Stress-ng 的简单学习
背景
想研究一下国产和不同架构,不通型号CPU的算力
也作为后续生产交付的基线准备.
学习各种不同工具进行简要测试.
安装
git clone https://github.com/ColinIanKing/stress-ng.git
cd stress-ng
make
然后就会在当前目录创建一个可执行文件
stress-ng
x86和aarch64是一样的.
主要测试脚本
Run 4 CPU, 2 virtual memory, 1 disk and 8 fork stressors for 2 minutes and print measurements:
stress-ng --cpu 4 --vm 2 --hdd 1 --fork 8 --timeout 2m --metrics
Run matrix stressor on all online CPUs for 60 seconds and measure temperature:
stress-ng --matrix -1 --tz -t 60
Run a mix of 4 I/O stressors and check for changes in disk S.M.A.R.T. metadata:
sudo stress-ng --iomix 4 --smart -t 30s
Benchmark system calls using the VDSO:
stress-ng --vdso 1 -t 5 --metrics
Generate and measure branch misses using perf metrics:
sudo stress-ng --branch 1 --perf -t 10 --stdout | grep Branch
鲲鹏虚拟机的结果
/stress-ng --cpu 4 --vm 2 --hdd 1 --fork 8 --timeout 2m --metrics
stress-ng: info: [1764043] setting to a 120 second (2 mins, 0.00 secs) run per stressor
stress-ng: info: [1764043] dispatching hogs: 4 cpu, 2 vm, 1 hdd, 8 fork
stress-ng: info: [1764043] stressor bogo ops real time usr time sys time bogo ops/s bogo ops/s CPU used per RSS Max
stress-ng: info: [1764043] (secs) (secs) (secs) (real time) (usr+sys time) instance (%) (KB)
stress-ng: info: [1764043] cpu 90568 120.01 476.10 0.25 754.67 190.13 99.23 5376
stress-ng: info: [1764043] vm 4479814 120.00 237.18 1.09 37331.39 18801.41 99.28 150656
stress-ng: info: [1764043] hdd 458752 125.92 15.30 6.96 3643.14 20609.34 17.68 1600
stress-ng: info: [1764043] fork 805043 120.00 88.03 90.05 6708.66 4520.63 18.55 3904
stress-ng: info: [1764043] miscellaneous metrics:
stress-ng: info: [1764043] hdd 5378.02 MB/sec read rate (geometic mean of 1 instances)
stress-ng: info: [1764043] hdd 5385.10 MB/sec write rate (geometic mean of 1 instances)
stress-ng: info: [1764043] hdd 5381.56 MB/sec read/write combined rate (geometic mean of 1 instances)
stress-ng: info: [1764043] successful run completed in 125.92s (2 mins, 5.92 secs)
飞腾2500+物理机
stress-ng: info: [4155956] setting to a 120 second (2 mins, 0.00 secs) run per stressor
stress-ng: info: [4155956] dispatching hogs: 4 cpu, 2 vm, 1 hdd, 8 fork
stress-ng: info: [4155956] stressor bogo ops real time usr time sys time bogo ops/s bogo ops/s CPU used per RSS Max
stress-ng: info: [4155956] (secs) (secs) (secs) (real time) (usr+sys time) instance (%) (KB)
stress-ng: info: [4155956] cpu 54036 120.08 480.27 0.04 449.99 112.50 100.00 6336
stress-ng: info: [4155956] vm 2327541 120.00 238.08 1.92 19395.68 9698.38 99.99 151104
stress-ng: info: [4155956] hdd 1706914 120.00 80.84 37.43 14223.72 14432.65 98.55 3776
stress-ng: info: [4155956] fork 799938 120.00 494.91 581.67 6666.13 743.03 112.14 4032
stress-ng: info: [4155956] miscellaneous metrics:
stress-ng: info: [4155956] hdd 2663.89 MB/sec read rate (geometic mean of 1 instances)
stress-ng: info: [4155956] hdd 3252.81 MB/sec write rate (geometic mean of 1 instances)
stress-ng: info: [4155956] hdd 2929.55 MB/sec read/write combined rate (geometic mean of 1 instances)
stress-ng: info: [4155956] successful run completed in 120.17s (2 mins, 0.17 secs)
飞腾2000+物理机
stress-ng: info: [3916081] setting to a 120 second (2 mins, 0.00 secs) run per stressor
stress-ng: info: [3916081] dispatching hogs: 4 cpu, 2 vm, 1 hdd, 8 fork
stress-ng: info: [3916081] stressor bogo ops real time usr time sys time bogo ops/s bogo ops/s CPU used per RSS Max
stress-ng: info: [3916081] (secs) (secs) (secs) (real time) (usr+sys time) instance (%) (KB)
stress-ng: info: [3916081] cpu 52085 120.10 479.29 0.11 433.70 108.65 99.80 6336
stress-ng: info: [3916081] vm 1881017 120.00 237.33 1.89 15674.69 7863.20 99.67 150912
stress-ng: info: [3916081] hdd 1614848 120.01 73.40 46.24 13455.87 13497.90 99.69 3776
stress-ng: info: [3916081] fork 1314325 120.00 522.63 579.41 10952.70 1192.63 114.80 4032
stress-ng: info: [3916081] miscellaneous metrics:
stress-ng: info: [3916081] hdd 2012.78 MB/sec read rate (geometic mean of 1 instances)
stress-ng: info: [3916081] hdd 2487.28 MB/sec write rate (geometic mean of 1 instances)
stress-ng: info: [3916081] hdd 2226.36 MB/sec read/write combined rate (geometic mean of 1 instances)
stress-ng: info: [3916081] successful run completed in 120.20s (2 mins, 0.20 secs)
飞腾2500 虚拟机
stress-ng: info: [814485] setting to a 120 second (2 mins, 0.00 secs) run per stressor
stress-ng: info: [814485] dispatching hogs: 4 cpu, 2 vm, 1 hdd, 8 fork
stress-ng: info: [814485] stressor bogo ops real time usr time sys time bogo ops/s bogo ops/s CPU used per RSS Max
stress-ng: info: [814485] (secs) (secs) (secs) (real time) (usr+sys time) instance (%) (KB)
stress-ng: info: [814485] cpu 39040 120.05 354.04 0.61 325.19 110.08 73.85 4736
stress-ng: info: [814485] vm 1612136 120.00 180.96 1.82 13434.14 8820.19 76.16 146880
stress-ng: info: [814485] hdd 1189177 120.02 57.64 28.67 9907.93 13777.31 71.91 1664
stress-ng: info: [814485] fork 310675 120.00 100.49 111.94 2588.94 1462.49 22.13 3904
stress-ng: info: [814485] miscellaneous metrics:
stress-ng: info: [814485] hdd 1788.38 MB/sec read rate (geometic mean of 1 instances)
stress-ng: info: [814485] hdd 2009.70 MB/sec write rate (geometic mean of 1 instances)
stress-ng: info: [814485] hdd 1893.48 MB/sec read/write combined rate (geometic mean of 1 instances)
stress-ng: info: [814485] successful run completed in 120.14s (2 mins, 0.14 secs)
海光7285
stress-ng: info: [2005838] setting to a 120 second (2 mins, 0.00 secs) run per stressor
stress-ng: info: [2005838] dispatching hogs: 4 cpu, 2 vm, 1 hdd, 8 fork
stress-ng: info: [2005838] stressor bogo ops real time usr time sys time bogo ops/s bogo ops/s CPU used per RSS Max
stress-ng: info: [2005838] (secs) (secs) (secs) (real time) (usr+sys time) instance (%) (KB)
stress-ng: info: [2005838] cpu 370238 120.00 479.22 0.00 3085.25 772.58 99.84 6008
stress-ng: info: [2005838] vm 5513614 120.00 237.47 2.11 45946.46 23013.67 99.82 149276
stress-ng: info: [2005838] hdd 2173494 120.05 86.50 32.82 18105.25 18215.65 99.39 2168
stress-ng: info: [2005838] fork 1593832 120.00 524.99 444.59 13281.92 1643.83 101.00 4332
stress-ng: info: [2005838] miscellaneous metrics:
stress-ng: info: [2005838] hdd 9165.57 MB/sec read rate (geometic mean of 1 instances)
stress-ng: info: [2005838] hdd 3352.49 MB/sec write rate (geometic mean of 1 instances)
stress-ng: info: [2005838] hdd 4897.99 MB/sec read/write combined rate (geometic mean of 1 instances)
stress-ng: info: [2005838] successful run completed in 120.10s (2 mins, 0.10 secs)
Golden 6150
stress-ng: info: [76329] setting to a 120 second (2 mins, 0.00 secs) run per stressor
stress-ng: info: [76329] dispatching hogs: 4 cpu, 2 vm, 1 hdd, 8 fork
stress-ng: info: [76329] stressor bogo ops real time usr time sys time bogo ops/s bogo ops/s CPU used per RSS Max
stress-ng: info: [76329] (secs) (secs) (secs) (real time) (usr+sys time) instance (%) (KB)
stress-ng: info: [76329] cpu 557584 120.00 479.66 0.34 4646.48 1161.64 100.00 3344
stress-ng: info: [76329] vm 3156543 120.02 229.66 10.38 26299.99 13149.95 100.00 148164
stress-ng: info: [76329] hdd 1828690 120.20 46.95 72.79 15213.18 15272.39 99.61 744
stress-ng: info: [76329] fork 551818 120.00 51.47 910.99 4598.47 573.34 100.26 964
stress-ng: info: [76329] miscellaneous metrics:
stress-ng: info: [76329] hdd 3928.41 MB/sec read rate (geometic mean of 1 instances)
stress-ng: info: [76329] hdd 1295.59 MB/sec write rate (geometic mean of 1 instances)
stress-ng: info: [76329] hdd 1945.16 MB/sec read/write combined rate (geometic mean of 1 instances)
stress-ng: info: [76329] successful run completed in 120.21s (2 mins, 0.21 secs)
Silver 4216
stress-ng: info: [5609] setting to a 120 second (2 mins, 0.00 secs) run per stressor
stress-ng: info: [5609] dispatching hogs: 4 cpu, 2 vm, 1 hdd, 8 fork
stress-ng: info: [5609] stressor bogo ops real time usr time sys time bogo ops/s bogo ops/s CPU used per RSS Max
stress-ng: info: [5609] (secs) (secs) (secs) (real time) (usr+sys time) instance (%) (KB)
stress-ng: info: [5609] cpu 445180 120.00 479.96 0.01 3709.81 927.51 99.99 6092
stress-ng: info: [5609] vm 2430504 120.01 237.08 2.92 20253.01 10126.97 100.00 149300
stress-ng: info: [5609] hdd 1322664 120.90 42.32 44.13 10939.86 15300.42 71.50 2228
stress-ng: info: [5609] fork 1788238 120.00 572.10 357.04 14901.97 1924.63 96.78 2240
stress-ng: info: [5609] miscellaneous metrics:
stress-ng: info: [5609] hdd 6977.70 MB/sec read rate (geometic mean of 1 instances)
stress-ng: info: [5609] hdd 1507.84 MB/sec write rate (geometic mean of 1 instances)
stress-ng: info: [5609] hdd 2465.46 MB/sec read/write combined rate (geometic mean of 1 instances)
stress-ng: info: [5609] successful run completed in 120.90s (2 mins, 0.90 secs)
Stress-ng 的简单学习的更多相关文章
- OI数学 简单学习笔记
基本上只是整理了一下框架,具体的学习给出了个人认为比较好的博客的链接. PART1 数论部分 最大公约数 对于正整数x,y,最大的能同时整除它们的数称为最大公约数 常用的:\(lcm(x,y)=xy\ ...
- 百度首席科学家 Andrew Ng谈深度学习的挑战和未来(转载)
转载:http://www.csdn.net/article/2014-07-10/2820600 人工智能被认为是下一个互联网大事件,当下,谷歌.微软.百度等知名的高科技公司争相投入资源,占领深度学 ...
- Log4j简单学习笔记
log4j结构图: 结构图展现出了log4j的主结构.logger:表示记录器,即数据来源:appender:输出源,即输出方式(如:控制台.文件...)layout:输出布局 Logger机滤器:常 ...
- shiro简单学习的简单总结
权限和我有很大渊源. 培训时候的最后一个项目是OA,权限那块却不知如何入手,最后以不是我写的那个模块应付面试. 最开始的是使用session装载用户登录信息,使用简单权限拦截器做到权限控制,利用资源文 ...
- CentOS 简单学习 firewalld的使用
1. centos7 开始 使用firewalld 代替了 iptables 命令工具为 firewall-cmd 帮助信息非常长,简单放到文末 2. 简单使用 首先开启 httpd 一般都自带安装了 ...
- Windows 下 Docker 的简单学习使用过程之一 dockertoolbox
1. Windows 下面运行 Docker 的两个主要工具1): Docker for Windows2): DockerToolbox区别:Docker For Windows 可以理解为是新一代 ...
- 在MVC中实现和网站不同服务器的批量文件下载以及NPOI下载数据到Excel的简单学习
嘿嘿,我来啦,最近忙啦几天,使用MVC把应该实现的一些功能实现了,说起来做项目,实属感觉蛮好的,即可以学习新的东西,又可以增加自己之前知道的知识的巩固,不得不说是双丰收啊,其实这周来就开始面对下载在挣 ...
- Linux——帮助命令简单学习笔记
Linux帮助命令简单学习笔记: 一: 命令名称:man 命令英文原意:manual 命令所在路径:/usr/bin/man 执行权限:所有用户 语法:man [命令或配置文件] 功能描述:获得帮助信 ...
- mongodb,redis简单学习
2.mongodb安装配置简单学习 配置好数据库路径就可以mongo命令执行交互操作了:先将服务器开起来:在开个cmd执行交互操作 ...
- html css的简单学习（三）
html css的简单学习(三) 前端开发工具:Dreamweaver.Hbuilder.WebStorm.Sublime.PhpStorm...=========================== ...
随机推荐
- 数仓性能调优：row_number() over(p)-rn=1性能瓶颈发现和改写套路
本文分享自华为云社区<GaussDB(DWS)性能调优:row_number() over(p)-rn=1性能瓶颈发现和改写套路>,作者:Zawami . 1.改写场景 本套路应用于子查询 ...
- LiteOS：剖析时间管理模块源代码
摘要:Huawei LiteOS的时间管理模块以系统时钟为基础,分为2部分,一部分是SysTick中断,为任务调度提供必要的时钟节拍:另外一部分是,给应用程序提供所有和时间有关的服务,如时间转换.统计 ...
- NDPQ(NDP+PQ)，定义分布式数据库新方向
摘要:云服务提供商构建新的云原生关系数据库系统,专门为云基础架构设计,通常采用将计算和存储分离到独立扩展的分布式层的设计. 本文分享自华为云社区<性能提升100倍!GaussDB(for MyS ...
- 火山引擎 VeDI 零售行业解决方案 聚焦精准营销场景提升品牌转化
更多技术交流.求职机会,欢迎关注字节跳动数据平台微信公众号,回复[1]进入官方交流群 你知道,为了能让你买到合适的商品,品牌商们有多努力吗? 精准营销并不是一个新词,但近年来,随着营销渠道/平台的更加 ...
- QA: Solon 怎么输出下载文件流？
Solon 的Mvc可以直接返回 DownloadedFile 或者 File 对象 @Mapping("/demo/down") @Controller public class ...
- PPT 图片框架排版万能能公式
图片作用 提升设计感 辅助表达 传递情感 如何选择一张高大上的图片? 星空.地球.城市.海洋.线条.粒子.山脉.壁纸(系统.手机厂商千挑万选的) https://cn.bing.com/images ...
- 数组递增的判断【python实现】
有时候需要对某一组数组的数据进行判断是否 递增 的场景,比如我在开发一些体育动作场景下,某些肢体动作是需要持续朝着垂直方向向上变化,那么z轴的值是会累增的.同理,逆向考虑,递减就是它的对立面. 下面是 ...
- Python | BitMap算法及其实现
BitMap概述 本文介绍 BitMap 算法的应用背景,算法思想和相关实现细节. 概括而言,BitMap 主要用来解决海量数据中元素查询,去重.以及排序等问题.这里对海量数据场景的强调,似乎暗示了这 ...
- Problem B - Card Constructions （构造）
题意: 你可以用图示的方法建造金字塔,但是每一次都要建最大的金字塔,问最后能建几个金字塔. 思路: 我们可以发现对于每一个金字塔都是两边增加了两天边,然后中间行数− 1 -1−1个三角形,所以就可以求 ...
- 解决SUM函数返回为NULL
解决SUM函数返回为NULL SUM函数的作用:计算某一字段中所有行的数值和, 使用SUM函数进行对符合条件的结果行数进行求和. 问题产生: sum 求和时会对 null 进行过滤,不计算,但如果没有 ...