Linux下Cell 3.2G vs. PPC G5 1.6G对比测试出炉

Prescott · 发表于 2006-11-23 12:06

http://www.geekpatrol.ca/2006/11/playstation-3-performance/

Cell的这个PPE还真是强大阿，居然绝大部分测试性能能输给1.6GHz的PowerPC G5。 :wacko:

这个geekbench有Windows版本阿，大家可以测试一下，看看Cell vs Conroe是什么水准 :D

下载地址：
http://www.geekpatrol.ca/geekbench/#download

PlayStation 3 Performance

On Sunday I saw a clip of Fedora Core 5 for PPC running on the PlayStation 3 over at Kotaku; I’d completely forgotten that Sony was going to make it easy to boot other operating systems on the PlayStation 3!

On Monday I started receiving requests for Geekbench for Linux PPC so people could run it on the PlayStation 3. I managed to get a beta version out last night and while it’s not quite ready for public release yet, one beta tester sent in the results for his PlayStation 3 which I thought I’d share. To give the results some context, I’m going to compare the PlayStation 3 results against one of the first Power Mac G5s running at 1.6GHz.

Setup
   Playstation 3
      o Cell Broadband Engine @ 3.2GHz
      o 256 MB RAM
      o Fedora Core 5
      o Geekbench 2006 (Build 243)

   Power Mac G5
      o PowerPC G5 @ 1.6GHz
      o 1280 MB RAM
      o Fedora Core 4
      o Geekbench 2006 (Build 243)

I’m reporting the baseline score, rather than the raw score, for each test (where 100 is the score a PowerMac G5 1.6GHz running Mac OS X would receive on the same test). As always, higher scores are better.

Results
Overall Score
PlayStation 3 105.2
Power Mac G5 106.9

Integer Performance

Emulate 6502 (single-threaded scalar)
PlayStation 3 42.1
Power Mac G5 73.9

Emulate 6502 (multi-threaded scalar)
PlayStation 3 57.3
Power Mac G5 73.8

Blowfish (single-threaded scalar)
PlayStation 3 118.7
Power Mac G5 107.0

Blowfish (multi-threaded scalar)
PlayStation 3 165.6
Power Mac G5  107.0

bzip2 Compress (single-threaded scalar)
PlayStation 3 89.8
Power Mac G5 162.8

bzip2 Compress (multi-threaded scalar)
PlayStation 3 124.1
Power Mac G5 168.4

bzip2 Decompress (single-threaded scalar)
PlayStation 3 76.6
Power Mac G5 129.9

bzip2 Decompress (multi-threaded scalar)
PlayStation 3 99.5
Power Mac G5 133.1

Floating Point Performance

Mandelbrot (single-threaded scalar)
PlayStation 3 49.0
Power Mac G5 100.0

Mandelbrot (multi-threaded scalar)
PlayStation 3 72.1
Power Mac G5 100.0

Dot Product (single-threaded scalar)
PlayStation 3 120.0
Power Mac G5 100.8

Dot Product (multi-threaded scalar)
PlayStation 3 119.3
Power Mac G5 100.3

JPEG Compress (single-threaded scalar)
PlayStation 3 70.7
Power Mac G5 103.0

JPEG Compress (multi-threaded scalar)
PlayStation 3 94.8
Power Mac G5 103.0

JPEG Decompress (single-threaded scalar)
PlayStation 3 61.6
Power Mac G5 119.0

JPEG Decompress (multi-threaded scalar)
PlayStation 3 72.9
Power Mac G5 119.2

Memory Performance

Read Sequential (single-threaded scalar)
PlayStation 3 51.9
Power Mac G5 116.7

Read Sequential (multi-threaded scalar)
PlayStation 3 56.9
Power Mac G5 116.0

Write Sequential (single-threaded scalar)
PlayStation 3 194.6
Power Mac G5 104.7

Write Sequential (multi-threaded scalar)
PlayStation 3 191.4
Power Mac G5 112.7

Stdlib Allocate (single-threaded scalar)
PlayStation 3 43.4
Power Mac G5  56.4

Stdlib Allocate (multi-threaded scalar)
PlayStation 3 51.2
Power Mac G5 55.6

Stdlib Write (single-threaded scalar)
PlayStation 3 331.5
Power Mac G5  92.7

Stdlib Write (multi-threaded scalar)
PlayStation 3 365.9
Power Mac G5 94.7

Stdlib Copy (single-threaded scalar)
PlayStation 3 64.5
Power Mac G5 63.5

Stdlib Copy (multi-threaded scalar)
PlayStation 3 102.1
Power Mac G5 72.7

Stream Performance

Stream Copy (single-threaded scalar)
PlayStation 3 89.7
Power Mac G5 114.1

Stream Copy (multi-threaded scalar)
PlayStation 3 109.9
Power Mac G5 111.8

Stream Scale (single-threaded scalar)
PlayStation 3 69.2
Power Mac G5 118.3

Stream Scale (multi-threaded scalar)
PlayStation 3 101.4
Power Mac G5 120.1

Stream Add (single-threaded scalar)
PlayStation 3 62.6
Power Mac G5 123.0

Stream Add (multi-threaded scalar)
PlayStation 3 93.2
Power Mac G5 118.0

Stream Triad (single-threaded scalar)
PlayStation 3 62.7
Power Mac G5 122.8

Stream Triad (multi-threaded scalar)
PlayStation 3 102.2
Power Mac G5 118.6

Conclusion

There was a comment on Slashdot last year that made the following assertion about the Cell processor:

The problem is that though the main CPU is PowerPC-based like current Apple chips, it is stripped down, and the Altivec support will be much lower than in current G5s. Unoptomized, Apple code would run like a G4 on this hardware.

Turns out the comment was right; Cell processor performance is comparable to low-end PowerPC G5 performance (which in turn is comparable to high-end PowerPC G4 performance). I can’t comment on Altivec performance, unfortunately, since Geekbench for Linux PPC doesn’t measure Altivec performance yet.

Geekbench also isn’t able to exploit the eight vector processors on the Cell processor. Any program designed and optimized for the Cell processor should be a lot faster than one designed for a generic processor (like, say, Geekbench). So while the Geekbench results might seem disappointing, keep in mind that Geekbench can’t exercise the PlayStation 3 to its full potential.

[ 本帖最后由 Prescott 于 2006-11-23 12:21 编辑 ]

只看该作者 · 发表于 2006-11-23 12:15

提示: 作者被禁止或删除内容自动屏蔽

武内空 · 发表于 2006-11-23 12:19

Playstation 3 游戏机？
Playstation 3
Cell Broadband Engine @ 3.2GHz
256 MB RAM
Fedora Core 5
Geekbench 2006 (Build 243)

Power Mac G5 苹果电脑？
Power Mac G5
PowerPC G5 @ 1.6GHz
1280 MB RAM
Fedora Core 4
Geekbench 2006 (Build 243)

Prescott · 发表于 2006-11-23 12:26

随便测测3.6G的P4什么水准。

Geekbench 2006 (build 238).  Email geekbench@geekpatrol.ca with fee

System Information
  Geekbench Version:       Geekbench 2006 (build 238)
  Geekbench Platform:       Windows x86 (32-bit)
  Geekbench Compiler:       Visual C++ 2005
  OS:                      Microsoft Windows XP Professional
  Model:                   INTEL_ D915GEV_
  Motherboard:             Intel Corporation D915GEV
  Processor:                Genuine Intel(R) CPU 3.60GHz
  Processor ID:             GenuineIntel Family 15 Model 3 Steppin
  Logical Processor Count: 2
  Physical Processor Count:  1
  Processor Frequency:    3600 MHz
  Bus Frequency:          200 MHz
  Memory:                   1014 MB

Integer Performance
  Emulate 6502
single-threaded scalar 198.4 (rate: 1.0, result: 375.1 MHz)
multi-threaded scalar 190.0 (rate: 1.0, result: 359.1 MHz)
  Blowfish
single-threaded scalar 213.3 (rate: 1.0, result: 88.0 MB/sec)
multi-threaded scalar 213.3 (rate: 1.0, result: 88.0 MB/sec)
  bzip2 Compress
single-threaded scalar 309.8 (rate: 1.0, result: 48.3 MB/sec)
multi-threaded scalar 311.8 (rate: 1.0, result: 48.4 MB/sec)
  bzip2 Decompress
single-threaded scalar 280.0 (rate: 1.0, result: 104.1 MB/sec
multi-threaded scalar 288.6 (rate: 1.0, result: 103.9 MB/sec

Floating Point Performance
  Mandelbrot
single-threaded scalar 117.3 (rate: 1.0, result: 831.4 Mflops
multi-threaded scalar 117.3 (rate: 1.0, result: 831.3 Mflops
  Dot Product
single-threaded scalar 50.5 (rate: 1.0, result: 260.3 Mflops
multi-threaded scalar    50.2 (rate: 1.0, result: 260.8 Mflops
single-threaded vector 149.3 (rate: 8.1, result: 2.1 Gflops)
multi-threaded vector 148.4 (rate: 8.2, result: 2.1 Gflops)
  JPEG Compress
single-threaded scalar 138.1 (rate: 1.0, result: 12.8 Mpixels
multi-threaded scalar 138.2 (rate: 1.0, result: 12.8 Mpixels
  JPEG Decompress
single-threaded scalar 187.9 (rate: 1.0, result: 31.2 Mpixels
multi-threaded scalar 181.4 (rate: 1.0, result: 30.1 Mpixels

Memory Performance
  Read Sequential
single-threaded scalar 262.8 (rate: 1.0, result: 3.3 GB/sec)
multi-threaded scalar 266.8 (rate: 0.5, result: 1.6 GB/sec)
  Write Sequential
single-threaded scalar 266.7 (rate: 1.0, result: 2.0 GB/sec)
multi-threaded scalar 266.1 (rate: 0.5, result: 1022.1 MB/se
  Stdlib Allocate
single-threaded scalar 83.4 (rate: 1.0, result: 2.9 Mallocs/
multi-threaded scalar    82.2 (rate: 1.0, result: 2.9 Mallocs/
  Stdlib Write
single-threaded scalar 98.6 (rate: 1.0, result: 2.5 GB/sec)
multi-threaded scalar    87.2 (rate: 0.8, result: 2.0 GB/sec)
  Stdlib Copy
single-threaded scalar 112.0 (rate: 1.0, result: 1.2 GB/sec)
multi-threaded scalar 117.0 (rate: 1.0, result: 1.2 GB/sec)

Stream Performance
  Stream Copy
single-threaded scalar 187.2 (rate: 1.0, result: 2.3 GB/sec)
multi-threaded scalar 188.8 (rate: 1.0, result: 2.4 GB/sec)
single-threaded vector 182.5 (rate: 1.1, result: 2.5 GB/sec)
multi-threaded vector 183.3 (rate: 1.1, result: 2.5 GB/sec)
  Stream Scale
single-threaded scalar 191.8 (rate: 1.0, result: 2.2 GB/sec)
multi-threaded scalar 192.5 (rate: 1.0, result: 2.3 GB/sec)
single-threaded vector 176.3 (rate: 1.1, result: 2.4 GB/sec)
multi-threaded vector 178.6 (rate: 1.1, result: 2.5 GB/sec)
  Stream Add
single-threaded scalar 192.4 (rate: 1.0, result: 2.5 GB/sec)
multi-threaded scalar 187.7 (rate: 1.0, result: 2.5 GB/sec)
single-threaded vector 187.6 (rate: 1.0, result: 2.6 GB/sec)
multi-threaded vector 186.8 (rate: 1.1, result: 2.7 GB/sec)
  Stream Triad
single-threaded scalar 190.2 (rate: 1.0, result: 2.5 GB/sec)
multi-threaded scalar 187.4 (rate: 1.0, result: 2.5 GB/sec)
single-threaded vector 149.7 (rate: 1.0, result: 2.6 GB/sec)
multi-threaded vector 148.6 (rate: 1.1, result: 2.6 GB/sec)

Overall Score: 178.1

xeon-pan · 发表于 2006-11-23 12:31

比p4还差？:huh: :huh:

HardCoded · 发表于 2006-11-23 12:33

呵呵，如我所料，果然是C4级别的通用计算水平，牛啊

hopetoknow2 · 发表于 2006-11-23 12:36

Ultra 20 M2: 1.8G
* AMD Dual-Core Opteron 1210
* 512 MB DDR2-667 (1 DIMM)
* Windows XP Professional x64 Edition
* Geekbench 2006 (build 230)

Overall Score: 147.6

Integer Performance
  Emulate 6502
single-threaded scalar 76.8
multi-threaded scalar 153.7
  Blowfish
single-threaded scalar 118.6
multi-threaded scalar 237.3
  bzip2 Compress
single-threaded scalar 180.6
multi-threaded scalar 364.3
  bzip2 Decompress
single-threaded scalar 140.8
multi-threaded scalar 300.9

Floating Point Performance
  Mandelbrot
single-threaded scalar 121.6
multi-threaded scalar 243.2
  Dot Product
single-threaded scalar 57.2
multi-threaded scalar    113.4
  JPEG Compress
single-threaded scalar 117.9
multi-threaded scalar 236.2
  JPEG Decompress
single-threaded scalar 113.2
multi-threaded scalar 220.6

Memory Performance
  Read Sequential
single-threaded scalar 150.1
multi-threaded scalar 108.9
  Write Sequential
single-threaded scalar 167.0
multi-threaded scalar 214.4
  Stdlib Allocate
single-threaded scalar 87.6
multi-threaded scalar    37.0
  Stdlib Write
single-threaded scalar 60.6
multi-threaded scalar    89.4
  Stdlib Copy
single-threaded scalar 78.2
multi-threaded scalar 99.4

Stream Performance
  Stream Copy
single-threaded scalar 141.3
multi-threaded scalar 168.9
  Stream Scale
single-threaded scalar 132.5
multi-threaded scalar 172.9
  Stream Add
single-threaded scalar 100.8
multi-threaded scalar 156.5
Stream Triad
single-threaded scalar 91
multi-threaded scalar 153

[ 本帖最后由 hopetoknow2 于 2006-11-23 12:53 编辑 ]

只看该作者 · 发表于 2006-11-23 12:38

提示: 作者被禁止或删除内容自动屏蔽

HardCoded · 发表于 2006-11-23 12:39

原帖由 xeon-pan 于 2006-11-23 12:31 发表
比p4还差？:huh: :huh:

:lol: 请CELL先和K7比，然后才有资格和P4比。和Conroe比？那简直就是秒杀

xeon-pan · 发表于 2006-11-23 12:42

原帖由 HardCoded 于 2006-11-23 12:39 发表

:lol: 请CELL先和K7比，然后才有资格和P4比。和Conroe比？那简直就是秒杀

你没弄明白我的意思吧....当初sony吹cell比高端p4块100000000000..........00000000000倍的。。。:lol:

HardCoded · 发表于 2006-11-23 12:45

原帖由 xeon-pan 于 2006-11-23 12:42 发表

你没弄明白我的意思吧....当初sony吹cell比高端p4块100000000000..........00000000000倍的。。。:lol:

:lol: 我知道，人家还能模拟地球呢

skywalker_hao · 发表于 2006-11-23 12:54

Mac Pro (ID 10052)
Overall Score: 383.2
System Information
Metric Value
Geekbench Version Geekbench 2006 (build 242)
Geekbench Platform Mac OS X x86 (64-bit)
Geekbench Compiler GCC 4.0.1 (Apple Computer, Inc. build 5363)
OS Mac OS X (Darwin 8.8.1)
Model Mac Pro
Motherboard MacPro1,1
Processor Intel(R) Xeon(R) CPU 5150 @ 2.66GHz
Processor ID GenuineIntel Family 6 Model 15 Stepping 6
Logical Processor Count 4
Physical Processor Count 4
Processor Frequency 2660 MHz
Bus Frequency 1332 MHz
Memory 8192 MB

Integer Performance
Benchmark Score Rate Result
Emulate 6502
single-threaded scalar 218.4 1.0 412.9 MHz
Emulate 6502
multi-threaded scalar 868.2 4.0 1.6 GHz
Blowfish
single-threaded scalar 205.2 1.0 84.7 MB/sec
Blowfish
multi-threaded scalar 817.6 4.0 337.2 MB/sec
bzip2 Compress
single-threaded scalar 282.2 1.0 44.0 MB/sec
bzip2 Compress
multi-threaded scalar 1057.5 3.7 164.0 MB/sec
bzip2 Decompress
single-threaded scalar 298.2 1.0 110.9 MB/sec
bzip2 Decompress
multi-threaded scalar 1099.4 3.6 396.0 MB/sec

Floating Point Performance
Benchmark Score Rate Result
Mandelbrot
single-threaded scalar 180.0 1.0 1.3 Gflops
Mandelbrot
multi-threaded scalar 718.5 4.0 5.1 Gflops
Dot Product
single-threaded scalar 321.9 1.0 1.7 Gflops
Dot Product
multi-threaded scalar 1110.7 3.5 5.8 Gflops
Dot Product
single-threaded vector 155.9 1.3 2.2 Gflops
Dot Product
multi-threaded vector 507.3 4.4 7.3 Gflops
JPEG Compress
single-threaded scalar 196.8 1.0 18.3 Mpixels/sec
JPEG Compress
multi-threaded scalar 780.2 4.0 72.3 Mpixels/sec
JPEG Decompress
single-threaded scalar 187.4 1.0 31.1 Mpixels/sec
JPEG Decompress
multi-threaded scalar 640.6 3.4 106.2 Mpixels/sec

Memory Performance
Benchmark Score Rate Result
Read Sequential
single-threaded scalar 318.0 1.0 4.0 GB/sec
Read Sequential
multi-threaded scalar 174.4 0.3 1.1 GB/sec
Write Sequential
single-threaded scalar 524.3 1.0 4.0 GB/sec
Write Sequential
multi-threaded scalar 290.4 0.3 1.1 GB/sec
Stdlib Allocate
single-threaded scalar 360.3 1.0 12.7 Mallocs/sec
Stdlib Allocate
multi-threaded scalar 60.6 0.2 2.2 Mallocs/sec
Stdlib Write
single-threaded scalar 132.8 1.0 3.4 GB/sec
Stdlib Write
multi-threaded scalar 215.8 1.5 5.1 GB/sec
Stdlib Copy
single-threaded scalar 260.9 1.0 2.8 GB/sec
Stdlib Copy
multi-threaded scalar 453.2 1.6 4.7 GB/sec

Stream Performance
Benchmark Score Rate Result
Stream Copy
single-threaded scalar 212.7 1.0 2.7 GB/sec
Stream Copy
multi-threaded scalar 358.7 1.7 4.5 GB/sec
Stream Copy
single-threaded vector 204.5 1.0 2.8 GB/sec
Stream Copy
multi-threaded vector 329.2 1.7 4.5 GB/sec
Stream Scale
single-threaded scalar 224.4 1.0 2.6 GB/sec
Stream Scale
multi-threaded scalar 381.2 1.7 4.5 GB/sec
Stream Scale
single-threaded vector 202.9 1.1 2.8 GB/sec
Stream Scale
multi-threaded vector 328.0 1.7 4.5 GB/sec
Stream Add
single-threaded scalar 225.8 1.0 2.9 GB/sec
Stream Add
multi-threaded scalar 360.0 1.7 4.9 GB/sec
Stream Add
single-threaded vector 216.3 1.0 3.0 GB/sec
Stream Add
multi-threaded vector 343.3 1.7 4.9 GB/sec
Stream Triad
single-threaded scalar 225.6 1.0 2.9 GB/sec
Stream Triad
multi-threaded scalar 362.6 1.7 4.9 GB/sec
Stream Triad
single-threaded vector 174.3 1.0 3.0 GB/sec
Stream Triad
multi-threaded vector 276.2 1.7 4.9 GB/sec
:p

[ 本帖最后由 skywalker_hao 于 2006-11-23 12:56 编辑 ]

hopetoknow2 · 发表于 2006-11-23 12:59

距离超级大

hopetoknow2 · 发表于 2006-11-23 12:59

再来一次，超级大

Edison · 发表于 2006-11-23 13:00

单独1个2 -issue in-order FGMT的PPE跑出来Dot Product就几乎把3-issue OOO SMT的P4踩在脚下了。

z1978 · 发表于 2006-11-23 13:09

原帖由 Edison 于 2006-11-23 13:00 发表
单独1个2 -issue in-order FGMT的PPE跑出来Dot Product就几乎把3-issue OOO SMT的P4踩在脚下了。

DOT PRODUCT是决定什么呢

Prescott · 发表于 2006-11-23 13:12

原帖由 Edison 于 2006-11-23 13:00 发表
单独1个2 -issue in-order FGMT的PPE跑出来Dot Product就几乎把3-issue OOO SMT的P4踩在脚下了。

让x87跑dot product，是不是有些残忍？
当SSE3不存在是不是？Vector之后Cell还不是一样被踩。
(_(

[ 本帖最后由 Prescott 于 2006-11-23 13:14 编辑 ]

Edison · 发表于 2006-11-23 13:12

原帖由 z1978 于 2006-11-23 13:09 发表
DOT PRODUCT是决定什么呢

物理运算、图形分析计算都会大量包含dot product。

Edison · 发表于 2006-11-23 13:16

原帖由 Prescott 于 2006-11-23 13:12 发表
让x87跑dot product，是不是有些残忍？
当SSE3不存在是不是？Vector之后Cell还不是一样被踩。(_(

你要忽略VMX、SPE的话，为什么不把SSE也忽略掉呢？

FENG950 · 发表于 2006-11-23 13:17

这个成绩有什么，不过只是PPE的能力反应而已，Cell如果只是一块PPE的用途，那它和别的CPU比也没有任何特别之处了。对比P4的性能，所指当然是某方面的性能比较，当初的宣传是指所有方面都比高端P4快吗？何况这PPE还是一个2发射按序核心，就算C4一级，也都不只2 issue了吧？这里的测试软件，是否有按有序核心作标准进行优化设计了呢？

帐号		自动登录	找回密码
密码			注册

potomac 该用户已被删除	2^# 发表于 2006-11-23 12:15 \| 只看该作者提示: 作者被禁止或删除内容自动屏蔽
potomac 该用户已被删除
	回复支持反对使用道具举报显身卡

potomac 该用户已被删除	8^# 发表于 2006-11-23 12:38 \| 只看该作者提示: 作者被禁止或删除内容自动屏蔽
potomac 该用户已被删除
	回复支持反对使用道具举报显身卡

Linux下Cell 3.2G vs. PPC G5 1.6G对比测试出炉

浏览过的版块