Describe the current behavior / 问题描述 (Mandatory / 必填)

densenet121 imagenet2012数据集，修改了batch_size（与竞品对齐），在ascend环境，pynative模式，MS与竞品性能比<100%
模型地址：https://gitee.com/mindspore/models/tree/master/research/cv/densenet

Environment / 环境信息 (Mandatory / 必填)

• Hardware Environment(Ascend/GPU/CPU) / 硬件环境:

Please delete the backend not involved / 请删除不涉及的后端:
/device ascend

• Software Environment / 软件环境 (Mandatory / 必填):
  -- MindSpore version (e.g., 1.7.0.Bxxx) :
  -- Python version (e.g., Python 3.7.5) :
  -- OS platform and distribution (e.g., Linux Ubuntu 16.04):
  -- GCC/Compiler version (if compiled from source):
  ascend（8p）mindspore:2.0.0.20221220,commit_id:470b760e
  ascend(单卡)：mindspore:2.0.0.20221206,commit_id:0a257187
• Excute Mode / 执行模式 (Mandatory / 必填)(PyNative/Graph):

Please delete the mode not involved / 请删除不涉及的模式:
/mode pynative
/mode graph

Related testcase / 关联用例 (Mandatory / 必填)

用例仓地址：solution_test/remaining/test_scripts/mindspore/net/densenet121
用例：
ms_densnet121_check_fps_1p.py
ms_densnet121_check_loss_8p.py

Steps to reproduce the issue / 重现步骤 (Mandatory / 必填)

1. get code from models
2. cd models/research/cv/densenet
3. 将 densenet121_config.yaml中的per_batch_size改为256，train.py中context.GRAPH_MODE改为pynative
4. sh scripts/run_distribute_train.sh 8 /data/hccl.json densenet121 imagenet /path/to/dataset ckpt_file

Describe the expected behavior / 预期结果 (Mandatory / 必填)

ms/竞品性能比 >=100%

Related log / screenshot / 日志 / 截图 (Mandatory / 必填)

Ascend:
竞品：单卡 PT1.8+910 BATCH_SIZE:256
性能：1145.63
mindspore：单卡
graph：1288
pynative：485.51

性能比：ms/pytorch=485.51/1145.63=42.38%

竞品：8卡 PT1.8+910 BATCH_SIZE:2048
性能：7505.33
mindspore：8卡
graph：10464.62
pynative：2031.14

性能比：ms/pytorch=2031.14/7505.33=27.06%

Special notes for this issue/备注 (Optional / 选填)

走给褚金锦