Amazon EC2の新インスタンスタイプが来たのでUNIXBenchとってみた。使ったインスタンスはc3.large。2コアみえてるけど、性能特性から判断すると1コア2スレッドのように思われる。（シングルスレッド性能とマルチスレッド性能があまり違わない）最上位のインスタンスは32仮想コアとなっているが、Intel Xeon E5-2680 v2は1CPUあたり10コア20スレッド、2CPU構成で20コア40スレッドであるので、物理32コアは割り当てられない。c3.8xlargeはは16コア32スレッドが割り当てられていると考えるのが妥当だろう。と考えると、近い将来、20コア40スレッドを利用できるc3.10xlargeが登場するかもしれませんね。

========================================================================
BYTE UNIX Benchmarks (Version 5.1.3)

System: ip-10-166-132-61: GNU/Linux
OS: GNU/Linux — 3.4.68-59.97.amzn1.x86_64 — #1 SMP Tue Nov 5 07:40:09 UTC 2013
Machine: x86_64 (x86_64)
Language: en_US.utf8 (charmap=”UTF-8″, collate=”UTF-8″)
CPU 0: Intel(R) Xeon(R) CPU E5-2680 v2 @ 2.80GHz (5600.1 bogomips)
Hyper-Threading, x86-64, MMX, Physical Address Ext, SYSENTER/SYSEXIT, SYSCALL/SYSRET
CPU 1: Intel(R) Xeon(R) CPU E5-2680 v2 @ 2.80GHz (5600.1 bogomips)
Hyper-Threading, x86-64, MMX, Physical Address Ext, SYSENTER/SYSEXIT, SYSCALL/SYSRET
10:20:15 up 5 min, 1 user, load average: 0.07, 0.09, 0.05; runlevel 3

————————————————————————
Benchmark Run: Mon Nov 18 2013 10:20:15 - 10:48:04
2 CPUs in system; running 1 parallel copy of tests

Dhrystone 2 using register variables 36663117.7 lps (10.0 s, 7 samples)
Double-Precision Whetstone 2262.1 MWIPS (9.6 s, 7 samples)
Execl Throughput 1650.9 lps (30.0 s, 2 samples)
File Copy 1024 bufsize 2000 maxblocks 340189.8 KBps (30.0 s, 2 samples)
File Copy 256 bufsize 500 maxblocks 87830.5 KBps (30.0 s, 2 samples)
File Copy 4096 bufsize 8000 maxblocks 1220206.8 KBps (30.0 s, 2 samples)
Pipe Throughput 447887.5 lps (10.0 s, 7 samples)
Pipe-based Context Switching 49181.2 lps (10.0 s, 7 samples)
Process Creation 3370.8 lps (30.0 s, 2 samples)
Shell Scripts (1 concurrent) 3255.6 lpm (60.0 s, 2 samples)
Shell Scripts (8 concurrent) 508.6 lpm (60.1 s, 2 samples)
System Call Overhead 388469.3 lps (10.0 s, 7 samples)

System Benchmarks Index Values BASELINE RESULT INDEX
Dhrystone 2 using register variables 116700.0 36663117.7 3141.7
Double-Precision Whetstone 55.0 2262.1 411.3
Execl Throughput 43.0 1650.9 383.9
File Copy 1024 bufsize 2000 maxblocks 3960.0 340189.8 859.1
File Copy 256 bufsize 500 maxblocks 1655.0 87830.5 530.7
File Copy 4096 bufsize 8000 maxblocks 5800.0 1220206.8 2103.8
Pipe Throughput 12440.0 447887.5 360.0
Pipe-based Context Switching 4000.0 49181.2 123.0
Process Creation 126.0 3370.8 267.5
Shell Scripts (1 concurrent) 42.4 3255.6 767.8
Shell Scripts (8 concurrent) 6.0 508.6 847.6
System Call Overhead 15000.0 388469.3 259.0
========
System Benchmarks Index Score 559.9

————————————————————————
Benchmark Run: Mon Nov 18 2013 10:48:04 - 11:15:56
2 CPUs in system; running 2 parallel copies of tests

Dhrystone 2 using register variables 39899802.8 lps (10.0 s, 7 samples)
Double-Precision Whetstone 3939.3 MWIPS (9.9 s, 7 samples)
Execl Throughput 2274.5 lps (30.0 s, 2 samples)
File Copy 1024 bufsize 2000 maxblocks 506600.5 KBps (30.0 s, 2 samples)
File Copy 256 bufsize 500 maxblocks 130887.1 KBps (30.0 s, 2 samples)
File Copy 4096 bufsize 8000 maxblocks 1792458.3 KBps (30.0 s, 2 samples)
Pipe Throughput 706859.3 lps (10.0 s, 7 samples)
Pipe-based Context Switching 114695.3 lps (10.0 s, 7 samples)
Process Creation 4438.5 lps (30.0 s, 2 samples)
Shell Scripts (1 concurrent) 3729.4 lpm (60.0 s, 2 samples)
Shell Scripts (8 concurrent) 511.7 lpm (60.0 s, 2 samples)
System Call Overhead 596167.3 lps (10.0 s, 7 samples)

System Benchmarks Index Values BASELINE RESULT INDEX
Dhrystone 2 using register variables 116700.0 39899802.8 3419.0
Double-Precision Whetstone 55.0 3939.3 716.2
Execl Throughput 43.0 2274.5 529.0
File Copy 1024 bufsize 2000 maxblocks 3960.0 506600.5 1279.3
File Copy 256 bufsize 500 maxblocks 1655.0 130887.1 790.9
File Copy 4096 bufsize 8000 maxblocks 5800.0 1792458.3 3090.4
Pipe Throughput 12440.0 706859.3 568.2
Pipe-based Context Switching 4000.0 114695.3 286.7
Process Creation 126.0 4438.5 352.3
Shell Scripts (1 concurrent) 42.4 3729.4 879.6
Shell Scripts (8 concurrent) 6.0 511.7 852.9
System Call Overhead 15000.0 596167.3 397.4
========
System Benchmarks Index Score 800.6

メモ

c3.8xlargeのベンチマークはこちら
元RX-7乗りの適当な日々 新しいHPC向けAmazon EC2インスタンス”c3.8xlarge”が大セールしていたので簡単なベンチマークをとった