08
2025
05
21:50:12

UNIXBENCH测试

UNXIBENCH5

一、            概述

   UNIXBENCH也是linuxunix测试系统基准性能的经典工具,UNIXBENCH测试包含了系统调用、读写、进程、2D3D、管道、运算、C库等系统基准性能,它的优点在于提供了对系统性能的一种评价体系,为系统评分,如此方便对系统作对比测试;但UNIXBENCH在网络性能测试欠缺。

二、            测试内容

   unixbench5测试内容包含了system2d3dmisc测试项。system测试系统的一些基本功能;2d测试2D图形加速;3d测试3D加速;misc测试诸如字符,整型等测试。详情见下面。

  system:

   dhry2reg        Dhrystone 2 using register variables

   whetstone-double Double-Precision Whetstone

   syscall         System Call Overhead

   pipe            Pipe Throughput

   context1        Pipe-based Context Switching

   spawn           Process Creation

   execl           Execl Throughput

   fstime-w        File Write 1024 bufsize 2000 maxblocks

   fstime-r        File Read 1024 bufsize 2000 maxblocks

   fstime          File Copy 1024 bufsize 2000 maxblocks

   fsbuffer-w      File Write 256 bufsize 500 maxblocks

   fsbuffer-r      File Read 256 bufsize 500 maxblocks

   fsbuffer        File Copy 256 bufsize 500 maxblocks

   fsdisk-w        File Write 4096 bufsize 8000 maxblocks

   fsdisk-r        File Read 4096 bufsize 8000 maxblocks

   fsdisk          File Copy 4096 bufsize 8000 maxblocks

   shell1          Shell Scripts (1 concurrent) (runs "looper 60 multi.sh 1")

   shell8          Shell Scripts (8 concurrent) (runs "looper 60 multi.sh 8")

   shell16         Shell Scripts (8 concurrent) (runs "looper 60 multi.sh 16")

2d:

   2d-rects        2D graphics: rectangles

   2d-lines        2D graphics: lines

   2d-circle       2D graphics: circles

   2d-ellipse      2D graphics: ellipses

   2d-shapes       2D graphics: polygons

   2d-aashapes     2D graphics: aa polygons

   2d-polys        2D graphics: complex polygons

   2d-text         2D graphics: text

   2d-blit         2D graphics: images and blits

   2d-window       2D graphics: windows

 

  3d:

   ubgears         3D graphics: gears

 

  misc:

   C               C Compiler Throughput ("looper 60 $cCompiler cctest.c")

   arithoh         Arithoh (huh?)

   short           Arithmetic Test (short) (this is arith.c configured for

                    "short" variables; ditto for the ones below)

   int             Arithmetic Test (int)

   long            Arithmetic Test (long)

   float           Arithmetic Test (float)

   double          Arithmetic Test (double)

   dc              Dc: sqrt(2) to 99 decimal places (runs

                    "looper 30 dc < dc.dat", using your system's copy of "dc")

   hanoi           Recursion Test -- Tower of Hanoi

   grep            Grep for a string in a large file, using your system's

                    copy of "grep"

   sysexec         Exercise fork() and exec().

 

 

 

三、           安装

   unixbench5包含systemgraphic测试,如果你需要测试graphic,则需要修改Makefile,不要注释掉"GRAPHIC_TESTS = defined,同时需要系统提供x11perf命令gl_glibs库。

   进入源包,执行

   make

四、           快速使用

  执行system测试方法:Run

  执行graphic测试方法:Run graphics

  执行system,graphics测试方法:Run gindex

 

五、           详细使用方法

1.   Run用法

    Run [ -q | -v ] [-i <n> ] [-c <n> [-c <n> ...]] [test ...]

选项说明:

-q 

不显示测试过程

-v

显示测试过程

-i <count>

执行次数,最低3次,默认10

-c <n>

每次测试并行ncopies(并行任务)

备注-c选项可以用来执行多次,如:

   Run -c 1 -c 4表示执行两次,第一次单个copies,第二次4copies的测试任务。

test

arithmetic      Runs arithoh, short, int, long, float, double,

                    and whetstone-double

   dhry            Alias for dhry2reg

   dhrystone       Alias for dhry2reg

   whets           Alias for whetstone-double

   whetstone       Alias for whetstone-double

   load            Runs shell1, shell8, and shell16

   misc            Runs C, dc, and hanoi

   speed           Runs the arithmetic and system groups

   oldsystem       Runs execl, fstime, fsbuffer, fsdisk, pipe, context1,

                    spawn, and syscall

   system          Runs oldsystem plus shell1, shell8, and shell16

   fs              Runs fstime-w, fstime-r, fstime, fsbuffer-w,

                    fsbuffer-r, fsbuffer, fsdisk-w, fsdisk-r, and fsdisk

   shell           Runs shell1, shell8, and shell16

 

   index           Runs the tests which constitute the official index:

                    the oldsystem group, plus dhry2reg, whetstone-double,

                    shell1, and shell8

                    See "The BYTE Index" below for more information.

   graphics        Runs the tests which constitute the graphics index:

                    2d-rects, 2d-ellipse, 2d-aashapes, 2d-text, 2d-blit,

                    2d-window, and ubgears

   gindex          Runs the index and graphics groups, to generate both

                    sets of index results

 

   all             Runs all tests

 

 

2.   测试说明

   对于多cpu系统的性能测试策略,需要统计单任务,多任务及其并行的性能增强。

4cpuPC为例,需要测试两次,4CPU就是要并行执行4copies

      Run -q -c 1 -c 4

3.   测试结果:

 BYTE UNIX Benchmarks (Version 5.1.2)

 

  System: localhost.localdomain: GNU/Linux

   OS: GNU/Linux -- 2.6.36.3newspec+ -- #5 SMP PREEMPT Wed Nov 30 13:46:30 CST 2011

  Machine: mips64 (mips64)

  Language: en_US.utf8 (charmap="UTF-8", collate="UTF-8")

   CPU 0: ICT Loongson-3A V0.5 FPU V0.1 (0.0 bogomips)

         

   CPU 1: ICT Loongson-3A V0.5 FPU V0.1 (0.0 bogomips)

         

   CPU 2: ICT Loongson-3A V0.5 FPU V0.1 (0.0 bogomips)                        

         

   CPU 3: ICT Loongson-3A V0.5 FPU V0.1 (0.0 bogomips)

         

  12:07:50 up 3:27,  3 users,  load average: 1.61, 1.07, 0.94; runlevel 5

 

------------------------------------------------------------------------

Benchmark Run:   4月 17 2012 12:07:50 - 12:35:51

4 CPUs in system; running 1 parallel copy of tests

 

Dhrystone 2 using register variables       1820525.3 lps  (10.0 s, 7 samples)

Double-Precision Whetstone                     352.7 MWIPS (10.0 s, 7 samples)

Execl Throughput                               739.7 lps  (29.8 s, 2 samples)

File Copy 1024 bufsize 2000 maxblocks        75780.3 KBps  (30.0 s, 2 samples)

File Copy 256 bufsize 500 maxblocks          27969.7 KBps  (30.0 s, 2 samples)

File Copy 4096 bufsize 8000 maxblocks        89566.6 KBps  (30.0 s, 2 samples)

Pipe Throughput                             208836.4 lps  (10.0 s, 7 samples)

Pipe-based Context Switching                 65757.7 lps  (10.0 s, 7 samples)

Process Creation                              2263.6 lps  (30.0 s, 2 samples)

Shell Scripts (1 concurrent)                   923.0 lpm  (60.1 s, 2 samples)

Shell Scripts (8 concurrent)                   206.6 lpm  (60.2 s, 2 samples)

System Call Overhead                        299206.2 lps  (10.0 s, 7 samples)

 

System Benchmarks Index Values              BASELINE      RESULT   INDEX

Dhrystone 2 using register variables        116700.0   1820525.3   156.0

Double-Precision Whetstone                      55.0       352.7    64.1

Execl Throughput                                43.0       739.7   172.0

File Copy 1024 bufsize 2000 maxblocks         3960.0     75780.3   191.4

File Copy 256 bufsize 500 maxblocks           1655.0     27969.7   169.0

File Copy 4096 bufsize 8000 maxblocks         5800.0     89566.6   154.4

Pipe Throughput                              12440.0    208836.4   167.9

Pipe-based Context Switching                  4000.0     65757.7   164.4

Process Creation                               126.0      2263.6   179.7

Shell Scripts (1 concurrent)                    42.4       923.0   217.7

Shell Scripts (8 concurrent)                     6.0       206.6   344.4

System Call Overhead                         15000.0    299206.2   199.5

                                                                  ========

System Benchmarks Index Score                                        171.3

 

------------------------------------------------------------------------

Benchmark Run:   4月 17 2012 12:35:51 - 13:04:03

4 CPUs in system; running 4 parallel copies of tests

 

Dhrystone 2 using register variables       7135020.6 lps  (10.0 s, 7 samples)

Double-Precision Whetstone                    1387.4 MWIPS (10.0 s, 7 samples)

Execl Throughput                              2658.2 lps  (29.8 s, 2 samples)

File Copy 1024 bufsize 2000 maxblocks       102436.7 KBps  (30.0 s, 2 samples)

File Copy 256 bufsize 500 maxblocks          33850.5 KBps  (30.0 s, 2 samples)

File Copy 4096 bufsize 8000 maxblocks       121124.7 KBps  (30.0 s, 2 samples)

Pipe Throughput                             815411.4 lps  (10.0 s, 7 samples)

Pipe-based Context Switching                255881.9 lps  (10.0 s, 7 samples)

Process Creation                              5750.4 lps  (30.0 s, 2 samples)

Shell Scripts (1 concurrent)                  1617.7 lpm  (60.1 s, 2 samples)

Shell Scripts (8 concurrent)                   159.2 lpm  (60.7 s, 2 samples)

System Call Overhead                       1083489.3 lps  (10.0 s, 7 samples)

 

System Benchmarks Index Values              BASELINE      RESULT   INDEX

Dhrystone 2 using register variables        116700.0   7135020.6   611.4

Double-Precision Whetstone                      55.0      1387.4   252.2

Execl Throughput                                43.0      2658.2   618.2

File Copy 1024 bufsize 2000 maxblocks         3960.0    102436.7   258.7

File Copy 256 bufsize 500 maxblocks           1655.0     33850.5   204.5

File Copy 4096 bufsize 8000 maxblocks         5800.0    121124.7   208.8

Pipe Throughput                              12440.0    815411.4   655.5

Pipe-based Context Switching                  4000.0    255881.9   639.7

Process Creation                               126.0      5750.4   456.4

Shell Scripts (1 concurrent)                    42.4      1617.7   381.5

Shell Scripts (8 concurrent)                     6.0       159.2   265.3

System Call Overhead                         15000.0   1083489.3   722.3

                                                                  ========

System Benchmarks Index Score                                        395.7

 

4.结果分析

      从上面的测试结果可以看出,但个并行的任务的得分为171.3,4个并行任务的得分为395.7。对比测试时需要关注这个值。

六、           高级用法

      unixbench也是一种灵活的测试框架,测试人员可以根据需求添加测试项。




推荐本站淘宝优惠价购买喜欢的宝贝:

image.png

本文链接:https://www.hqyman.cn/post/11160.html 非本站原创文章欢迎转载,原创文章需保留本站地址!

分享到:
打赏





休息一下~~


« 上一篇 下一篇 »

发表评论:

◎欢迎参与讨论,请在这里发表您的看法、交流您的观点。

请先 登录 再评论,若不是会员请先 注册

您的IP地址是: