处理器性能评测报告

我们针对系统评测的不同角度会采用不同的评测程序。在目前已有的评测程序中,为下列评测目标找到某些合适的评测程序(列出即可)

测评目标测评程序测评目标测评程序
CPU整点性能SPECint2017CPU浮点性能SPECfp2017
计算机事务处理能力TPC-C嵌入式系统计算能力CoreMark
2D处理能力3DMark3D处理能力3DMark
并行计算性能LINPACK系统响应速度Loadrunner
编译优化能力SPECint2017操作系统性能Sysbench
多媒体处理能力PassMark SoundCheck™IO处理能力Iozone
浏览器性能wrk网络传输速率Netperf
Java运行环境性能Specjbb2005邮件服务性能servermark
文件服务器性能servermarkWeb服务器性能servermark
服务器功耗和性能SPECpower

阅读文献(Reinhold P.Weicker, An Overview of Common Benchmarks, IEEE Computer, December 1990.)并回答下面的问题

  1. 简述用于性能评测的MIPS指标之含义,以及它是如何被计算的

    • MIPS 有多种含义,使用时必须特别指出

      • Native MIPS, 即每秒执行的百万指令数。由于 RISC 指令系统的精简设计, RISC 在相同情况下是 CISCMIPS 的数倍。
      • Peak MIPS, 指处理器的MIPS峰值,在数值上等于处理器频率。
      • EDN MIPS, Dhrystone MIPS指运行特定程序得到的 Native MIPS 或者 VAX MIPS
      • VAX MIPS, 指定 VAX 11/7801MIPS,规定程序功能、语言、编译器,将待测机器的运行时间与 VAX 11/780 作比,比值就是待测机器的 MIPS
  2. 使用linux下的剖视工具(例如gprof)对dhrystonewhetstone进行剖视,参考论文Table 1形式给出数据,你的结果和该论文是否一致,为什么?

使用 -pg -static 进行编译.

procedure profile for dhrystone(cc_dry2)

从图片中可以看出,我的结果与论文基本相同.猜测原因是整型运算和库函数运行速度目前基本只与处理器频率相关,因此其占比没有很大变化.

procedure profile for whetstone3

从图片中可以看出, 我的结果与论文中不同, 我认为是计算机能力造成的.我的计算机和论文作者的计算机的浮点运算能力之间存在较大差异,导致了结果的不同.

  1. 文中讨论了处理器之外可能对性能造成影响的因素,请分别使用两种不同的语言(例如CJava)使用同一算法实现快速排序、矩阵乘法、求Ackermann函数,验证文中的观点。(请保留你的程序,我们在后面可能还会用到它)

    1. 我的程序采用了 CPython. 论文中认为除CPU外,编程语言、编译器、动态库、高速缓存以及内存大小都会影响到程序到性能

      • 编程语言: 在 CPython 两种语言中, 众所周知 C 语言一般比 Python 语言要快.
      • 编译器: 不同到编译选项也会直接影响程序性能,优化选项 -O0-O3 性能一般会增加, -register 选项会要求运算只使用寄存器, -ffast-math 会针对浮点运算进行某些优化.
      • 动态库: 通过剖视可以看出, 程序执行过程中总有一部分时间需要调用和执行库函数,不同库有不同实现方式,会对性能有所影响.
      • 高速缓冲: 一般缓存越大,性能越快. 这一点同样适用于内存.
      1. 快排中,对一个长度为$10^5$的整数数列进行快排,该数列事先生成,两种语言各循环100次, C语言耗时15s, Python耗时39s
      2. 矩阵乘法算法中,采用朴素的$O(n^3)$算法, 对两个 1000*1000 的整数矩阵进行矩阵乘法,各循环100次, C 耗时 54s, Python 不使用 numpy 库函数耗时103s, 利用 numpy 库函数仅耗时 13s
      3. Ackermann 函数, 对A(40,20)循环$10^6$次, C 耗时2s, python 耗时 1167s (一度以为卡死了...但做了下面的benchmark我又觉得这个时间还行).
      4. 从以上结果可以看出, C 语言基本上都比 Python 的要快上一个数量级.

性能评测 基于某个给定的计算机系统平台,使用dhrystone、whetstone、SPEC CPU2000开展评测、分析、研究并给出报告。

工作背景和评测目标

本次实验通过使用dhrystone,whetstone,SPEC CPU2000对个人PC进行测评,目标是掌握Benchmark的使用和测评估计办法。经过此次实验,我对Benchmark的应用有了较为深入的认识,并且认识到了Benchmark的局限性。

评测环境

项目详细指标和参数
处理器型号及相关参数Intel(R) Corm(TM) i7-8700K CPU @3.7GHz
6 x 32 KB 8-way set associative instruction caches
6 x 32 KB 8-way set associative data caches
6 x 256 KB 4-way set associative caches
12 MB 16-way set associative shared cache
内存1GB DDR4 3200MHz
外存64G SSD
操作系统及版本Linux Ubuntu-Linux 5.3.0-40-generic #32~18.04.1-Ubuntu SMP Mon Feb 3 14:05:59 UTC 2020 x86_64 x86_64 x86_64 GNU/Linux
编译器版本gcc version 7.4.0 (Ubuntu 7.4.0-1ubuntu1~18.04.1)

测评步骤及结果

1. 在linux下基于dhrystone-2.1所提供的Makefile编译dhrystone

使用 tar zxvf dhrystone-2.1.tar.gz 解压,然后进入解压目录下进行 make 命令, 报错.
根据报错提示,将 dhry_1.c 中的第47行的 times 函数声明修改为 extern clock_t times(struct tms *__buffer); 后, make 不再报错,生成四个新的可执行文件.

2. 分别采用$10^8$、$3\times10^8$、$5\times10^8$、$7\times10^8$、$9\times10^8$为输入次数,运行编译生成的两个程序,记录、处理相关数据并做出解释。

由于源程序输出精度限制,将 dhry_1.c 中第281行代码,修改为 printf ("%6.6f \n", Microseconds), 重新 make 后,用规定的循环轮数运行生成的四个可执行文件.

结果如下表

循环轮数类型gcc_dry2gcc_dry2regcc_dry2cc_dry2reg
$1\times10^8$ms/run0.04420.04380.03350.0345
$1\times10^8$Dhry/s2.26x$10^7$2.28x$10^7$2.99x$10^7$2.90x$10^7$
$3\times10^8$ms/run0.04360.04480.03400.0337
$3\times10^8$Dhry/s2.30x$10^7$2.23x$10^7$2.94x$10^7$2.97x$10^7$
$5\times10^8$ms/run0.04360.04430.03400.0340
$5\times10^8$Dhry/s2.29x$10^7$2.26x$10^7$2.94x$10^7$2.94x$10^7$
$7\times10^8$ms/run0.04360.04370.03410.0339
$7\times10^8$Dhry/s2.30x$10^7$2.29x$10^7$2.93x$10^7$2.95x$10^7$
$9\times10^8$ms/run0.04480.04380.03440.0337
$9\times10^8$Dhry/s2.23x$10^7$2.28x$10^7$2.90x$10^7$2.97x$10^7$

gcc编译的选项对于dhrystone的处理能力比cc稍差一些,对于不同的输入次数, dhrystone per second 会随着此数的增多而略微上升.

3. 对dhrystone代码做少量修改,使其运行结果不变但“性能”提升。

由于 dhry_2.cProc_8 函数中第101行存在一个只进行两次的循环,所以将其改写,去掉循环以减少循环的跳转预测错误引起的时间损失.
结果如下表
结果如下表

循环轮数类型gcc_dry2_oldgcc_dry2_newcc_dry2_oldcc_dry2_new
$1\times10^8$ms/run0.04420.04250.03350.0336
$1\times10^8$Dhry/s2.26x$10^7$2.35x$10^7$2.99x$10^7$2.97x$10^7$
$3\times10^8$ms/run0.04360.04280.03400.0341
$3\times10^8$Dhry/s2.30x$10^7$2.34x$10^7$2.94x$10^7$2.93x$10^7$
$5\times10^8$ms/run0.04360.04360.03400.0341
$5\times10^8$Dhry/s2.29x$10^7$2.30x$10^7$2.94x$10^7$2.94x$10^7$
$7\times10^8$ms/run0.04360.04230.03410.0344
$7\times10^8$Dhry/s2.30x$10^7$2.36x$10^7$2.93x$10^7$2.91x$10^7$
$9\times10^8$ms/run0.04480.04230.03440.0342
$9\times10^8$Dhry/s2.23x$10^7$2.36x$10^7$2.90x$10^7$2.92x$10^7$

从上表中可以看出,针对 Proc_8 循环语句的修改, 小幅度的提升了性能,降低了 CPU Time, 该修改是有效的.

4. 采用dhrystone进行评测有哪些可改进的地方?对其做出修改、评测和说明。

使用编译优化选项对跳转指令进行预测优化, 分别修改 Makefile 中的 GCCOPTIM 选项为 -O0,-O1,-O2,-O3 使用 gcc_dry2 进行实验,结果如下表.

循环轮数类型-O0-O1-O2-O3
$1\times10^8$ms/run0.08680.04350.03520.0348
$1\times10^8$Dhry/s1.15x$10^7$2.30x$10^7$2.84x$10^7$2.87x$10^7$
$3\times10^8$ms/run0.08690.04270.03490.0351
$3\times10^8$Dhry/s1.15x$10^7$2.34x$10^7$2.87x$10^7$2.85x$10^7$
$5\times10^8$ms/run0.08720.04710.03510.0346
$5\times10^8$Dhry/s1.15x$10^7$2.12x$10^7$2.85x$10^7$2.89x$10^7$
$7\times10^8$ms/run0.08740.04230.03520.0341
$7\times10^8$Dhry/s1.14x$10^7$2.37x$10^7$2.84x$10^7$2.94x$10^7$
$9\times10^8$ms/run0.08680.04340.03540.0343
$9\times10^8$Dhry/s1.15x$10^7$2.30x$10^7$2.82x$10^7$2.92x$10^7$

从表中可以看出 -O1 相对于 -O0 来说, CPU Time 有明显的下降, 并且 -O1,-O2,-O3 来说没有较大提升.

5. 在linux下使用编译器分别采用-O0-O2-O3选项对whetstone程序进行编译并执行,记录评测结果。

使用命令 gcc whetstone.c -lm -o whetstone -O0 依次生成 -O0, -O2, -O3 程序, 得到whet0, whet1, whet2 三个可执行文件(数字编号对应优化).

6. 分别采用$10^6$、$10^7$、$10^8$、$10^9$为输入次数,运行编译生成的可执行程序,记录、处理相关数据并做出解释。

得到实验结果如下表

循环轮数类型whet0whet2whet3
$1\times10^6$Dura311615
$1\times10^6$MIPS3225.86250.06666.7
$1\times10^7$Dura299112104
$1\times10^7$MIPS3344.58928.69615.4
$1\times10^8$Dura301110971007
$1\times10^8$MIPS3321.29115.89930.5
$1\times10^9$Dura28825105699763
$1\times10^9$MIPS3469.29461.610242.8

根据上表,发现循环轮数足够大时, whetstone 检测趋于稳定. 此外 -O0-O3 各个优化选项带来的影响基本符合预期, -O2-O3 性能提升很小,因为 -O3-O2 最主要差别为对于 inline 的优化,而 whetstone 中没有这一类函数.

7. 进一步改进whetstone程序性能(例如新的编译选项),用实验结果回答。

由于 whetstone 是针对浮点运算的测评,利用 -ffast-math 选项对 whetstone 性能进行优化, -ffast-mathmath 库进行优化,大幅提高运行速度,但也可能造成一些运算错误.

实验结果如下表

循环轮数类型whet0whet2whet3
$1\times10^6$Dura1411
$1\times10^6$MIPS7142.9100000.0100000.0
$1\times10^7$Dura1391213
$1\times10^7$MIPS7194.283333.376923.1
$1\times10^8$Dura1377129138
$1\times10^8$MIPS7262.277519.472463.8
$1\times10^9$Dura1352212171312
$1\times10^9$MIPS7395.482196.376219.5

从上表中, 可以发现增加 -ffast-math 后运行速度有了极大的提升.

8. 完成SPEC CPU2000的安装。

9. 修改自己的config文件,分别用低强度优化(例如O2)和高强度优化(例如O3)完成完整的SPEC CPU2000的评测,提交评测报告文件。

这里给出我改的cfg文件,并且将 gccg++gfortran 都降至4.8版本,分别用低优化和高优化进行bench

linux-amd64-gcc4.cfg

###############################################################################
# This is a sample config file for AMD64 linux using GCC 4.0 compiler.
# This does not provide any peak level tuning, but can be used as a 
# starter config file.
###############################################################################
# To run: runspec -c linux-amd64-gcc4.cfg -T all -n 3 int fp
# The extension 'gcc4-high-opt' uses high optimization and is enabled by
# default.  If you want lower levels of optimization, use the extension 
# 'gcc4-low-opt', e.g.
# runspec -c linux-amd64-gcc4.cfg -T all -n 3 -e gcc4-low-opt int fp
###############################################################################

# These variables below are examples and should be modified according to the
# system under test.
company_name    = XYZ Inc.
hw_model    = Opteron (TM) 150 
hw_cpu        = AMD Opteron (TM) 150
hw_cpu_mhz    = 2400
hw_disk        = IDE, WD2000
hw_fpu        = Integrated
hw_memory    = 2 x 512 PC3200 DDR SDRAM CL2.0 (Corsair CMX512R-3200C2)
hw_avail    = May-2003
test_date    = 
sw_file        = Linux/ext3
sw_os        = SuSE Linux 9.2 for AMD64
hw_vendor    = 
tester_name    = XYZ Inc.
license_num    = 0000

hw_ncpu        = 1
hw_ncpuorder    = 1
hw_ocache    = N/A
hw_other    = None
hw_parallel    = No
hw_pcache    = 64KBI + 64KBD on chip
hw_scache    = 1024KB(I+D) on chip
hw_tcache    = N/A
sw_state    = Multi-user SuSE Run level 3

VENDOR        =
action        = validate
tune        = base
output_format    = asc,html,config
ext        = gcc4-high-opt

check_md5    = 1
reportable    = 1

teeout=yes
teerunout=yes

#
# These are listed as benchmark-tuning-extension-machine
#
default=default=default=default:
CC    = gcc -fpermissive -DHAS_ERRLIST -DSPEC_STDCPP
CXX    = g++ -fpermissive -DHAS_ERRLIST -DSPEC_STDCPP
FC    = gfortran
F77    = gfortran


################################################################
# Portability Flags
################################################################

255.vortex=default=default=default:
notes0045=   255.vortex=  CPORTABILITY=-DSPEC_CPU2000_LP64
CPORTABILITY    = -DSPEC_CPU2000_LP64

186.crafty=default=default=default:
notes0050=   186.crafty:  CPORTABILITY=-DLINUX_i386
CPORTABILITY    = -DLINUX_i386

252.eon=default=default=default:
notes0051=   252.eon:     CXXPORTABILITY=-DHAS_ERRLIST -DSPEC_CPU2000_LP64 
CXXPORTABILITY    = -DHAS_ERRLIST -DSPEC_CPU2000_LP64

253.perlbmk=default=default=default: 
notes0052=   253.perlbmk: CPORTABILITY=-DSPEC_CPU2000_LINUX_I386 -DSPEC_CPU2000_NEED_BOOL
notes0053=                             -DSPEC_CPU2000_LP64
CPORTABILITY    = -DSPEC_CPU2000_NEED_BOOL -DSPEC_CPU2000_LINUX_I386 -DSPEC_CPU2000_LP64
 
254.gap=default=default=default:
notes0055=   254.gap:     CPORTABILITY=-DSYS_IS_USG -DSYS_HAS_IOCTL_PROTO -DSYS_HAS_TIME_PROTO
notes0056=                             -DSYS_HAS_CALLOC_PROTO -DSYS_HAS_MALLOC_PROTO
notes0057=                             -DSPEC_CPU2000_LP64
CPORTABILITY    = -DSYS_HAS_MALLOC_PROTO -DSYS_HAS_CALLOC_PROTO -DSYS_IS_USG -DSYS_HAS_IOCTL_PROTO -DSYS_HAS_TIME_PROTO -DSPEC_CPU2000_LP64

178.galgel=default=default=default:
notes0050=    178.galgel:     -ffixed-form
FPORTABILITY= -ffixed-form


################################################################
# Baseline Tuning Flags
################################################################

#
# int2000
# Base tuning default optimization
#
int=base=gcc4-low-opt=default:
notes0080=  Baseline C:   gcc -O2
COPTIMIZE    = -O2
feedback=0

252.eon=base=gcc4-low-opt=default:
notes0085=  Baseline C++: g++ -O2
CXXOPTIMIZE    = -O2

# High Optimization:
int=base=gcc4-high-opt=default:
notes0080=  Baseline C:   gcc -O3 
notes0085=  Baseline C++: g++ -O3
COPTIMIZE    = -O3
CXXOPTIMIZE    = -O3


#
# fp2000
# Base tuning default optimization
#
#    Fortran benchmarks
#
fp=base=gcc4-low-opt=default:
notes0085=  Baseline: Fortran gfortran -O2
notes0080=  Baseline: C       gcc -O2
FOPTIMIZE    = -O2
F77OPTIMIZE    = -O2
COPTIMIZE    = -O2

#High Optimization:

fp=base=gcc4-high-opt=default:
notes0080=  Baseline: C       gcc      -O3 
notes0085=  Baseline: Fortran gfortran -O3
FOPTIMIZE    = -O3 
F77OPTIMIZE    = -O3
COPTIMIZE    = -O3


################################################################
# Peak Tuning Flags
################################################################

#
# int2000
# Peak tuning
#

int=peak=default=default:
notes0087=   All peak: basepeak=yes
basepeak = yes


int=default=default=default:
notes0030=  Portability:
sw_avail= Dec-2003
sw_compiler0000= GNU gcc 4.0 compiler for AMD64

fp=default=default=default:
sw_avail= Jan-2004
sw_compiler0000= GNU gcc 4.0 and gfortran 4.0 compiler for AMD64

__MD5__
164.gzip=base=gcc4-high-opt=default:
# Last updated Sat Feb 29 20:01:55 2020
optmd5=96f7305b2efe7cd6a086a3500462c7d6
exemd5=2550749a90a2523204948317004f6c16

175.vpr=base=gcc4-high-opt=default:
# Last updated Sat Feb 29 20:01:55 2020
optmd5=3ba2478ba3445aab700488b14bdfc04a
exemd5=9c74f8c1842951f19b3c1b2e48ab50ba

176.gcc=base=gcc4-high-opt=default:
# Last updated Sat Feb 29 20:01:55 2020
optmd5=3869c2abdc72c6dede932f899746bab1
exemd5=45e9075f2fc7720fb90430798cc507d7

181.mcf=base=gcc4-high-opt=default:
# Last updated Sat Feb 29 20:01:55 2020
optmd5=bcd9714775fb8b9dfc766973339bc9fa
exemd5=20b03564736d487e384a1c3f90c577af

186.crafty=base=gcc4-high-opt=default:
# Last updated Sat Feb 29 20:01:55 2020
optmd5=e7e175d76af05a01280aa10308a66f1f
exemd5=12472c760f2654d2d264b0ef18a70726

197.parser=base=gcc4-high-opt=default:
# Last updated Sat Feb 29 20:01:55 2020
optmd5=da82519e9fe7cae0c83c252057a1739e
exemd5=60c73c9944ccc16786796729a4ec55ec

252.eon=base=gcc4-high-opt=default:
# Last updated Sat Feb 29 20:01:55 2020
optmd5=a57f4fee33aed5ba3911b4de570ba28e
exemd5=8f010418230469acb2569d0ea23f6470

253.perlbmk=base=gcc4-high-opt=default:
# Last updated Sat Feb 29 20:01:55 2020
optmd5=fab2769aca012ebde3a21dc21ed71dde
exemd5=5c9a8f3511b1b712bac4ecc60e21d33d

254.gap=base=gcc4-high-opt=default:
# Last updated Sat Feb 29 20:01:55 2020
optmd5=3bc92b9947eb96341ef63508a8e70c28
exemd5=6ef05078ba00eac8f50b1af8fd6c8976

255.vortex=base=gcc4-high-opt=default:
# Last updated Sat Feb 29 20:01:55 2020
optmd5=da9d19bbb8d7ec39dd0848959aa370f7
exemd5=ca63ea932c2d5cae4c80e9573265d21c

256.bzip2=base=gcc4-high-opt=default:
# Last updated Sat Feb 29 20:01:55 2020
optmd5=14db51db373187a0f6736a9d125ec07d
exemd5=862fc9b5a106bc3e4f12a56905d9e1cc

300.twolf=base=gcc4-high-opt=default:
# Last updated Sat Feb 29 20:01:55 2020
optmd5=1ce94c85c8e2d9217efdf6895de6020b
exemd5=535f64680cf4fdd1c9cad761459bd7f5

168.wupwise=base=gcc4-high-opt=default:
# Last updated Sat Feb 29 20:05:07 2020
optmd5=5c25d15abd005cc698c92a664df5184a
exemd5=3745063442df6a46ac6eb9db5097dbbd

168.wupwise=peak=gcc4-high-opt=default:
# Last updated Sat Feb 29 20:05:07 2020
optmd5=063d1ea855415d7646c130c6e0416b61
exemd5=ea62c4b615c8a01bed3eae917a40504d

171.swim=base=gcc4-high-opt=default:
# Last updated Sat Feb 29 20:05:07 2020
optmd5=a4404c279ad9ef66119bca1a18cadddb
exemd5=a65db57f397667eff6f15d83f5cf9dae

171.swim=peak=gcc4-high-opt=default:
# Last updated Sat Feb 29 20:05:07 2020
optmd5=2bd88232e6a81d10148efcdea591c5a2
exemd5=b06ecafe63b0ee604f5d9fc149e94fcb

172.mgrid=base=gcc4-high-opt=default:
# Last updated Sat Feb 29 20:05:07 2020
optmd5=46bf28f6e0ccfcab7ae615e72e4697ab
exemd5=380bbb12ae3afc40c770949c26ded4f1

172.mgrid=peak=gcc4-high-opt=default:
# Last updated Sat Feb 29 20:05:07 2020
optmd5=e912cad32f119f12a0d567dcd80bed1b
exemd5=f549a0c3c8951f8dbfa4b394fe5b37f0

173.applu=base=gcc4-high-opt=default:
# Last updated Sat Feb 29 20:05:07 2020
optmd5=724abee4112d4b1443c1e850d6db1105
exemd5=a5cef86974cf57b8103341aed9f94509

173.applu=peak=gcc4-high-opt=default:
# Last updated Sat Feb 29 20:05:07 2020
optmd5=69883d821c6385916c594d49d35df5d1
exemd5=277f3a4e9426f11acaf14d2be943da96

177.mesa=base=gcc4-high-opt=default:
# Last updated Sat Feb 29 20:05:07 2020
optmd5=d51cec6ce8fcbcd52ea64dbf63575ceb
exemd5=766dcf0c31fcbcffe6a851025f0a7e33

177.mesa=peak=gcc4-high-opt=default:
# Last updated Sat Feb 29 20:05:07 2020
optmd5=d887ed3e6e0832b9abbb33d87c8b2598
exemd5=56f986a54dce751a9c128b8a9703022a

178.galgel=base=gcc4-high-opt=default:
# Last updated Sat Feb 29 20:05:07 2020
optmd5=216232f1c342b8cf033a8ed4afa7900d
exemd5=e3cb0b15ff7796aba5fdfb91472a6c1c

178.galgel=peak=gcc4-high-opt=default:
# Last updated Sat Feb 29 20:05:07 2020
optmd5=fa2ae5d1563af1dd5b2a3b7e779accf7
exemd5=ed6ed19866e9ed9f75aa8d4226577a47

179.art=base=gcc4-high-opt=default:
# Last updated Sat Feb 29 20:05:07 2020
optmd5=6babcc2f0b62506715f2d9a29e7e25a9
exemd5=320425d183802eca2b9269d66b9054f8

179.art=peak=gcc4-high-opt=default:
# Last updated Sat Feb 29 20:05:07 2020
optmd5=c9f4bfad75e783afd051a0ecdf1aa3e3
exemd5=8178f20f56c18f57390e7240decd6ca0

183.equake=base=gcc4-high-opt=default:
# Last updated Sat Feb 29 20:05:07 2020
optmd5=30dc22ffdf76386d218fbc142b265cfc
exemd5=90860a44c83a9e23693b32633c67924f

183.equake=peak=gcc4-high-opt=default:
# Last updated Sat Feb 29 20:05:07 2020
optmd5=b04acdba2329a0dc64b97e9285596632
exemd5=de870126e6dae03de05d1aee4d28508d

187.facerec=base=gcc4-high-opt=default:
# Last updated Sat Feb 29 20:05:07 2020
optmd5=35a9d9f189dd3bcf03df8b1ec7450e67
exemd5=eaf558c54678420de61fd9bc02e191bb

187.facerec=peak=gcc4-high-opt=default:
# Last updated Sat Feb 29 20:05:07 2020
optmd5=0db47db4c3ba62397682ba6a72ec9b97
exemd5=eec508203b69f4c716fd7b921105e1d2

188.ammp=base=gcc4-high-opt=default:
# Last updated Sat Feb 29 20:05:07 2020
optmd5=13119a6ce16610d8c9cb57f7668eef2a
exemd5=93819b1283f1c26a8963c9bf7c1d5c34

188.ammp=peak=gcc4-high-opt=default:
# Last updated Sat Feb 29 20:05:07 2020
optmd5=28cd70cc627ff27c732fedc254b6e248
exemd5=e88fdc1bddcd3cb6073281f7e32da011

189.lucas=base=gcc4-high-opt=default:
# Last updated Sat Feb 29 20:05:07 2020
optmd5=3c46180d4d2aeb9a64902de7f83e4d31
exemd5=dc4486c74a1d64531c2c4b7051701f87

189.lucas=peak=gcc4-high-opt=default:
# Last updated Sat Feb 29 20:05:07 2020
optmd5=57818530c2124120646da96a1d67f422
exemd5=49aef15fd0edebb99dafd23cdae84db4

191.fma3d=base=gcc4-high-opt=default:
# Last updated Sat Feb 29 20:05:07 2020
optmd5=b40236010d15cb4a8ddb460ba65f1803
exemd5=ee5f636eb7b3a999bb188fbf9264d9b1

191.fma3d=peak=gcc4-high-opt=default:
# Last updated Sat Feb 29 20:05:07 2020
optmd5=48d7f92813543e40df7af626cecb1928
exemd5=d95efad10ac8cab0f3e3d30c96f21818

200.sixtrack=base=gcc4-high-opt=default:
# Last updated Sat Feb 29 20:05:07 2020
optmd5=6f4ebf9dcbd9d1c3bcfab2c31dac645c
exemd5=18c3ad3aa5ebe88fedc504502ca692f4

200.sixtrack=peak=gcc4-high-opt=default:
# Last updated Sat Feb 29 20:05:07 2020
optmd5=e55964a2beba18f56ab8fc598b25da7d
exemd5=69d37b6e6dd53389990b7f1f51031e35

301.apsi=base=gcc4-high-opt=default:
# Last updated Sat Feb 29 20:05:07 2020
optmd5=1f2a54d2d4e1fcbeffab87b429217105
exemd5=04bbe69751f307d7a8390a70ddf844b8

301.apsi=peak=gcc4-high-opt=default:
# Last updated Sat Feb 29 20:05:07 2020
optmd5=9c94dddc242d2e02e001a3d6830a9745
exemd5=bc18b5c0a052868587af0bd4bf092092

164.gzip=base=gcc4-low-opt=default:
# Last updated Sat Feb 29 22:28:54 2020
optmd5=b410bae0d5b4cb300ecab8c9089b51cb
exemd5=9e034310b3dadebcc84d6bab8f252eb3

168.wupwise=base=gcc4-low-opt=default:
# Last updated Sat Feb 29 22:28:54 2020
optmd5=6d716a34f257dfaabcb9e1e6e9a8bbfb
exemd5=30b6f27c6e17ccd7f31b8f71b205aea1

168.wupwise=peak=gcc4-low-opt=default:
# Last updated Sat Feb 29 22:28:54 2020
optmd5=629ebaf8eb12ee0fbe39291808ed54df
exemd5=ea62c4b615c8a01bed3eae917a40504d

171.swim=base=gcc4-low-opt=default:
# Last updated Sat Feb 29 22:28:54 2020
optmd5=b8e3544181a92b94c8e8207e8bdd3593
exemd5=93e9a76c926e772821691ecc31fbff5a

171.swim=peak=gcc4-low-opt=default:
# Last updated Sat Feb 29 22:28:54 2020
optmd5=18320100daaa43ffd8e02f2283a18300
exemd5=b06ecafe63b0ee604f5d9fc149e94fcb

172.mgrid=base=gcc4-low-opt=default:
# Last updated Sat Feb 29 22:28:54 2020
optmd5=d5337d68c449d31a4407315f475a74cd
exemd5=821cf74de649e049562ab8aaf4b07ebc

172.mgrid=peak=gcc4-low-opt=default:
# Last updated Sat Feb 29 22:28:54 2020
optmd5=a18aa5298f6e5434b30009db996491e7
exemd5=f549a0c3c8951f8dbfa4b394fe5b37f0

173.applu=base=gcc4-low-opt=default:
# Last updated Sat Feb 29 22:28:54 2020
optmd5=499cb2affff242cbb155985e732618e6
exemd5=232b21a6dd1ee109f43f0745c53bb95e

173.applu=peak=gcc4-low-opt=default:
# Last updated Sat Feb 29 22:28:54 2020
optmd5=4a70e336ebe8f450fd1db9cd69b97d70
exemd5=277f3a4e9426f11acaf14d2be943da96

175.vpr=base=gcc4-low-opt=default:
# Last updated Sat Feb 29 22:28:54 2020
optmd5=bc307035c095a2f898c9db4f4c09340d
exemd5=1fbe49c1ef5391b9ebbf9c299b3752ba

176.gcc=base=gcc4-low-opt=default:
# Last updated Sat Feb 29 22:28:54 2020
optmd5=a16a1ba36d4b4db525b519044341bd72
exemd5=ce32b4b4c6b3b563295ea02958dab2c9

177.mesa=base=gcc4-low-opt=default:
# Last updated Sat Feb 29 22:28:54 2020
optmd5=6a7ea0fb4ecbbe39779eb09a6481af7f
exemd5=44df1961dccdc5925244a613813174e9

177.mesa=peak=gcc4-low-opt=default:
# Last updated Sat Feb 29 22:28:54 2020
optmd5=150fd0276a454cdadb2bd7dc07d629ed
exemd5=56f986a54dce751a9c128b8a9703022a

178.galgel=base=gcc4-low-opt=default:
# Last updated Sat Feb 29 22:28:54 2020
optmd5=3804f00fd90eea58fc137c8549ac97ee
exemd5=a2615e7580e91e9b656adf9c9f4ed0b7

178.galgel=peak=gcc4-low-opt=default:
# Last updated Sat Feb 29 22:28:54 2020
optmd5=29a25880c110c3c80839cca4c309f5cb
exemd5=ed6ed19866e9ed9f75aa8d4226577a47

179.art=base=gcc4-low-opt=default:
# Last updated Sat Feb 29 22:28:54 2020
optmd5=5900109d64ea34afa92f7abda2b86810
exemd5=5cf10a4be0a02efaf33dbe6a11eab29d

179.art=peak=gcc4-low-opt=default:
# Last updated Sat Feb 29 22:28:54 2020
optmd5=5d4fba8242edf0c74ae9b2725eb6d824
exemd5=8178f20f56c18f57390e7240decd6ca0

181.mcf=base=gcc4-low-opt=default:
# Last updated Sat Feb 29 22:28:54 2020
optmd5=740d53f67b6ea1df2fdc02e4787f3960
exemd5=edb424df15ad0ecf57b0469703bdfa8d

183.equake=base=gcc4-low-opt=default:
# Last updated Sat Feb 29 22:28:54 2020
optmd5=165ff9f467b01c870ec56aa09bf89317
exemd5=47ad172d28e117b7d98ad8c10d64ebfa

183.equake=peak=gcc4-low-opt=default:
# Last updated Sat Feb 29 22:28:54 2020
optmd5=c766db18ad6acce60fbce58d0c161a04
exemd5=de870126e6dae03de05d1aee4d28508d

186.crafty=base=gcc4-low-opt=default:
# Last updated Sat Feb 29 22:28:54 2020
optmd5=bd8470e2f0717b6c1116fab19f256ff1
exemd5=08876c4d3bec36c9e28469358beeb4ec

187.facerec=base=gcc4-low-opt=default:
# Last updated Sat Feb 29 22:28:54 2020
optmd5=3cfd6b2aeec4f40876ee8eafb4d53f1d
exemd5=d954b378b8d66d679b469ca0be6ac9df

187.facerec=peak=gcc4-low-opt=default:
# Last updated Sat Feb 29 22:28:54 2020
optmd5=8ff5a14b354841b6233e3129daa59a61
exemd5=eec508203b69f4c716fd7b921105e1d2

188.ammp=base=gcc4-low-opt=default:
# Last updated Sat Feb 29 22:28:54 2020
optmd5=46c649dbd651d28663373649824657a4
exemd5=79169fc71da478a31befd218ef0da399

188.ammp=peak=gcc4-low-opt=default:
# Last updated Sat Feb 29 22:28:54 2020
optmd5=27ed62374413191cf2edf09d4920cd46
exemd5=e88fdc1bddcd3cb6073281f7e32da011

189.lucas=base=gcc4-low-opt=default:
# Last updated Sat Feb 29 22:28:54 2020
optmd5=605977aa3e0c622f71d30f3275329331
exemd5=0d6ac97fc9125927dd4f3fdff1b13aff

189.lucas=peak=gcc4-low-opt=default:
# Last updated Sat Feb 29 22:28:54 2020
optmd5=65d55b5ccdece4438a4efd8d22f34868
exemd5=49aef15fd0edebb99dafd23cdae84db4

191.fma3d=base=gcc4-low-opt=default:
# Last updated Sat Feb 29 22:28:54 2020
optmd5=0b6b65e98d64d39354ca483c9e06b1c1
exemd5=a821be3d371fd1fb135690796ee529da

191.fma3d=peak=gcc4-low-opt=default:
# Last updated Sat Feb 29 22:28:54 2020
optmd5=db0bdf2723c8d203e34aa4a558c05c00
exemd5=d95efad10ac8cab0f3e3d30c96f21818

197.parser=base=gcc4-low-opt=default:
# Last updated Sat Feb 29 22:28:54 2020
optmd5=2208e78ed24fd1cbf52a9f944b247607
exemd5=c0c7e8bcb8dc6800657c13c74e6b56e6

200.sixtrack=base=gcc4-low-opt=default:
# Last updated Sat Feb 29 22:28:54 2020
optmd5=d91514cef4ea49a8c4586cc12aa95e9b
exemd5=a04d2f8d9198f3152aad6439d4193305

200.sixtrack=peak=gcc4-low-opt=default:
# Last updated Sat Feb 29 22:28:54 2020
optmd5=84cf5cea07e63d844f9e4af9d72afd4c
exemd5=69d37b6e6dd53389990b7f1f51031e35

252.eon=base=gcc4-low-opt=default:
# Last updated Sat Feb 29 22:28:54 2020
optmd5=3ddc9242465c9c3081f1a1408f9df44e
exemd5=dd0fde9ef90b079b09de4bf846bbda2c

253.perlbmk=base=gcc4-low-opt=default:
# Last updated Sat Feb 29 22:28:54 2020
optmd5=b85ac7c48652bd60c0dd886c665311f2
exemd5=0acfce1e19a854d2f0184dc88bc345e1

254.gap=base=gcc4-low-opt=default:
# Last updated Sat Feb 29 22:28:54 2020
optmd5=a186d23f7ab59c8af67f85311c570afe
exemd5=740a60c7c783d4a9660d935a44c934f3

255.vortex=base=gcc4-low-opt=default:
# Last updated Sat Feb 29 22:28:54 2020
optmd5=600d5c4aa160911b1ff36551598fa4f1
exemd5=15256871b0669c91fa9aeb134779e39a

256.bzip2=base=gcc4-low-opt=default:
# Last updated Sat Feb 29 22:28:54 2020
optmd5=950bff2e3d1f8f02f0aa33b40594ad1f
exemd5=e34a5f93462181bf6f051a1650db1e24

300.twolf=base=gcc4-low-opt=default:
# Last updated Sat Feb 29 22:28:54 2020
optmd5=bc30ac1b85dd975e82e7a87cd2c0101d
exemd5=96cc977ae9940a02d5d6c7021f4843ce

301.apsi=base=gcc4-low-opt=default:
# Last updated Sat Feb 29 22:28:54 2020
optmd5=19c21bf245ce3b60c6190f3e0d064669
exemd5=cc1ab14d0493adae7f860f22f314687a

301.apsi=peak=gcc4-low-opt=default:
# Last updated Sat Feb 29 22:28:54 2020
optmd5=93ac2de4f8e6bab66d3bc06dbacf676f
exemd5=bc18b5c0a052868587af0bd4bf092092

CFP2000.016.html CINT2000.016.html为高强度优化的测评报告文件.
CFP2000.017.html CINT2000.017.html为低强度优化的测评报告文件.

小结

本次实验,利用 Whetstone Benchmark 和 Dhrystone Benchmark 以及 SPEC CPU2000, 对测评电脑进行了多方面的测试. 从中可以看出,不同的测评程序对于同一电脑的结果均有所差别, 这说明 Benchmark 具有一定的局限性, 较容易被针对优化.

Last modification:May 9th, 2020 at 12:51 am
如果觉得我的文章对你有用,请随意赞赏