[ST][MS][NET][resnet50 imagenet][910 1p/8p]FPS[14529] can not reach 17000

name	about	labels
Bug Report	Use this template for reporting a bug	kind/bug

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

resnet50、resnet50_boost、resnet50_ge 网络在910环境训练imagenet2012数据集性能劣化
resnet50_ge 8p训练性能14530/fps达不到17000
resnet50_ge 1p训练性能1824/fps达不到2100
resnet50_boost 8p训练性能22367/fps达不到26000
resnet50_boost 1p训练性能2818/fps达不到3400
resnet50 8p训练性能14529/fps达不到17000
resnet50 1p训练性能1824/fps达不到2100

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) :r1.8 B010 commit_id:42306df4
-- 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):
run包:HiAI/HISI_C82/20220518
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 graph

Related testcase / 关联用例 (Mandatory / 必填)

test_ms_resnet50_imagenet_ge_train_check_loss_910_8p_0001.py
test_ms_resnet50_imagenet_ge_train_check_perf_910_1p_0002.py
test_ms_resnet50_imagenet_boost_train_check_fps_910_1p_0002.py
test_ms_resnet50_imagenet_boost_train_check_loss_910_8p_0003.py
test_ms_resnet50_imagenet_train_check_loss_910_8p_0002.py
test_ms_resnet50_imagenet_train_check_perf_910_1p_0003.py

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

cd solution_test/cases/02network/00cv/resnet50/train
pytest -s test_ms_resnet50_imagenet_train_check_loss_910_8p_0002.py

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

网络训练成功
resnet50_ge 8p训练性能能达到17000
resnet50_ge 1p训练性能能达到2100
resnet50_boost 8p训练性能能达到26000
resnet50_boost 1p训练性能能达到3400
resnet50 8p训练性能能达到17000
resnet50 1p训练性能能达到2100

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

epoch time: 145087.241 ms, per step time: 290.174 ms
epoch time: 70476.110 ms, per step time: 140.952 ms
epoch time: 70477.973 ms, per step time: 140.956 ms
epoch time: 70476.777 ms, per step time: 140.954 ms
epoch time: 70473.858 ms, per step time: 140.948 ms

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

走给曹杰文

Please assign maintainer to check this issue.
请为此issue分配处理人。
@zhongjicheng

Please add labels (comp or sig), also you can visit https://gitee.com/mindspore/community/blob/master/sigs/dx/docs/labels.md to find more.
为了让代码尽快被审核，请您为Pull Request打上 组件(comp)或兴趣组(sig) 标签，打上标签的PR可直接推送给责任人进行审核。
更多的标签可以查看https://gitee.com/mindspore/community/blob/master/sigs/dx/docs/labels.md
以组件相关代码提交为例，如果你提交的是data组件代码，你可以这样评论：
//comp/data
当然你也可以邀请data SIG组来审核代码，可以这样写：
//sig/data
另外你还可以给这个PR标记类型，例如是bugfix或者是特性需求：
//kind/bug or //kind/feature
恭喜你，你已经学会了使用命令来打标签，接下来就在下面的评论里打上标签吧！

卷积和BN的算子融合失败

输入图片说明

@liangzelang 请适配新run包ub算子融合

Appearance & Root Cause
问题：resnet50_ge 8p训练性能不达标
原因：融合算子编译接口变更，导致算子融合失败。

Fix Solution
解决方法：适配融合算子编译接口
https://e.gitee.com/mind_spore/repos/mindspore/mindspore/pulls/35111

回归版本：2022-05-30 r1.8 commit_id:5afb7c9d
回归步骤：参考issue复现步骤
基本功能：readme已更新
输入图片说明
测试结论：回归通过
回归人员：zhongjicheng
回归时间： 2022-06-06

GVP MindSpore / mindspore

内容风险标识