歴史のあるクラウドサービスは、どこもそうなってしまう傾向があるとは思いますが、ホストサーバでの実CPUのアーキテクチャ・世代の違いで、サーバインスタンスのCPUパフォーマンスに微妙な差がついてしまいます。
2006年よりサービス提供しているAmazon EC2でもその傾向があることは割と知られていて、同じ性能だと思って並べて使っていたサーバインスタンスが、同じ処理量にもかかわらず使っているCPUリソースに差がついている、なんてことが起こります。
con_mameさんも、以下のエントリで書かれていますね。
昔は、us-eastでm1.smallのインスタンスをよく使ったもので、その頃はいつもAMDのOpteronプロセッサでしたが、最近では、ほとんどIntel Xeonですし。
ということで、現時点(2013/10)で、EC2インスタンスで使われているCPUの種類と傾向を探ってみることにしました。(インスタンスガチャ)
対象としたインスタンスタイプとリージョン
上記のようなAmazon EC2で扱えるインスタンスタイプを見ていると、ハイエンドなインスタンスタイプを除くと、1core(1vCPU)あたりのECUが、
- 1ECU
- 2ECU
- 2.5ECU
- 3.25ECU
の4種類にわかれています。3年以上前の情報ですが、過去のベンチマークエントリ「Amazon EC2性能検証！気になるパフォーマンスをインスタンスタイプやリージョン毎に計測・比較してみた」の際に、インスタンスで使われていたCPUを見ると、1vCPUが1ECUのm1.smallは、m1.large等の1vCPUあたり2ECUのインスタンスと、同じCPUに見えて、スケジューリング等で半分だけ使えるような感じで制御されているっぽかったので、インスタンスガチャの対象は、
- 2ECU / 1vCPU => m1.medium
- 2.5ECU / 1vCPU => c1.medium
- 3.25ECU / 1vCPU => m2.xlarge
の3つにしました。
尚、リージョンについては、普段よく使っている ap-northeast-1 (東京) を対象にしました。ちなみにリージョン内の、Availabirity Zoneについては、全てのゾーンを満遍なく使うイメージです。
前提
上記の3インスタンスタイプとリージョンが対象です。
ガチャを引いた回数(インスタンスを起動した回数)は、各パターンごとに50回です。
(サンプリングとして少ないのは承知の上ですが、スポンサーがついたら時間帯も変動させつつ1000回くらいやりたいですｗ)
おまけ的なものとして、インスタンスタイプでCPUに差異が出た場合に、UnixBenchを走らせてみました。参考までにスコアを載せておきますが、これがCPU性能の全てというわけではないので、あくまで参考までに。
(尚、UnixBenchのスコアは、"System Benchmark Index Score"を記載しています。エントリの最後に全てのスコアを一応載せておきます。)
結果
結果は、以下の通りとなりました。
m1.medium (2ECU/1vCPU) @ Tokyo
| 出現回数 | CPU型番 | UnixBench Score |
|---|---|---|
| 47 | Intel(R) Xeon(R) CPU E5-2650 0 @ 2.00GHz | 335.5 |
| 3 | Intel(R) Xeon(R) CPU E5645 @ 2.40GHz | 368.9 |
c1.medium (2.5ECU/1vCPU) @ Tokyo
| 出現回数 | CPU型番 | UnixBench Score (1para/2para) |
|---|---|---|
| 28 | Intel(R) Xeon(R) CPU E5-2650 0 @ 2.00GHz | 336.4 / 614.7 |
| 22 | Intel(R) Xeon(R) CPU E5506 @ 2.13GHz | 380.8 / 662.8 |
m2.xlarge (3.25ECU/1vCPU) @ Tokyo
| 出現回数 | CPU型番 | UnixBench Score (1para/2para) |
|---|---|---|
| 46 | Intel(R) Xeon(R) CPU E5-2665 0 @ 2.40GHz | 445.0 / 811.2 |
| 4 | Intel(R) Xeon(R) CPU X5550 @ 2.67GHz | 504.9 / 937.6 |
最近のEC2インスタンスガチャでのCPU出現頻度は、僕の手元でやった感じだと、↑のような確率でした。
利用CPUは、割とSandy Bridge世代に寄ってきていることが確認できましたが、やはり旧世代のCPUについても、ちょこちょこ出てくるような印象です。(特に2.5ECU/1vCPUのインスタンス)
あと、2ECUである"m1.medium"と、2.5ECUである"c1.medium"で同じ「Xeon E5-2650」が使われているのも意外でした。0.5ECU分はどこで差をつけているのかわからないですが、仮想ハイパーバイザとかのレイヤで何か特別なCPUスケジューリングをしているのかもしれません。(UnixBenchのスコアも大体同じだったので分からず)
ちなみに、最新のCPUアーキテクチャの方が、より性能が良いはずなので、クロック数を落としたものが旧型CPUと同列で並んでいることには納得ですが、UnixBenchのスコアが意外と並ばないことが気になりました。(あくまで特定処理のベンチマークではあるので、この通りのCPUパフォーマンスとはならないとは思いますが。参考までに)
ひょっとしたら、場合(処理)によっては、旧世代のCPUを積んだインスタンスの方が性能が良い可能性はあります。(レアキャラですね...)
おまけ: m1.small (1ECU / 1vCPU) @ US-east
おまけとして、それはそれは昔よく使っていたus-east-1の"m1.small"で20回だけインスタンスガチャをやってみました。その結果が以下。
| 出現回数 | CPU型番 | UnixBench Score |
|---|---|---|
| 8 | Intel(R) Xeon(R) CPU E5507 @ 2.27GHz | 167.8 |
| 7 | Intel(R) Xeon(R) CPU E5645 @ 2.40GHz | 182.5 |
| 5 | Intel(R) Xeon(R) CPU E5430 @ 2.66GHz | 174.4 |
us-east-1は最古リージョンだけあって、まだHarpertown(E54xx)世代のCPUが生き残っていますね！
と、いつもの如く適当に書き殴ったようなエントリですが、最近やったEC2インスタンスガチャの結果でしたー。
他に、こんなレアキャラ出ましたー！的な情報がございましたら、是非お教えくださいｗ
最後に、UnixBenchの実行結果(詳細)をAppendixとして最後に残しておきます。
それでは！=͟͟͞͞(๑•̀=͟͟͞͞(๑•̀д•́=͟͟͞͞(๑•̀д•́๑)=͟͟͞͞(๑•̀д•́
まとめ
クラウドAMAZON EC2/S3のすべて (ITpro BOOKs)
- 作者: 並河祐貴,安達輝雄,ITpro/日経SYSTEMS
- 出版社/メーカー: 日経BP社
- 発売日: 2009/11/05
- メディア: 単行本
- 購入: 4人 クリック: 372回
- この商品を含むブログ (18件) を見る
Appendix (UnixBenchの実行結果)
m1.medium (2ECU/1vCPU) @ Tokyo
========================================================================
BYTE UNIX Benchmarks (Version 5.1.3)
System: ip-10-96-2-51: GNU/Linux
OS: GNU/Linux -- 3.4.43-43.43.amzn1.x86_64 -- #1 SMP Mon May 6 18:04:41 UTC 2013
Machine: x86_64 (x86_64)
Language: en_US.utf8 (charmap="UTF-8", collate="UTF-8")
CPU 0: Intel(R) Xeon(R) CPU E5-2650 0 @ 2.00GHz (3600.0 bogomips)
Hyper-Threading, x86-64, MMX, Physical Address Ext, SYSENTER/SYSEXIT, SYSCALL/SYSRET
07:24:16 up 6 min, 1 user, load average: 0.14, 0.09, 0.05; runlevel 3
------------------------------------------------------------------------
Benchmark Run: Fri Sep 27 2013 07:24:16 - 07:52:14
1 CPU in system; running 1 parallel copy of tests
Dhrystone 2 using register variables 20788313.1 lps (10.0 s, 7 samples)
Double-Precision Whetstone 1203.9 MWIPS (10.0 s, 7 samples)
Execl Throughput 1032.0 lps (30.0 s, 2 samples)
File Copy 1024 bufsize 2000 maxblocks 208407.3 KBps (30.0 s, 2 samples)
File Copy 256 bufsize 500 maxblocks 54001.5 KBps (30.0 s, 2 samples)
File Copy 4096 bufsize 8000 maxblocks 702701.7 KBps (30.0 s, 2 samples)
Pipe Throughput 276267.5 lps (10.0 s, 7 samples)
Pipe-based Context Switching 50442.6 lps (10.0 s, 7 samples)
Process Creation 2086.6 lps (30.0 s, 2 samples)
Shell Scripts (1 concurrent) 1644.8 lpm (60.0 s, 2 samples)
Shell Scripts (8 concurrent) 223.6 lpm (60.1 s, 2 samples)
System Call Overhead 233529.0 lps (10.0 s, 7 samples)
System Benchmarks Index Values BASELINE RESULT INDEX
Dhrystone 2 using register variables 116700.0 20788313.1 1781.3
Double-Precision Whetstone 55.0 1203.9 218.9
Execl Throughput 43.0 1032.0 240.0
File Copy 1024 bufsize 2000 maxblocks 3960.0 208407.3 526.3
File Copy 256 bufsize 500 maxblocks 1655.0 54001.5 326.3
File Copy 4096 bufsize 8000 maxblocks 5800.0 702701.7 1211.6
Pipe Throughput 12440.0 276267.5 222.1
Pipe-based Context Switching 4000.0 50442.6 126.1
Process Creation 126.0 2086.6 165.6
Shell Scripts (1 concurrent) 42.4 1644.8 387.9
Shell Scripts (8 concurrent) 6.0 223.6 372.7
System Call Overhead 15000.0 233529.0 155.7
========
System Benchmarks Index Score 335.5========================================================================
BYTE UNIX Benchmarks (Version 5.1.3)
System: ip-10-96-2-19: GNU/Linux
OS: GNU/Linux -- 3.4.43-43.43.amzn1.x86_64 -- #1 SMP Mon May 6 18:04:41 UTC 2013
Machine: x86_64 (x86_64)
Language: en_US.utf8 (charmap="UTF-8", collate="UTF-8")
CPU 0: Intel(R) Xeon(R) CPU E5645 @ 2.40GHz (4000.1 bogomips)
Hyper-Threading, x86-64, MMX, Physical Address Ext, SYSENTER/SYSEXIT, SYSCALL/SYSRET
07:24:55 up 6 min, 1 user, load average: 0.15, 0.07, 0.05; runlevel 3
------------------------------------------------------------------------
Benchmark Run: Fri Sep 27 2013 07:24:55 - 07:52:51
1 CPU in system; running 1 parallel copy of tests
Dhrystone 2 using register variables 21065967.3 lps (10.0 s, 7 samples)
Double-Precision Whetstone 1485.3 MWIPS (9.9 s, 7 samples)
Execl Throughput 1138.8 lps (30.0 s, 2 samples)
File Copy 1024 bufsize 2000 maxblocks 229330.2 KBps (30.0 s, 2 samples)
File Copy 256 bufsize 500 maxblocks 59391.5 KBps (30.0 s, 2 samples)
File Copy 4096 bufsize 8000 maxblocks 695318.8 KBps (30.0 s, 2 samples)
Pipe Throughput 316523.7 lps (10.0 s, 7 samples)
Pipe-based Context Switching 56549.1 lps (10.0 s, 7 samples)
Process Creation 2220.4 lps (30.0 s, 2 samples)
Shell Scripts (1 concurrent) 1804.0 lpm (60.0 s, 2 samples)
Shell Scripts (8 concurrent) 243.4 lpm (60.2 s, 2 samples)
System Call Overhead 270786.4 lps (10.0 s, 7 samples)
System Benchmarks Index Values BASELINE RESULT INDEX
Dhrystone 2 using register variables 116700.0 21065967.3 1805.1
Double-Precision Whetstone 55.0 1485.3 270.1
Execl Throughput 43.0 1138.8 264.8
File Copy 1024 bufsize 2000 maxblocks 3960.0 229330.2 579.1
File Copy 256 bufsize 500 maxblocks 1655.0 59391.5 358.9
File Copy 4096 bufsize 8000 maxblocks 5800.0 695318.8 1198.8
Pipe Throughput 12440.0 316523.7 254.4
Pipe-based Context Switching 4000.0 56549.1 141.4
Process Creation 126.0 2220.4 176.2
Shell Scripts (1 concurrent) 42.4 1804.0 425.5
Shell Scripts (8 concurrent) 6.0 243.4 405.6
System Call Overhead 15000.0 270786.4 180.5
========
System Benchmarks Index Score 368.9
c1.medium (2.5ECU/1vCPU) @ Tokyo
========================================================================
BYTE UNIX Benchmarks (Version 5.1.3)
System: ip-10-96-2-211: GNU/Linux
OS: GNU/Linux -- 3.4.62-53.42.amzn1.x86_64 -- #1 SMP Fri Sep 20 07:23:24 UTC 2013
Machine: x86_64 (x86_64)
Language: en_US.utf8 (charmap="UTF-8", collate="UTF-8")
CPU 0: Intel(R) Xeon(R) CPU E5-2650 0 @ 2.00GHz (3600.0 bogomips)
Hyper-Threading, x86-64, MMX, Physical Address Ext, SYSENTER/SYSEXIT, SYSCALL/SYSRET
CPU 1: Intel(R) Xeon(R) CPU E5-2650 0 @ 2.00GHz (3600.0 bogomips)
Hyper-Threading, x86-64, MMX, Physical Address Ext, SYSENTER/SYSEXIT, SYSCALL/SYSRET
07:10:29 up 55 min, 1 user, load average: 0.14, 0.06, 0.06; runlevel 3
------------------------------------------------------------------------
Benchmark Run: Tue Oct 01 2013 07:10:30 - 07:38:27
2 CPUs in system; running 1 parallel copy of tests
Dhrystone 2 using register variables 20801975.3 lps (10.0 s, 7 samples)
Double-Precision Whetstone 1203.7 MWIPS (10.0 s, 7 samples)
Execl Throughput 961.1 lps (30.0 s, 2 samples)
File Copy 1024 bufsize 2000 maxblocks 196894.5 KBps (30.0 s, 2 samples)
File Copy 256 bufsize 500 maxblocks 51055.3 KBps (30.0 s, 2 samples)
File Copy 4096 bufsize 8000 maxblocks 645640.3 KBps (30.0 s, 2 samples)
Pipe Throughput 269836.6 lps (10.0 s, 7 samples)
Pipe-based Context Switching 31917.7 lps (10.0 s, 7 samples)
Process Creation 1959.0 lps (30.0 s, 2 samples)
Shell Scripts (1 concurrent) 2138.1 lpm (60.0 s, 2 samples)
Shell Scripts (8 concurrent) 401.6 lpm (60.1 s, 2 samples)
System Call Overhead 233219.4 lps (10.0 s, 7 samples)
System Benchmarks Index Values BASELINE RESULT INDEX
Dhrystone 2 using register variables 116700.0 20801975.3 1782.5
Double-Precision Whetstone 55.0 1203.7 218.8
Execl Throughput 43.0 961.1 223.5
File Copy 1024 bufsize 2000 maxblocks 3960.0 196894.5 497.2
File Copy 256 bufsize 500 maxblocks 1655.0 51055.3 308.5
File Copy 4096 bufsize 8000 maxblocks 5800.0 645640.3 1113.2
Pipe Throughput 12440.0 269836.6 216.9
Pipe-based Context Switching 4000.0 31917.7 79.8
Process Creation 126.0 1959.0 155.5
Shell Scripts (1 concurrent) 42.4 2138.1 504.3
Shell Scripts (8 concurrent) 6.0 401.6 669.3
System Call Overhead 15000.0 233219.4 155.5
========
System Benchmarks Index Score 336.4
------------------------------------------------------------------------
Benchmark Run: Tue Oct 01 2013 07:38:27 - 08:06:24
2 CPUs in system; running 2 parallel copies of tests
Dhrystone 2 using register variables 41563180.4 lps (10.0 s, 7 samples)
Double-Precision Whetstone 2407.9 MWIPS (10.0 s, 7 samples)
Execl Throughput 1851.3 lps (30.0 s, 2 samples)
File Copy 1024 bufsize 2000 maxblocks 356448.4 KBps (30.0 s, 2 samples)
File Copy 256 bufsize 500 maxblocks 92261.5 KBps (30.0 s, 2 samples)
File Copy 4096 bufsize 8000 maxblocks 1172433.5 KBps (30.0 s, 2 samples)
Pipe Throughput 538535.2 lps (10.0 s, 7 samples)
Pipe-based Context Switching 94941.3 lps (10.0 s, 7 samples)
Process Creation 3613.7 lps (30.0 s, 2 samples)
Shell Scripts (1 concurrent) 3004.9 lpm (60.0 s, 2 samples)
Shell Scripts (8 concurrent) 410.2 lpm (60.1 s, 2 samples)
System Call Overhead 449388.8 lps (10.0 s, 7 samples)
System Benchmarks Index Values BASELINE RESULT INDEX
Dhrystone 2 using register variables 116700.0 41563180.4 3561.5
Double-Precision Whetstone 55.0 2407.9 437.8
Execl Throughput 43.0 1851.3 430.5
File Copy 1024 bufsize 2000 maxblocks 3960.0 356448.4 900.1
File Copy 256 bufsize 500 maxblocks 1655.0 92261.5 557.5
File Copy 4096 bufsize 8000 maxblocks 5800.0 1172433.5 2021.4
Pipe Throughput 12440.0 538535.2 432.9
Pipe-based Context Switching 4000.0 94941.3 237.4
Process Creation 126.0 3613.7 286.8
Shell Scripts (1 concurrent) 42.4 3004.9 708.7
Shell Scripts (8 concurrent) 6.0 410.2 683.7
System Call Overhead 15000.0 449388.8 299.6
========
System Benchmarks Index Score 614.7========================================================================
BYTE UNIX Benchmarks (Version 5.1.3)
System: ip-10-96-2-33: GNU/Linux
OS: GNU/Linux -- 3.4.62-53.42.amzn1.x86_64 -- #1 SMP Fri Sep 20 07:23:24 UTC 2013
Machine: x86_64 (x86_64)
Language: en_US.utf8 (charmap="UTF-8", collate="UTF-8")
CPU 0: Intel(R) Xeon(R) CPU E5506 @ 2.13GHz (4266.8 bogomips)
Hyper-Threading, x86-64, MMX, Physical Address Ext, SYSENTER/SYSEXIT, SYSCALL/SYSRET
CPU 1: Intel(R) Xeon(R) CPU E5506 @ 2.13GHz (4266.8 bogomips)
Hyper-Threading, x86-64, MMX, Physical Address Ext, SYSENTER/SYSEXIT, SYSCALL/SYSRET
07:10:14 up 54 min, 1 user, load average: 0.50, 0.13, 0.08; runlevel 3
------------------------------------------------------------------------
Benchmark Run: Tue Oct 01 2013 07:10:14 - 07:38:10
2 CPUs in system; running 1 parallel copy of tests
Dhrystone 2 using register variables 22517135.5 lps (10.0 s, 7 samples)
Double-Precision Whetstone 1549.1 MWIPS (10.1 s, 7 samples)
Execl Throughput 1082.8 lps (30.0 s, 2 samples)
File Copy 1024 bufsize 2000 maxblocks 222821.4 KBps (30.0 s, 2 samples)
File Copy 256 bufsize 500 maxblocks 61863.1 KBps (30.0 s, 2 samples)
File Copy 4096 bufsize 8000 maxblocks 637762.2 KBps (30.0 s, 2 samples)
Pipe Throughput 327767.7 lps (10.0 s, 7 samples)
Pipe-based Context Switching 33825.1 lps (10.0 s, 7 samples)
Process Creation 2119.6 lps (30.0 s, 2 samples)
Shell Scripts (1 concurrent) 2375.6 lpm (60.0 s, 2 samples)
Shell Scripts (8 concurrent) 431.4 lpm (60.1 s, 2 samples)
System Call Overhead 291513.1 lps (10.0 s, 7 samples)
System Benchmarks Index Values BASELINE RESULT INDEX
Dhrystone 2 using register variables 116700.0 22517135.5 1929.5
Double-Precision Whetstone 55.0 1549.1 281.7
Execl Throughput 43.0 1082.8 251.8
File Copy 1024 bufsize 2000 maxblocks 3960.0 222821.4 562.7
File Copy 256 bufsize 500 maxblocks 1655.0 61863.1 373.8
File Copy 4096 bufsize 8000 maxblocks 5800.0 637762.2 1099.6
Pipe Throughput 12440.0 327767.7 263.5
Pipe-based Context Switching 4000.0 33825.1 84.6
Process Creation 126.0 2119.6 168.2
Shell Scripts (1 concurrent) 42.4 2375.6 560.3
Shell Scripts (8 concurrent) 6.0 431.4 719.0
System Call Overhead 15000.0 291513.1 194.3
========
System Benchmarks Index Score 380.8
------------------------------------------------------------------------
Benchmark Run: Tue Oct 01 2013 07:38:10 - 08:06:06
2 CPUs in system; running 2 parallel copies of tests
Dhrystone 2 using register variables 42805578.9 lps (10.0 s, 7 samples)
Double-Precision Whetstone 2991.0 MWIPS (9.9 s, 7 samples)
Execl Throughput 2036.9 lps (30.0 s, 2 samples)
File Copy 1024 bufsize 2000 maxblocks 365120.0 KBps (30.0 s, 2 samples)
File Copy 256 bufsize 500 maxblocks 101638.8 KBps (30.0 s, 2 samples)
File Copy 4096 bufsize 8000 maxblocks 985152.5 KBps (30.0 s, 2 samples)
Pipe Throughput 623913.4 lps (10.0 s, 7 samples)
Pipe-based Context Switching 109529.7 lps (10.0 s, 7 samples)
Process Creation 3914.0 lps (30.0 s, 2 samples)
Shell Scripts (1 concurrent) 3217.4 lpm (60.0 s, 2 samples)
Shell Scripts (8 concurrent) 430.0 lpm (60.2 s, 2 samples)
System Call Overhead 511078.8 lps (10.0 s, 7 samples)
System Benchmarks Index Values BASELINE RESULT INDEX
Dhrystone 2 using register variables 116700.0 42805578.9 3668.0
Double-Precision Whetstone 55.0 2991.0 543.8
Execl Throughput 43.0 2036.9 473.7
File Copy 1024 bufsize 2000 maxblocks 3960.0 365120.0 922.0
File Copy 256 bufsize 500 maxblocks 1655.0 101638.8 614.1
File Copy 4096 bufsize 8000 maxblocks 5800.0 985152.5 1698.5
Pipe Throughput 12440.0 623913.4 501.5
Pipe-based Context Switching 4000.0 109529.7 273.8
Process Creation 126.0 3914.0 310.6
Shell Scripts (1 concurrent) 42.4 3217.4 758.8
Shell Scripts (8 concurrent) 6.0 430.0 716.7
System Call Overhead 15000.0 511078.8 340.7
========
System Benchmarks Index Score 662.8
m2.xlarge (3.25ECU/1vCPU) @ Tokyo
========================================================================
BYTE UNIX Benchmarks (Version 5.1.3)
System: ip-10-96-0-205: GNU/Linux
OS: GNU/Linux -- 3.4.43-43.43.amzn1.x86_64 -- #1 SMP Mon May 6 18:04:41 UTC 2013
Machine: x86_64 (x86_64)
Language: en_US.utf8 (charmap="UTF-8", collate="UTF-8")
CPU 0: Intel(R) Xeon(R) CPU E5-2665 0 @ 2.40GHz (4800.0 bogomips)
Hyper-Threading, x86-64, MMX, Physical Address Ext, SYSENTER/SYSEXIT, SYSCALL/SYSRET
CPU 1: Intel(R) Xeon(R) CPU E5-2665 0 @ 2.40GHz (4800.0 bogomips)
Hyper-Threading, x86-64, MMX, Physical Address Ext, SYSENTER/SYSEXIT, SYSCALL/SYSRET
09:01:33 up 1:03, 1 user, load average: 0.06, 0.59, 1.15; runlevel 3
------------------------------------------------------------------------
Benchmark Run: Fri Sep 27 2013 09:01:33 - 09:29:27
2 CPUs in system; running 1 parallel copy of tests
Dhrystone 2 using register variables 27235657.4 lps (10.0 s, 7 samples)
Double-Precision Whetstone 1602.5 MWIPS (9.9 s, 7 samples)
Execl Throughput 1249.9 lps (30.0 s, 2 samples)
File Copy 1024 bufsize 2000 maxblocks 265495.8 KBps (30.0 s, 2 samples)
File Copy 256 bufsize 500 maxblocks 69560.9 KBps (30.0 s, 2 samples)
File Copy 4096 bufsize 8000 maxblocks 898195.7 KBps (30.0 s, 2 samples)
Pipe Throughput 359900.4 lps (10.0 s, 7 samples)
Pipe-based Context Switching 41257.2 lps (10.0 s, 7 samples)
Process Creation 2538.5 lps (30.0 s, 2 samples)
Shell Scripts (1 concurrent) 2794.8 lpm (60.0 s, 2 samples)
Shell Scripts (8 concurrent) 520.3 lpm (60.1 s, 2 samples)
System Call Overhead 305004.6 lps (10.0 s, 7 samples)
System Benchmarks Index Values BASELINE RESULT INDEX
Dhrystone 2 using register variables 116700.0 27235657.4 2333.8
Double-Precision Whetstone 55.0 1602.5 291.4
Execl Throughput 43.0 1249.9 290.7
File Copy 1024 bufsize 2000 maxblocks 3960.0 265495.8 670.4
File Copy 256 bufsize 500 maxblocks 1655.0 69560.9 420.3
File Copy 4096 bufsize 8000 maxblocks 5800.0 898195.7 1548.6
Pipe Throughput 12440.0 359900.4 289.3
Pipe-based Context Switching 4000.0 41257.2 103.1
Process Creation 126.0 2538.5 201.5
Shell Scripts (1 concurrent) 42.4 2794.8 659.2
Shell Scripts (8 concurrent) 6.0 520.3 867.1
System Call Overhead 15000.0 305004.6 203.3
========
System Benchmarks Index Score 445.0
------------------------------------------------------------------------
Benchmark Run: Fri Sep 27 2013 09:29:27 - 09:57:21
2 CPUs in system; running 2 parallel copies of tests
Dhrystone 2 using register variables 54373456.8 lps (10.0 s, 7 samples)
Double-Precision Whetstone 3205.8 MWIPS (10.0 s, 7 samples)
Execl Throughput 2410.2 lps (29.8 s, 2 samples)
File Copy 1024 bufsize 2000 maxblocks 484540.0 KBps (30.0 s, 2 samples)
File Copy 256 bufsize 500 maxblocks 121182.4 KBps (30.0 s, 2 samples)
File Copy 4096 bufsize 8000 maxblocks 1596866.1 KBps (30.0 s, 2 samples)
Pipe Throughput 719233.8 lps (10.0 s, 7 samples)
Pipe-based Context Switching 124876.0 lps (10.0 s, 7 samples)
Process Creation 4664.3 lps (30.0 s, 2 samples)
Shell Scripts (1 concurrent) 3904.5 lpm (60.0 s, 2 samples)
Shell Scripts (8 concurrent) 538.3 lpm (60.1 s, 2 samples)
System Call Overhead 587314.6 lps (10.0 s, 7 samples)
System Benchmarks Index Values BASELINE RESULT INDEX
Dhrystone 2 using register variables 116700.0 54373456.8 4659.3
Double-Precision Whetstone 55.0 3205.8 582.9
Execl Throughput 43.0 2410.2 560.5
File Copy 1024 bufsize 2000 maxblocks 3960.0 484540.0 1223.6
File Copy 256 bufsize 500 maxblocks 1655.0 121182.4 732.2
File Copy 4096 bufsize 8000 maxblocks 5800.0 1596866.1 2753.2
Pipe Throughput 12440.0 719233.8 578.2
Pipe-based Context Switching 4000.0 124876.0 312.2
Process Creation 126.0 4664.3 370.2
Shell Scripts (1 concurrent) 42.4 3904.5 920.9
Shell Scripts (8 concurrent) 6.0 538.3 897.2
System Call Overhead 15000.0 587314.6 391.5
========
System Benchmarks Index Score 811.2========================================================================
BYTE UNIX Benchmarks (Version 5.1.3)
System: ip-10-96-0-249: GNU/Linux
OS: GNU/Linux -- 3.4.43-43.43.amzn1.x86_64 -- #1 SMP Mon May 6 18:04:41 UTC 2013
Machine: x86_64 (x86_64)
Language: en_US.utf8 (charmap="UTF-8", collate="UTF-8")
CPU 0: Intel(R) Xeon(R) CPU X5550 @ 2.67GHz (5333.5 bogomips)
Hyper-Threading, x86-64, MMX, Physical Address Ext, SYSENTER/SYSEXIT, SYSCALL/SYSRET
CPU 1: Intel(R) Xeon(R) CPU X5550 @ 2.67GHz (5333.5 bogomips)
Hyper-Threading, x86-64, MMX, Physical Address Ext, SYSENTER/SYSEXIT, SYSCALL/SYSRET
09:00:48 up 1:02, 1 user, load average: 0.01, 0.65, 1.18; runlevel 3
------------------------------------------------------------------------
Benchmark Run: Fri Sep 27 2013 09:00:48 - 09:28:42
2 CPUs in system; running 1 parallel copy of tests
Dhrystone 2 using register variables 28207297.1 lps (10.0 s, 7 samples)
Double-Precision Whetstone 1974.3 MWIPS (10.0 s, 7 samples)
Execl Throughput 1391.8 lps (30.0 s, 2 samples)
File Copy 1024 bufsize 2000 maxblocks 306464.8 KBps (30.0 s, 2 samples)
File Copy 256 bufsize 500 maxblocks 79316.2 KBps (30.0 s, 2 samples)
File Copy 4096 bufsize 8000 maxblocks 827535.5 KBps (30.0 s, 2 samples)
Pipe Throughput 423569.9 lps (10.0 s, 7 samples)
Pipe-based Context Switching 52534.3 lps (10.0 s, 7 samples)
Process Creation 2867.1 lps (30.0 s, 2 samples)
Shell Scripts (1 concurrent) 3109.1 lpm (60.0 s, 2 samples)
Shell Scripts (8 concurrent) 609.0 lpm (60.1 s, 2 samples)
System Call Overhead 366198.2 lps (10.0 s, 7 samples)
System Benchmarks Index Values BASELINE RESULT INDEX
Dhrystone 2 using register variables 116700.0 28207297.1 2417.1
Double-Precision Whetstone 55.0 1974.3 359.0
Execl Throughput 43.0 1391.8 323.7
File Copy 1024 bufsize 2000 maxblocks 3960.0 306464.8 773.9
File Copy 256 bufsize 500 maxblocks 1655.0 79316.2 479.3
File Copy 4096 bufsize 8000 maxblocks 5800.0 827535.5 1426.8
Pipe Throughput 12440.0 423569.9 340.5
Pipe-based Context Switching 4000.0 52534.3 131.3
Process Creation 126.0 2867.1 227.5
Shell Scripts (1 concurrent) 42.4 3109.1 733.3
Shell Scripts (8 concurrent) 6.0 609.0 1014.9
System Call Overhead 15000.0 366198.2 244.1
========
System Benchmarks Index Score 504.9
------------------------------------------------------------------------
Benchmark Run: Fri Sep 27 2013 09:28:42 - 09:56:35
2 CPUs in system; running 2 parallel copies of tests
Dhrystone 2 using register variables 56253029.0 lps (10.0 s, 7 samples)
Double-Precision Whetstone 3945.4 MWIPS (9.9 s, 7 samples)
Execl Throughput 2794.3 lps (29.9 s, 2 samples)
File Copy 1024 bufsize 2000 maxblocks 571035.5 KBps (30.0 s, 2 samples)
File Copy 256 bufsize 500 maxblocks 149270.8 KBps (30.0 s, 2 samples)
File Copy 4096 bufsize 8000 maxblocks 1572210.9 KBps (30.0 s, 2 samples)
Pipe Throughput 848740.1 lps (10.0 s, 7 samples)
Pipe-based Context Switching 150493.1 lps (10.0 s, 7 samples)
Process Creation 5465.5 lps (30.0 s, 2 samples)
Shell Scripts (1 concurrent) 4553.8 lpm (60.0 s, 2 samples)
Shell Scripts (8 concurrent) 614.5 lpm (60.1 s, 2 samples)
System Call Overhead 713330.9 lps (10.0 s, 7 samples)
System Benchmarks Index Values BASELINE RESULT INDEX
Dhrystone 2 using register variables 116700.0 56253029.0 4820.3
Double-Precision Whetstone 55.0 3945.4 717.3
Execl Throughput 43.0 2794.3 649.8
File Copy 1024 bufsize 2000 maxblocks 3960.0 571035.5 1442.0
File Copy 256 bufsize 500 maxblocks 1655.0 149270.8 901.9
File Copy 4096 bufsize 8000 maxblocks 5800.0 1572210.9 2710.7
Pipe Throughput 12440.0 848740.1 682.3
Pipe-based Context Switching 4000.0 150493.1 376.2
Process Creation 126.0 5465.5 433.8
Shell Scripts (1 concurrent) 42.4 4553.8 1074.0
Shell Scripts (8 concurrent) 6.0 614.5 1024.2
System Call Overhead 15000.0 713330.9 475.6
========
System Benchmarks Index Score 937.6
おまけ: m1.small (1ECU / 1vCPU) @ US-east
========================================================================
BYTE UNIX Benchmarks (Version 5.1.3)
System: ip-10-72-190-126: GNU/Linux
OS: GNU/Linux -- 3.4.43-43.43.amzn1.x86_64 -- #1 SMP Mon May 6 18:04:41 UTC 2013
Machine: x86_64 (x86_64)
Language: en_US.utf8 (charmap="UTF-8", collate="UTF-8")
CPU 0: Intel(R) Xeon(R) CPU E5507 @ 2.27GHz (4533.5 bogomips)
Hyper-Threading, x86-64, MMX, Physical Address Ext, SYSENTER/SYSEXIT, SYSCALL/SYSRET
06:25:47 up 44 min, 1 user, load average: 0.32, 0.09, 0.07; runlevel 3
------------------------------------------------------------------------
Benchmark Run: Fri Sep 27 2013 06:25:47 - 06:54:21
1 CPU in system; running 1 parallel copy of tests
Dhrystone 2 using register variables 9467107.1 lps (10.0 s, 7 samples)
Double-Precision Whetstone 676.1 MWIPS (10.1 s, 7 samples)
Execl Throughput 459.9 lps (29.8 s, 2 samples)
File Copy 1024 bufsize 2000 maxblocks 88355.5 KBps (30.0 s, 2 samples)
File Copy 256 bufsize 500 maxblocks 28947.0 KBps (30.0 s, 2 samples)
File Copy 4096 bufsize 8000 maxblocks 294647.8 KBps (30.1 s, 2 samples)
Pipe Throughput 154822.6 lps (10.0 s, 7 samples)
Pipe-based Context Switching 29339.0 lps (10.0 s, 7 samples)
Process Creation 1040.3 lps (30.0 s, 2 samples)
Shell Scripts (1 concurrent) 817.1 lpm (60.1 s, 2 samples)
Shell Scripts (8 concurrent) 110.4 lpm (60.3 s, 2 samples)
System Call Overhead 132007.9 lps (10.0 s, 7 samples)
System Benchmarks Index Values BASELINE RESULT INDEX
Dhrystone 2 using register variables 116700.0 9467107.1 811.2
Double-Precision Whetstone 55.0 676.1 122.9
Execl Throughput 43.0 459.9 107.0
File Copy 1024 bufsize 2000 maxblocks 3960.0 88355.5 223.1
File Copy 256 bufsize 500 maxblocks 1655.0 28947.0 174.9
File Copy 4096 bufsize 8000 maxblocks 5800.0 294647.8 508.0
Pipe Throughput 12440.0 154822.6 124.5
Pipe-based Context Switching 4000.0 29339.0 73.3
Process Creation 126.0 1040.3 82.6
Shell Scripts (1 concurrent) 42.4 817.1 192.7
Shell Scripts (8 concurrent) 6.0 110.4 184.0
System Call Overhead 15000.0 132007.9 88.0
========
System Benchmarks Index Score 167.8========================================================================
BYTE UNIX Benchmarks (Version 5.1.3)
System: ip-10-191-22-9: GNU/Linux
OS: GNU/Linux -- 3.4.43-43.43.amzn1.x86_64 -- #1 SMP Mon May 6 18:04:41 UTC 2013
Machine: x86_64 (x86_64)
Language: en_US.utf8 (charmap="UTF-8", collate="UTF-8")
CPU 0: Intel(R) Xeon(R) CPU E5645 @ 2.40GHz (4000.1 bogomips)
Hyper-Threading, x86-64, MMX, Physical Address Ext, SYSENTER/SYSEXIT, SYSCALL/SYSRET
06:21:55 up 40 min, 1 user, load average: 0.19, 0.06, 0.06; runlevel 3
------------------------------------------------------------------------
Benchmark Run: Fri Sep 27 2013 06:21:55 - 06:50:07
1 CPU in system; running 1 parallel copy of tests
Dhrystone 2 using register variables 10534867.1 lps (10.0 s, 7 samples)
Double-Precision Whetstone 743.3 MWIPS (10.0 s, 7 samples)
Execl Throughput 571.4 lps (29.9 s, 2 samples)
File Copy 1024 bufsize 2000 maxblocks 117143.0 KBps (30.0 s, 2 samples)
File Copy 256 bufsize 500 maxblocks 30216.8 KBps (30.0 s, 2 samples)
File Copy 4096 bufsize 8000 maxblocks 358094.2 KBps (30.0 s, 2 samples)
Pipe Throughput 160246.4 lps (10.0 s, 7 samples)
Pipe-based Context Switching 29581.2 lps (10.0 s, 7 samples)
Process Creation 1113.2 lps (30.0 s, 2 samples)
Shell Scripts (1 concurrent) 795.5 lpm (60.1 s, 2 samples)
Shell Scripts (8 concurrent) 116.8 lpm (60.1 s, 2 samples)
System Call Overhead 123195.9 lps (10.0 s, 7 samples)
System Benchmarks Index Values BASELINE RESULT INDEX
Dhrystone 2 using register variables 116700.0 10534867.1 902.7
Double-Precision Whetstone 55.0 743.3 135.1
Execl Throughput 43.0 571.4 132.9
File Copy 1024 bufsize 2000 maxblocks 3960.0 117143.0 295.8
File Copy 256 bufsize 500 maxblocks 1655.0 30216.8 182.6
File Copy 4096 bufsize 8000 maxblocks 5800.0 358094.2 617.4
Pipe Throughput 12440.0 160246.4 128.8
Pipe-based Context Switching 4000.0 29581.2 74.0
Process Creation 126.0 1113.2 88.4
Shell Scripts (1 concurrent) 42.4 795.5 187.6
Shell Scripts (8 concurrent) 6.0 116.8 194.7
System Call Overhead 15000.0 123195.9 82.1
========
System Benchmarks Index Score 182.5========================================================================
BYTE UNIX Benchmarks (Version 5.1.3)
System: domU-12-31-39-13-E2-B5: GNU/Linux
OS: GNU/Linux -- 3.4.43-43.43.amzn1.x86_64 -- #1 SMP Mon May 6 18:04:41 UTC 2013
Machine: x86_64 (x86_64)
Language: en_US.utf8 (charmap="UTF-8", collate="UTF-8")
CPU 0: Intel(R) Xeon(R) CPU E5430 @ 2.66GHz (5320.0 bogomips)
Hyper-Threading, x86-64, MMX, Physical Address Ext, SYSENTER/SYSEXIT, SYSCALL/SYSRET
06:26:40 up 45 min, 1 user, load average: 0.38, 0.11, 0.08; runlevel 3
------------------------------------------------------------------------
Benchmark Run: Fri Sep 27 2013 06:26:40 - 06:54:52
1 CPU in system; running 1 parallel copy of tests
Dhrystone 2 using register variables 9249588.4 lps (10.0 s, 7 samples)
Double-Precision Whetstone 601.1 MWIPS (10.0 s, 7 samples)
Execl Throughput 571.4 lps (29.6 s, 2 samples)
File Copy 1024 bufsize 2000 maxblocks 101319.9 KBps (30.0 s, 2 samples)
File Copy 256 bufsize 500 maxblocks 26590.5 KBps (30.1 s, 2 samples)
File Copy 4096 bufsize 8000 maxblocks 272539.7 KBps (30.0 s, 2 samples)
Pipe Throughput 170356.6 lps (10.0 s, 7 samples)
Pipe-based Context Switching 30221.2 lps (10.0 s, 7 samples)
Process Creation 1117.8 lps (30.0 s, 2 samples)
Shell Scripts (1 concurrent) 904.4 lpm (60.0 s, 2 samples)
Shell Scripts (8 concurrent) 121.0 lpm (60.3 s, 2 samples)
System Call Overhead 135041.5 lps (10.0 s, 7 samples)
System Benchmarks Index Values BASELINE RESULT INDEX
Dhrystone 2 using register variables 116700.0 9249588.4 792.6
Double-Precision Whetstone 55.0 601.1 109.3
Execl Throughput 43.0 571.4 132.9
File Copy 1024 bufsize 2000 maxblocks 3960.0 101319.9 255.9
File Copy 256 bufsize 500 maxblocks 1655.0 26590.5 160.7
File Copy 4096 bufsize 8000 maxblocks 5800.0 272539.7 469.9
Pipe Throughput 12440.0 170356.6 136.9
Pipe-based Context Switching 4000.0 30221.2 75.6
Process Creation 126.0 1117.8 88.7
Shell Scripts (1 concurrent) 42.4 904.4 213.3
Shell Scripts (8 concurrent) 6.0 121.0 201.6
System Call Overhead 15000.0 135041.5 90.0
========
System Benchmarks Index Score 174.4
