そういえばNanoPi NEO/NEO2をUnixBenchで測ったけど、普通のPCだとどれくらいだろうと思ったので見てみた。
Pentium N3700のPC
まずは常用している端末。これはPentium N3700 1.6GHzというCPUを搭載しているPC。以前にルーター用として購入したASRock N3700-ITXが退役したお下がり。人に言わすと「何？その苦行」レベルの性能だけど、ブラウザでネット見てメールしてエディタで何か書く程度だとほとんど不満なく使えてんるだけどねぇ。
========================================================================
BYTE UNIX Benchmarks (Version 5.1.3)
System: linux-uy1c: GNU/Linux
OS: GNU/Linux -- 4.10.13-1-default -- #1 SMP PREEMPT Thu Apr 27 12:23:31 UTC 2017 (e5d11ce)
Machine: x86_64 (x86_64)
Language: en_US.utf8 (charmap="UTF-8", collate="UTF-8")
CPU 0: Intel(R) Pentium(R) CPU N3700 @ 1.60GHz (3200.0 bogomips)
Hyper-Threading, x86-64, MMX, Physical Address Ext, SYSENTER/SYSEXIT, SYSCALL/SYSRET, Intel virtualization
CPU 1: Intel(R) Pentium(R) CPU N3700 @ 1.60GHz (3201.8 bogomips)
Hyper-Threading, x86-64, MMX, Physical Address Ext, SYSENTER/SYSEXIT, SYSCALL/SYSRET, Intel virtualization
CPU 2: Intel(R) Pentium(R) CPU N3700 @ 1.60GHz (3202.0 bogomips)
Hyper-Threading, x86-64, MMX, Physical Address Ext, SYSENTER/SYSEXIT, SYSCALL/SYSRET, Intel virtualization
CPU 3: Intel(R) Pentium(R) CPU N3700 @ 1.60GHz (3202.0 bogomips)
Hyper-Threading, x86-64, MMX, Physical Address Ext, SYSENTER/SYSEXIT, SYSCALL/SYSRET, Intel virtualization
09:28am up 0:14, 1 user, load average: 0.66, 0.41, 0.34; runlevel 5
------------------------------------------------------------------------
Benchmark Run: Wed May 10 2017 09:28:30 - 09:56:46
4 CPUs in system; running 1 parallel copy of tests
Dhrystone 2 using register variables 12634825.9 lps (10.0 s, 7 samples)
Double-Precision Whetstone 2041.6 MWIPS (10.0 s, 7 samples)
Execl Throughput 2246.9 lps (29.9 s, 2 samples)
File Copy 1024 bufsize 2000 maxblocks 288839.9 KBps (30.0 s, 2 samples)
File Copy 256 bufsize 500 maxblocks 82246.9 KBps (30.0 s, 2 samples)
File Copy 4096 bufsize 8000 maxblocks 693385.3 KBps (30.0 s, 2 samples)
Pipe Throughput 1084798.9 lps (10.0 s, 7 samples)
Pipe-based Context Switching 142273.5 lps (10.0 s, 7 samples)
Process Creation 1282.9 lps (30.0 s, 2 samples)
Shell Scripts (1 concurrent) 1570.3 lpm (60.0 s, 2 samples)
Shell Scripts (8 concurrent) 681.3 lpm (60.1 s, 2 samples)
System Call Overhead 1465447.9 lps (10.0 s, 7 samples)
System Benchmarks Index Values BASELINE RESULT INDEX
Dhrystone 2 using register variables 116700.0 12634825.9 1082.7
Double-Precision Whetstone 55.0 2041.6 371.2
Execl Throughput 43.0 2246.9 522.5
File Copy 1024 bufsize 2000 maxblocks 3960.0 288839.9 729.4
File Copy 256 bufsize 500 maxblocks 1655.0 82246.9 497.0
File Copy 4096 bufsize 8000 maxblocks 5800.0 693385.3 1195.5
Pipe Throughput 12440.0 1084798.9 872.0
Pipe-based Context Switching 4000.0 142273.5 355.7
Process Creation 126.0 1282.9 101.8
Shell Scripts (1 concurrent) 42.4 1570.3 370.3
Shell Scripts (8 concurrent) 6.0 681.3 1135.5
System Call Overhead 15000.0 1465447.9 977.0
========
System Benchmarks Index Score 570.2
------------------------------------------------------------------------
Benchmark Run: Wed May 10 2017 09:56:46 - 10:25:27
4 CPUs in system; running 4 parallel copies of tests
Dhrystone 2 using register variables 47027951.1 lps (10.0 s, 7 samples)
Double-Precision Whetstone 8166.1 MWIPS (10.0 s, 7 samples)
Execl Throughput 4708.4 lps (29.8 s, 2 samples)
File Copy 1024 bufsize 2000 maxblocks 378624.0 KBps (30.0 s, 2 samples)
File Copy 256 bufsize 500 maxblocks 112926.8 KBps (30.0 s, 2 samples)
File Copy 4096 bufsize 8000 maxblocks 844039.4 KBps (30.0 s, 2 samples)
Pipe Throughput 4188091.1 lps (10.0 s, 7 samples)
Pipe-based Context Switching 642523.3 lps (10.0 s, 7 samples)
Process Creation 10798.0 lps (30.0 s, 2 samples)
Shell Scripts (1 concurrent) 5015.7 lpm (60.0 s, 2 samples)
Shell Scripts (8 concurrent) 804.5 lpm (60.1 s, 2 samples)
System Call Overhead 4137375.5 lps (10.0 s, 7 samples)
System Benchmarks Index Values BASELINE RESULT INDEX
Dhrystone 2 using register variables 116700.0 47027951.1 4029.8
Double-Precision Whetstone 55.0 8166.1 1484.7
Execl Throughput 43.0 4708.4 1095.0
File Copy 1024 bufsize 2000 maxblocks 3960.0 378624.0 956.1
File Copy 256 bufsize 500 maxblocks 1655.0 112926.8 682.3
File Copy 4096 bufsize 8000 maxblocks 5800.0 844039.4 1455.2
Pipe Throughput 12440.0 4188091.1 3366.6
Pipe-based Context Switching 4000.0 642523.3 1606.3
Process Creation 126.0 10798.0 857.0
Shell Scripts (1 concurrent) 42.4 5015.7 1182.9
Shell Scripts (8 concurrent) 6.0 804.5 1340.8
System Call Overhead 15000.0 4137375.5 2758.3
========
System Benchmarks Index Score 1496.5
NanoPi NEO2のインデックススコア(前回)がシングルで298.9、4パラレルで771.2だったので単純にその数値で比較するとどちらも2倍に僅かに届かない程度。つまりオイラのPCはNanoPi NEO2の2倍速いんだぜってこと。
NanoPi NEO(無印)と比べてみた。
NanoPi NEOのインデックススコア(前回)がシングルで164.1パラレルで429.0だったので単純にその数値で比較するとどちらも共に3.48倍。つまりオイラのPCはNanoPi NEOの3.5倍も速いんだぜYeah!ってこと。
Core i3 2120TのPC
殆ど仕事らしい仕事をしていないサーバでも試してみた。
CPUはIntel Core i3-2120T 2.60GHz。これは、たしかSandy Bridge世代のi3の省電力版。要するにi3とか名乗っていてもドラクエの最初に出てくる敵のスライムなみのザコ。
========================================================================
BYTE UNIX Benchmarks (Version 5.1.3)
System: router.localnet: FreeBSD
OS: FreeBSD -- 11.0-RELEASE-p9 -- FreeBSD 11.0-RELEASE-p9 #0: Tue Apr 11 08:48:40 UTC 2017 root@amd64-builder.daemonology.net:/usr/obj/usr/src/sys/GENERIC
Machine: amd64 (GENERIC)
Language: en_US.utf8 (charmap=, collate=)
CPU: no details available
11:21AM up 11 mins, 1 users, load averages: 0.20, 0.27, 0.18; runlevel
------------------------------------------------------------------------
Benchmark Run: Wed May 10 2017 11:21:42 - 11:49:53
unknown CPUs in system; running 1 parallel copy of tests
Dhrystone 2 using register variables 26643080.7 lps (10.0 s, 7 samples)
Double-Precision Whetstone 4328.3 MWIPS (9.6 s, 7 samples)
Execl Throughput 2552.4 lps (29.9 s, 2 samples)
File Copy 1024 bufsize 2000 maxblocks 114162.2 KBps (30.0 s, 2 samples)
File Copy 256 bufsize 500 maxblocks 100555.9 KBps (30.0 s, 2 samples)
File Copy 4096 bufsize 8000 maxblocks 59423.0 KBps (30.0 s, 2 samples)
Pipe Throughput 1467491.2 lps (10.0 s, 7 samples)
Pipe-based Context Switching 210785.1 lps (10.0 s, 7 samples)
Process Creation 7921.7 lps (30.0 s, 2 samples)
Shell Scripts (1 concurrent) 7561.4 lpm (60.0 s, 2 samples)
Shell Scripts (8 concurrent) 1814.9 lpm (60.0 s, 2 samples)
System Call Overhead 1145558.2 lps (10.0 s, 7 samples)
System Benchmarks Index Values BASELINE RESULT INDEX
Dhrystone 2 using register variables 116700.0 26643080.7 2283.0
Double-Precision Whetstone 55.0 4328.3 787.0
Execl Throughput 43.0 2552.4 593.6
File Copy 1024 bufsize 2000 maxblocks 3960.0 114162.2 288.3
File Copy 256 bufsize 500 maxblocks 1655.0 100555.9 607.6
File Copy 4096 bufsize 8000 maxblocks 5800.0 59423.0 102.5
Pipe Throughput 12440.0 1467491.2 1179.7
Pipe-based Context Switching 4000.0 210785.1 527.0
Process Creation 126.0 7921.7 628.7
Shell Scripts (1 concurrent) 42.4 7561.4 1783.4
Shell Scripts (8 concurrent) 6.0 1814.9 3024.9
System Call Overhead 15000.0 1145558.2 763.7
========
System Benchmarks Index Score 748.3
------------------------------------------------------------------------
Benchmark Run: Wed May 10 2017 11:49:53 - 12:18:16
unknown CPUs in system; running 4 parallel copies of tests
Dhrystone 2 using register variables 59413922.1 lps (10.0 s, 7 samples)
Double-Precision Whetstone 14660.3 MWIPS (9.8 s, 7 samples)
Execl Throughput 6034.2 lps (29.9 s, 2 samples)
File Copy 1024 bufsize 2000 maxblocks 99115.6 KBps (30.0 s, 2 samples)
File Copy 256 bufsize 500 maxblocks 64172.5 KBps (30.1 s, 2 samples)
File Copy 4096 bufsize 8000 maxblocks 83750.5 KBps (30.1 s, 2 samples)
Pipe Throughput 3397970.8 lps (10.0 s, 7 samples)
Pipe-based Context Switching 773384.7 lps (10.0 s, 7 samples)
Process Creation 18785.9 lps (30.0 s, 2 samples)
Shell Scripts (1 concurrent) 14593.1 lpm (60.0 s, 2 samples)
Shell Scripts (8 concurrent) 1867.0 lpm (60.1 s, 2 samples)
System Call Overhead 2839176.8 lps (10.0 s, 7 samples)
System Benchmarks Index Values BASELINE RESULT INDEX
Dhrystone 2 using register variables 116700.0 59413922.1 5091.2
Double-Precision Whetstone 55.0 14660.3 2665.5
Execl Throughput 43.0 6034.2 1403.3
File Copy 1024 bufsize 2000 maxblocks 3960.0 99115.6 250.3
File Copy 256 bufsize 500 maxblocks 1655.0 64172.5 387.7
File Copy 4096 bufsize 8000 maxblocks 5800.0 83750.5 144.4
Pipe Throughput 12440.0 3397970.8 2731.5
Pipe-based Context Switching 4000.0 773384.7 1933.5
Process Creation 126.0 18785.9 1490.9
Shell Scripts (1 concurrent) 42.4 14593.1 3441.8
Shell Scripts (8 concurrent) 6.0 1867.0 3111.7
System Call Overhead 15000.0 2839176.8 1892.8
========
System Benchmarks Index Score 1367.0
全く期待してなかったけど本当に良くない。シングルでは748.3ということなのでN3700よりは流石にマシだけど、4パラレルだと1367ということでN3700の1496.5と比較して約9%低いスコアとなっている。
i3は2コア4スレッドなので4パラレルで良い筈。ちなみに2パラレルだとインデックススコアは1072.9だった。
もうちっとマシなCPUでも試したかったけど空いてなかったので今回はこれだけ。
っていうか俺のPCって15ドルのおもちゃの2倍の性能しかないのかよ・・・orz
関連記事:- NanoPi NEO2ベンチマーク (公式イメージで)
- NanoPi NEO2 最大クロック引き下げ後のUnixBench
- NanoPi NEO2ベンチマーク
- NanoPi NEOベンチマーク
- UnixBench 自分の常用端末でも試してみた ←いまここ