2012年4月4日水曜日

さくらのVPS(2G)でUnixBench(5.1.3)を実行してみた。結果:値段そのままなのに、大幅に性能アップ!

はじめに

というわけで、早速、さくらのVPS (2G)で、UnixBenchを実行してみました。
なお、1.5Gで評価した結果は、"さくらのVPSでUnixBench(5.1.3)を実施してみた - part2(結果編)"を参照して下さい。
========================================================================
   BYTE UNIX Benchmarks (Version 5.1.3)

   System: localhost.localdomain: GNU/Linux
   OS: GNU/Linux -- 3.3.0-8.fc16.x86_64 -- #1 SMP Thu Mar 29 18:37:19 UTC 2012
   Machine: x86_64 (x86_64)
   Language: en_US.utf8 (charmap="UTF-8", collate="UTF-8")
   CPU 0: Intel(R) Xeon(R) CPU E5645 (4800.2 bogomips)
          Hyper-Threading, x86-64, MMX, Physical Address Ext, SYSCALL/SYSRET
   CPU 1: Intel(R) Xeon(R) CPU E5645 (4800.2 bogomips)
          Hyper-Threading, x86-64, MMX, Physical Address Ext, SYSCALL/SYSRET
   CPU 2: Intel(R) Xeon(R) CPU E5645 (4800.2 bogomips)
          Hyper-Threading, x86-64, MMX, Physical Address Ext, SYSCALL/SYSRET
   20:22:38 up  2:57,  2 users,  load average: 0.53, 0.41, 0.21; runlevel 3

------------------------------------------------------------------------
Benchmark Run: Tue Apr 03 2012 20:22:38 - 20:49:59
3 CPUs in system; running 1 parallel copy of tests

Dhrystone 2 using register variables       24322317.4 lps   (10.0 s, 7 samples)
Double-Precision Whetstone                     2961.2 MWIPS (7.0 s, 7 samples)
Execl Throughput                               1451.8 lps   (30.0 s, 2 samples)
File Copy 1024 bufsize 2000 maxblocks        787751.5 KBps  (30.0 s, 2 samples)
File Copy 256 bufsize 500 maxblocks          223048.0 KBps  (30.0 s, 2 samples)
File Copy 4096 bufsize 8000 maxblocks       1639344.8 KBps  (30.0 s, 2 samples)
Pipe Throughput                             1385086.0 lps   (10.0 s, 7 samples)
Pipe-based Context Switching                 226423.8 lps   (10.0 s, 7 samples)
Process Creation                               3676.0 lps   (30.0 s, 2 samples)
Shell Scripts (1 concurrent)                   3725.1 lpm   (60.0 s, 2 samples)
Shell Scripts (8 concurrent)                   1646.1 lpm   (60.0 s, 2 samples)
System Call Overhead                        1578386.4 lps   (10.0 s, 7 samples)

System Benchmarks Index Values               BASELINE       RESULT    INDEX
Dhrystone 2 using register variables         116700.0   24322317.4   2084.2
Double-Precision Whetstone                       55.0       2961.2    538.4
Execl Throughput                                 43.0       1451.8    337.6
File Copy 1024 bufsize 2000 maxblocks          3960.0     787751.5   1989.3
File Copy 256 bufsize 500 maxblocks            1655.0     223048.0   1347.7
File Copy 4096 bufsize 8000 maxblocks          5800.0    1639344.8   2826.5
Pipe Throughput                               12440.0    1385086.0   1113.4
Pipe-based Context Switching                   4000.0     226423.8    566.1
Process Creation                                126.0       3676.0    291.7
Shell Scripts (1 concurrent)                     42.4       3725.1    878.6
Shell Scripts (8 concurrent)                      6.0       1646.1   2743.5
System Call Overhead                          15000.0    1578386.4   1052.3
                                                                   ========
System Benchmarks Index Score                                        1024.6

------------------------------------------------------------------------
Benchmark Run: Tue Apr 03 2012 20:49:59 - 21:17:27
3 CPUs in system; running 3 parallel copies of tests

Dhrystone 2 using register variables       72068041.9 lps   (10.0 s, 7 samples)
Double-Precision Whetstone                     8846.1 MWIPS (7.2 s, 7 samples)
Execl Throughput                               8582.7 lps   (30.0 s, 2 samples)
File Copy 1024 bufsize 2000 maxblocks       1071208.3 KBps  (30.0 s, 2 samples)
File Copy 256 bufsize 500 maxblocks          285035.8 KBps  (30.0 s, 2 samples)
File Copy 4096 bufsize 8000 maxblocks       2289930.0 KBps  (30.0 s, 2 samples)
Pipe Throughput                             4059296.6 lps   (10.0 s, 7 samples)
Pipe-based Context Switching                 654058.4 lps   (10.0 s, 7 samples)
Process Creation                              28607.3 lps   (30.0 s, 2 samples)
Shell Scripts (1 concurrent)                  14152.2 lpm   (60.0 s, 2 samples)
Shell Scripts (8 concurrent)                   1953.0 lpm   (60.0 s, 2 samples)
System Call Overhead                        4106210.8 lps   (10.0 s, 7 samples)

System Benchmarks Index Values               BASELINE       RESULT    INDEX
Dhrystone 2 using register variables         116700.0   72068041.9   6175.5
Double-Precision Whetstone                       55.0       8846.1   1608.4
Execl Throughput                                 43.0       8582.7   1996.0
File Copy 1024 bufsize 2000 maxblocks          3960.0    1071208.3   2705.1
File Copy 256 bufsize 500 maxblocks            1655.0     285035.8   1722.3
File Copy 4096 bufsize 8000 maxblocks          5800.0    2289930.0   3948.2
Pipe Throughput                               12440.0    4059296.6   3263.1
Pipe-based Context Switching                   4000.0     654058.4   1635.1
Process Creation                                126.0      28607.3   2270.4
Shell Scripts (1 concurrent)                     42.4      14152.2   3337.8
Shell Scripts (8 concurrent)                      6.0       1953.0   3254.9
System Call Overhead                          15000.0    4106210.8   2737.5
                                                                   ========
System Benchmarks Index Score                                        2670.2

まとめ

テスト内容旧システム(1.5G)新システム(2G)
1 parallel copy of tests731.71024.6
2or3 parallel copy of tests1256.52670.2
CPUCore2 Duo T7700@2.4Ghz(2666.8MHz)Xeon E5645(2400.0MHz)
パラレルテストでは、1.5Gと比較して、2倍以上の結果がでており、CPUもパワーアップしていて、
さらに、HDD容量2倍以上(200GB)、メモリ512MBアップで、お値段そのままなので、乗り換える価値大と思います。

ただし、まだ、サービス開始直後なので、今後、他のVMの使用状況によっては、
この結果が維持されるとは限らないので、注意が必要と思います。

0 件のコメント:

コメントを投稿