使用SuperPi和Sysbench测试CPU性能

1. 准备工作

准备CentOS、Ubuntu测试机器，CentOS选用7.9版本，Ubuntu采用22.04版本。

1.1. CentOS配置镜像源及安装基础软件

配置国内的公共镜像源，安装基础软件。

1.1.1. 配置镜像源

# 备份
$ sudo mkdir /etc/yum.repos.d/bak
$ sudo mv /etc/yum.repos.d/*.repo /etc/yum.repos.d/bak/
# 配置镜像源文件
$ sudo curl -o /etc/yum.repos.d/CentOS-Base.repo https://mirrors.aliyun.com/repo/Centos-7.repo
$ sudo curl -o /etc/yum.repos.d/epel.repo https://mirrors.aliyun.com/repo/epel-7.repo
# 生成缓存
$ sudo yum makecache

1.1.2. 安装基础依赖包

# 安装开发工具包
$ sudo yum groupinstall -y "Development Tools"
$ sudo yum install -y wget

1.2. Ubuntu配置镜像源及安装基础软件

1.2.1. 配置镜像源

# 替换为阿里云镜像源
$ sudo sed -i 's/https:\/\/archive.ubuntu.com/http:\/\/mirrors.aliyun.com/g' /etc/apt/sources.list

1.2.2. 安装基础依赖包

# 安装构建工具包
$ sudo apt-get install -y build-essential

可用镜像源链接:

• 阿里云：https://developer.aliyun.com/mirror/
• 中科大：https://mirrors.ustc.edu.cn/

2. CPU性能测试

CPU性能测试工具常用的有SuperPI，Sysbench等。

2.1. SuperPI测试

SuperPI是著名的圆周率计算软件，通常用于CPU的单线程性能对比测试。

2.1.1. 安装及测试

# 创建下载安装目录
$ git clone https://github.com/Fibonacci43/SuperPI.git
# 或者使用国内的代码仓库
$ git clone https://gitee.com/lg19891024/SuperPI.git
$ cd SuperPI/
$ ls
CMakeLists.txt  fftsg_h.c  install.txt  Makefile  pi_fftcs.c  README.md  readme.txt
$ gcc -O -funroll-loops -fomit-frame-pointer pi_fftcs.c fftsg_h.c -lm -o pi_css5
# 编译后文件如下，生成了pi_css5
$ ls
CMakeLists.txt  fftsg_h.c  install.txt  Makefile  pi_css5  pi_fftcs.c  README.md  readme.txt

# 执行测试。将计算圆周率小数点后2的26次方位的数值，即64M（6700万）位。
$ ./pi_css5 $((1<<26))    # 等同于 ./pi_css5 67108864

2.1.2. 结果分析

$ ./pi_css5 $((1<<26))
Calculation of PI using FFT and AGM, ver. LG1.1.2-MP1.5.2a.memsave
initializing...
nfft= 16777216
radix= 10000
error_margin= 0.365078
calculating 67108864 digits of PI...
AGM iteration
precision= 48: 12.31 sec
precision= 80: 11.61 sec
precision= 176: 11.63 sec
precision= 352: 12.28 sec
precision= 688: 11.99 sec
precision= 1392: 11.68 sec
precision= 2784: 12.12 sec
precision= 5584: 14.04 sec
precision= 11168: 13.61 sec
precision= 22336: 13.58 sec
precision= 44688: 13.30 sec
precision= 89408: 13.26 sec
precision= 178816: 12.34 sec
precision= 357648: 12.45 sec
precision= 715312: 12.10 sec
precision= 1430640: 12.62 sec
precision= 2861280: 12.06 sec
precision= 5722592: 11.74 sec
precision= 11445200: 11.57 sec
precision= 22890416: 11.84 sec
precision= 45780848: 12.16 sec
precision= 91561728: 11.88 sec
writing pi67108864.txt...
306.04 sec. (real time)

计算的圆周率写入到文件pi67108864.txt。计算67108864位圆周率共耗时306.04s，通过对比时间长短来比较不同机器CPU的单线程性能。

2.2. sysbench

sysbench是一个开源的、模块化的、跨平台的多线程性能测试工具，可以用来进行CPU、内存、磁盘I/O、线程、数据库的性能测试。

2.2.1. 安装

2.2.1.1. 镜像源安装

centos安装sysbench需要配置epel源。

# CentOS安装
sudo yum -y install sysbench

# Ubuntu安装
sudo apt -y install sysbench

2.2.1.2. 源码编译安装

下载编译安装

$ sudo wget https://github.com/akopytov/sysbench/archive/master.zip
$ sudo unzip master.zip
$ sudo cd master/

# 或使用国内代码仓库
$ git clone https://gitee.com/mirrors/sysbench.git
$ cd sysbench/

$ sudo ./autogen.sh
# 如果仅测试cpu性能不涉及mysql，添加下面参数。否则编译配置不通过。
$ sudo ./configure --without-mysql
# 安装
$ sudo make && sudo make install

参数解读：

$ sysbench --help
Usage:
  sysbench [options]... [testname] [command]

Commands implemented by most tests: prepare run cleanup help

General options:
  --threads=N                     number of threads to use [1]
  --events=N                      limit for total number of events [0]
  --time=N                        limit for total execution time in seconds [10]
  --warmup-time=N                 execute events for this many seconds with statistics disabled before the actual benchmark run with statistics enabled [0]
  --forced-shutdown=STRING        number of seconds to wait after the --time limit before forcing shutdown, or 'off' to disable [off]
  --thread-stack-size=SIZE        size of stack per thread [64K]
  --thread-init-timeout=N         wait time in seconds for worker threads to initialize [30]
  --rate=N                        average transactions rate. 0 for unlimited rate [0]
  --report-interval=N             periodically report intermediate statistics with a specified interval in seconds. 0 disables intermediate reports [0]
  --report-checkpoints=[LIST,...] dump full statistics and reset all counters at specified points in time. The argument is a list of comma-separated values representing the amount of time in seconds elapsed from start of test when report checkpoint(s) must be performed. Report checkpoints are off by default. []
  --debug[=on|off]                print more debugging info [off]
  --validate[=on|off]             perform validation checks where possible [off]
  --help[=on|off]                 print help and exit [off]
  --version[=on|off]              print version and exit [off]
  --config-file=FILENAME          File containing command line options
  --luajit-cmd=STRING             perform LuaJIT control command. This option is equivalent to 'luajit -j'. See LuaJIT documentation for more information

Pseudo-Random Numbers Generator options:
  --rand-type=STRING   random numbers distribution {uniform, gaussian, pareto, zipfian} to use by default [uniform]
  --rand-seed=N        seed for random number generator. When 0, the current time is used as an RNG seed. [0]
  --rand-pareto-h=N    shape parameter for the Pareto distribution [0.2]
  --rand-zipfian-exp=N shape parameter (exponent, theta) for the Zipfian distribution [0.8]

Log options:
  --verbosity=N verbosity level {5 - debug, 0 - only critical messages} [3]

  --percentile=N       percentile to calculate in latency statistics (1-100). Use the special value of 0 to disable percentile calculations [95]
  --histogram[=on|off] print latency histogram in report [off]

General database options:

  --db-driver=STRING  specifies database driver to use ('help' to get list of available drivers)
  --db-ps-mode=STRING prepared statements usage mode {auto, disable} [auto]
  --db-debug[=on|off] print database-specific debug information [off]


Compiled-in database drivers:

Compiled-in tests:
  fileio - File I/O test
  cpu - CPU performance test
  memory - Memory functions speed test
  threads - Threads subsystem performance test
  mutex - Mutex performance test

See 'sysbench <testname> help' for a list of options for each test.

2.2.2. 测试

# 查看cpu测试帮助信息
$ sysbench cpu help

常用参数：
--cpu-max-prime: 素数生成数量的上限，默认为10000
--threads: 线程数，默认为1
--time: 运行时间，单位为妙，默认为10。每完成一轮素数计算为1个event，相同时间情况下比较谁的events多。
--events: 上限次数，完成events后，即使时间剩余也停止运行。相同events情况下比较谁的时间小。

# 执行测试
$ sysbench cpu --cpu-max-prime=20000 --threads=2 run

2.2.3. 结果分析

$ sysbench cpu --cpu-max-prime=2000 --threads=2 run
sysbench 1.1.0-2ca9e3f (using bundled LuaJIT 2.1.0-beta3)

Running the test with following options:
Number of threads: 2
Initializing random number generator from current time


Prime numbers limit: 2000

Initializing worker threads...

Threads started!

CPU speed:
    events per second: 14323.35

Throughput:
    events/s (eps):                      14323.3494
    time elapsed:                        10.0002s
    total number of events:              143236

Latency (ms):
         min:                                    0.12
         avg:                                    0.14
         max:                                   12.21
         95th percentile:                        0.16
         sum:                                19917.39

Threads fairness:
    events (avg/stddev):           71618.0000/908.00
    execution time (avg/stddev):   9.9587/0.00

两台服务器进行CPU性能比较，当素数上限和线程数一致时：

• 相同时间，比较events；
• 相同events，比较时间；
• 时间和events都相同，比较标准差。

参考资料：

• 官方仓库及文档：https://github.com/akopytov/sysbench/

本文由mdnice多平台发布