-    --seed=49 \
-    --workers=128 \
-    --learning-rate=${learning_rate} \
-    --mom=0.9 \
-    --weight-decay=1.0e-04  \
-    --print-freq=30 \
-    --dist-url='tcp://127.0.0.1:50000' \
-    --dist-backend='hccl' \
-    --multiprocessing-distributed \
-    --world-size=1 \
-    --rank=0 \
-    --device='npu' \
-    --epochs=${train_epochs} \
-    --checkpoint-freq=-1 \
-    --label-smoothing=0.1 \
-    --batch-size=${batch_size} > ${test_path_dir}/output/$ASCEND_DEVICE_ID/train_$ASCEND_DEVICE_ID.log 2>&1 &
-
+KERNEL_NUM=$(($(nproc)/8))
+for i in $(seq 0 7)
+do
+  if [ $(uname -m) = "aarch64" ]
+  then
+    PID_START=$((KERNEL_NUM*i))
+    PID_END=$((PID_START + KERNEL_NUM - 1))
+    taskset -c $PID_START-$PID_END python3 ./main-8p.py \
+      -a inception_v3 \
+      --amp \
+      --loss-scale 128 \
+      --data ${data_path} \
+      --addr=$(hostname -I |awk '{print $1}') \
+      --seed=49 \
+      --workers=128 \
+      --learning-rate=${learning_rate} \
+      --mom=0.9 \
+      --weight-decay=1.0e-04  \
+      --print-freq=30 \
+      --dist-url='tcp://127.0.0.1:50000' \
+      --dist-backend='hccl' \
+      --multiprocessing-distributed \
+      --world-size=1 \
+      --rank=0 \
+      --device='npu' \
+      --epochs=${train_epochs} \
+      --checkpoint-freq=-1 \
+      --label-smoothing=0.1 \
+      --batch-size=${batch_size} > ${test_path_dir}/output/$ASCEND_DEVICE_ID/train_$ASCEND_DEVICE_ID.log 2>&1 &
+  else
+    python3 ./main-8p.py \
+      -a inception_v3 \
+      --amp \
+      --loss-scale 128 \
+      --data ${data_path} \
+      --addr=$(hostname -I |awk '{print $1}') \
+      --seed=49 \
+      --workers=128 \
+      --learning-rate=${learning_rate} \
+      --mom=0.9 \
+      --weight-decay=1.0e-04  \
+      --print-freq=30 \
+      --dist-url='tcp://127.0.0.1:50000' \
+      --dist-backend='hccl' \
+      --multiprocessing-distributed \
+      --world-size=1 \
+      --rank=0 \
+      --device='npu' \
+      --epochs=${train_epochs} \
+      --checkpoint-freq=-1 \
+      --label-smoothing=0.1 \
+      --batch-size=${batch_size} > ${test_path_dir}/output/$ASCEND_DEVICE_ID/train_$ASCEND_DEVICE_ID.log 2>&1 &
+  fi
+done
 wait
 
 # 训练结束时间，不需要修改
diff --git a/PyTorch/contrib/cv/classification/InceptionV3_ID1596_for_PyTorch/test/train_performance_8p.sh b/PyTorch/contrib/cv/classification/InceptionV3_ID1596_for_PyTorch/test/train_performance_8p.sh
index 3d1080d971bbaf6a3cc97a6cedf4aee96fd611a9..b59f64220449fbf5440c14cd112f01b947d2e451 100644
--- a/PyTorch/contrib/cv/classification/InceptionV3_ID1596_for_PyTorch/test/train_performance_8p.sh
+++ b/PyTorch/contrib/cv/classification/InceptionV3_ID1596_for_PyTorch/test/train_performance_8p.sh
@@ -66,27 +66,58 @@ if [ x"${etp_flag}" != x"true" ];then
     source ${test_path_dir}/env_npu.sh
 fi
 
-python3 ./main-8p.py \
-    -a inception_v3 \
-    --amp \
-    --loss-scale 128 \
-    --data ${data_path} \
-    --addr=$(hostname -I |awk '{print $1}') \
-    --seed=49 \
-    --workers=128 \
-    --learning-rate=${learning_rate} \
-    --mom=0.9 \
-    --weight-decay=1.0e-04  \
-    --print-freq=30 \
-    --dist-url='tcp://127.0.0.1:50000' \
-    --dist-backend='hccl' \
-    --multiprocessing-distributed \
-    --world-size=1 \
-    --rank=0 \
-    --device='npu' \
-    --epochs=${train_epochs} \
-    --label-smoothing=0.1 \
-    --batch-size=${batch_size} > ${test_path_dir}/output/$ASCEND_DEVICE_ID/train_$ASCEND_DEVICE_ID.log 2>&1 &
+KERNEL_NUM=$(($(nproc)/8))
+for i in $(seq 0 7)
+do
+  if [ $(uname -m) = "aarch64" ]
+  then
+    PID_START=$((KERNEL_NUM*i))
+    PID_END=$((PID_START + KERNEL_NUM - 1))
+    taskset -c $PID_START-$PID_END python3 ./main-8p.py \
+      -a inception_v3 \
+      --amp \
+      --loss-scale 128 \
+      --data ${data_path} \
+      --addr=$(hostname -I |awk '{print $1}') \
+      --seed=49 \
+      --workers=128 \
+      --learning-rate=${learning_rate} \
+      --mom=0.9 \
+      --weight-decay=1.0e-04  \
+      --print-freq=30 \
+      --dist-url='tcp://127.0.0.1:50000' \
+      --dist-backend='hccl' \
+      --multiprocessing-distributed \
+      --world-size=1 \
+      --rank=0 \
+      --device='npu' \
+      --epochs=${train_epochs} \
+      --label-smoothing=0.1 \
+      --batch-size=${batch_size} > ${test_path_dir}/output/$ASCEND_DEVICE_ID/train_$ASCEND_DEVICE_ID.log 2>&1 &
+  else
+    python3 ./main-8p.py \
+      -a inception_v3 \
+      --amp \
+      --loss-scale 128 \
+      --data ${data_path} \
+      --addr=$(hostname -I |awk '{print $1}') \
+      --seed=49 \
+      --workers=128 \
+      --learning-rate=${learning_rate} \
+      --mom=0.9 \
+      --weight-decay=1.0e-04  \
+      --print-freq=30 \
+      --dist-url='tcp://127.0.0.1:50000' \
+      --dist-backend='hccl' \
+      --multiprocessing-distributed \
+      --world-size=1 \
+      --rank=0 \
+      --device='npu' \
+      --epochs=${train_epochs} \
+      --label-smoothing=0.1 \
+      --batch-size=${batch_size} > ${test_path_dir}/output/$ASCEND_DEVICE_ID/train_$ASCEND_DEVICE_ID.log 2>&1 &
+  fi
+done
 wait
 
 # 训练结束时间，不需要修改
diff --git a/PyTorch/contrib/cv/semantic_segmentation/FCN8s/test/train_full_8p.sh b/PyTorch/contrib/cv/semantic_segmentation/FCN8s/test/train_full_8p.sh
index 4a924972ff9a77cce7a89868d612d4c7330becb8..bc86be9a2362f12945cd3a1a80adbae4a4d399b3 100644
--- a/PyTorch/contrib/cv/semantic_segmentation/FCN8s/test/train_full_8p.sh
+++ b/PyTorch/contrib/cv/semantic_segmentation/FCN8s/test/train_full_8p.sh
@@ -81,6 +81,7 @@ echo "end_time: ${end_time}"
 e2e_time=$(( $end_time - $start_time ))
 
 FPS=`grep -a 'fps: '  $cur_path/output/${Network}/${ASCEND_DEVICE_ID}/train_${ASCEND_DEVICE_ID}.log|awk -F "fps: " '{print $NF}'|awk -F "," '{print $1}'|awk 'NR==1{max=$1;next}{max=max>$1?max:$1}END{print max}'`
+FPS='${FPS}'*8
 echo "Final Performance images/sec : $FPS"
 
 #输出训练精度,需要模型审视修改
diff --git a/PyTorch/contrib/cv/semantic_segmentation/FCN8s/test/train_performance_8p.sh b/PyTorch/contrib/cv/semantic_segmentation/FCN8s/test/train_performance_8p.sh
index 900e7652b8e8e647876d3cb29392b1de98dbea4d..ee843b84b7e8b7eed9fbdcc927f0efe90eebfb72 100644
--- a/PyTorch/contrib/cv/semantic_segmentation/FCN8s/test/train_performance_8p.sh
+++ b/PyTorch/contrib/cv/semantic_segmentation/FCN8s/test/train_performance_8p.sh
@@ -81,6 +81,7 @@ echo "end_time: ${end_time}"
 e2e_time=$(( $end_time - $start_time ))
 
 FPS=`grep -a 'fps: '  $cur_path/output/${Network}/${ASCEND_DEVICE_ID}/train_${ASCEND_DEVICE_ID}.log|awk -F "fps: " '{print $NF}'|awk -F "," '{print $1}'|awk 'NR==1{max=$1;next}{max=max>$1?max:$1}END{print max}'`
+FPS='${FPS}'*8
 echo "Final Performance images/sec : $FPS"
 
 #输出训练精度,需要模型审视修改