はじめに
Raspberry Pi 3 Model B+を入手して、はやくなったよーと話を聞いていたけど、どれくらい早くなっているのか気になったのでベンチマークを取ってみました。
実行環境
- OS : 2018-04-18-raspbian-stretch-lite
- microSD : microSDXC/SDHC Class 10 , 8GB
実行コマンド
SSH経由で実行
$ sudo apt update
$ sudo apt upgrade
$ sudo apt install git
$ git clone https://github.com/kdlucas/byte-unixbench
$ cd byte-unixbench/UnixBench
$ ./Run
Raspberry Pi 3 Model BとRaspberry Pi 3 Model B+、若干アーキティチャーが違うらしく、3BのSDカードを3B+で使わないでくださいというメモが同封されていたので、機械を変えるごとにSDカードのフォーマットとSDカードにイメージを書き込んでいます。
ベンチマーク
結果のサマリ。
| # | Type | Ampere | running 1 parallel copy of tests | running 4 parallel copies of test |
|---|---|---|---|---|
| 1. | 3B+ | 2.0A | 308.1 | 442.4 |
| 2. | 3B+ | 3.0A | 355.8 | 783.8 |
| 3. | 3B | 2.0A | 276.5 | 552.8 |
| 4. | 3B | 3.0A | 312.2 | 779.1 |
- 5V 3.0Aの電源を使用した場合、3B+が早い。
- 5V 2.0Aの電源を使用した場合のシングルスレッド、3B+が早い。
- 5V 2.0Aの電源を使用した場合のマルチスレッド、3Bが早い。
Raspberry Pi 3 Model B+ 【電源 5V, 2.0Aの場合】
電流値が足りなくて、コンソール(HDMI出力)で警告がたくさん出てました。
========================================================================
BYTE UNIX Benchmarks (Version 5.1.3)
System: raspberrypi: GNU/Linux
OS: GNU/Linux -- 4.14.34-v7+ -- #1110 SMP Mon Apr 16 15:18:51 BST 2018
Machine: armv7l (unknown)
Language: en_US.utf8 (charmap="ANSI_X3.4-1968", collate="ANSI_X3.4-1968")
CPU 0: ARMv7 Processor rev 4 (v7l) (0.0 bogomips)
CPU 1: ARMv7 Processor rev 4 (v7l) (0.0 bogomips)
CPU 2: ARMv7 Processor rev 4 (v7l) (0.0 bogomips)
CPU 3: ARMv7 Processor rev 4 (v7l) (0.0 bogomips)
10:13:22 up 5 min, 1 user, load average: 0.23, 0.35, 0.19; runlevel Apr
------------------------------------------------------------------------
Benchmark Run: Sun Jun 17 2018 10:13:22 - 10:41:30
4 CPUs in system; running 1 parallel copy of tests
Dhrystone 2 using register variables 5056831.9 lps (10.0 s, 7 samples)
Double-Precision Whetstone 1237.5 MWIPS (9.9 s, 7 samples)
Execl Throughput 1083.0 lps (29.6 s, 2 samples)
File Copy 1024 bufsize 2000 maxblocks 157288.9 KBpsg (30.0 s, 2 samples)
File Copy 256 bufsize 500 maxblocks 47232.4 KBps (30.0 s, 2 samples)
File Copy 4096 bufsize 8000 maxblocks 356708.5 KBps (30.0 s, 2 samples)
Pipe Throughput 343233.0 lps (10.0 s, 7 samples)
Pipe-based Context Switching 61349.4 lps (10.0 s, 7 samples)
Process Creation 1794.6 lps (30.0 s, 2 samples)
Shell Scripts (1 concurrent) 1198.4 lpm (60.0 s, 2 samples)
Shell Scripts (8 concurrent) 337.5 lpm (60.1 s, 2 samples)
System Call Overhead 670443.4 lps (10.0 s, 7 samples)
System Benchmarks Index Values BASELINE RESULT INDEX
Dhrystone 2 using register variables 116700.0 5056831.9 433.3
Double-Precision Whetstone 55.0 1237.5 225.0
Execl Throughput 43.0 1083.0 251.8
File Copy 1024 bufsize 2000 maxblocks 3960.0 157288.9 397.2
File Copy 256 bufsize 500 maxblocks 1655.0 47232.4 285.4
File Copy 4096 bufsize 8000 maxblocks 5800.0 356708.5 615.0
Pipe Throughput 12440.0 343233.0 275.9
Pipe-based Context Switching 4000.0 61349.4 153.4
Process Creation 126.0 1794.6 142.4
Shell Scripts (1 concurrent) 42.4 1198.4 282.6
Shell Scripts (8 concurrent) 6.0 337.5 562.4
System Call Overhead 15000.0 670443.4 447.0
========
System Benchmarks Index Score 308.1
------------------------------------------------------------------------
Benchmark Run: Sun Jun 17 2018 10:41:30 - 11:09:42
4 CPUs in system; running 4 parallel copies of tests
Dhrystone 2 using register variables 8910258.5 lps (10.0 s, 7 samples)
Double-Precision Whetstone 2943.2 MWIPS (7.9 s, 7 samples)
Execl Throughput 1307.8 lps (29.8 s, 2 samples)
File Copy 1024 bufsize 2000 maxblocks 122553.6 KBps (30.0 s, 2 samples)
File Copy 256 bufsize 500 maxblocks 33132.5 KBps (30.0 s, 2 samples)
File Copy 4096 bufsize 8000 maxblocks 329416.8 KBps (30.0 s, 2 samples)
Pipe Throughput 605402.2 lps (10.0 s, 7 samples)
Pipe-based Context Switching 122263.8 lps (10.0 s, 7 samples)
Process Creation 3046.6 lps (30.0 s, 2 samples)
Shell Scripts (1 concurrent) 2664.8 lpm (60.1 s, 2 samples)
Shell Scripts (8 concurrent) 359.6 lpm (60.5 s, 2 samples)
System Call Overhead 1424579.5 lps (10.0 s, 7 samples)
System Benchmarks Index Values BASELINE RESULT INDEX
Dhrystone 2 using register variables 116700.0 8910258.5 763.5
Double-Precision Whetstone 55.0 2943.2 535.1
Execl Throughput 43.0 1307.8 304.1
File Copy 1024 bufsize 2000 maxblocks 3960.0 122553.6 309.5
File Copy 256 bufsize 500 maxblocks 1655.0 33132.5 200.2
File Copy 4096 bufsize 8000 maxblocks 5800.0 329416.8 568.0
Pipe Throughput 12440.0 605402.2 486.7
Pipe-based Context Switching 4000.0 122263.8 305.7
Process Creation 126.0 3046.6 241.8
Shell Scripts (1 concurrent) 42.4 2664.8 628.5
Shell Scripts (8 concurrent) 6.0 359.6 599.3
System Call Overhead 15000.0 1424579.5 949.7
========
System Benchmarks Index Score 442.4
Raspberry Pi 3 Model B+ 【電源 5V, 3.0Aの場合】
========================================================================
BYTE UNIX Benchmarks (Version 5.1.3)
System: raspberrypi: GNU/Linux
OS: GNU/Linux -- 4.14.34-v7+ -- #1110 SMP Mon Apr 16 15:18:51 BST 2018
Machine: armv7l (unknown)
Language: en_US.utf8 (charmap="ANSI_X3.4-1968", collate="ANSI_X3.4-1968")
CPU 0: ARMv7 Processor rev 4 (v7l) (0.0 bogomips)
CPU 1: ARMv7 Processor rev 4 (v7l) (0.0 bogomips)
CPU 2: ARMv7 Processor rev 4 (v7l) (0.0 bogomips)
CPU 3: ARMv7 Processor rev 4 (v7l) (0.0 bogomips)
10:47:36 up 2 min, 1 user, load average: 0.05, 0.05, 0.01; runlevel Jun
------------------------------------------------------------------------
Benchmark Run: Mon Jun 18 2018 10:47:36 - 11:15:41
4 CPUs in system; running 1 parallel copy of tests
Dhrystone 2 using register variables 5061681.5 lps (10.0 s, 7 samples)
Double-Precision Whetstone 1238.6 MWIPS (9.9 s, 7 samples)
Execl Throughput 1073.1 lps (29.9 s, 2 samples)
File Copy 1024 bufsize 2000 maxblocks 158305.2 KBps (30.0 s, 2 samples)
File Copy 256 bufsize 500 maxblocks 47425.0 KBps (30.0 s, 2 samples)
File Copy 4096 bufsize 8000 maxblocks 371694.3 KBps (30.0 s, 2 samples)
Pipe Throughput 342347.0 lps (10.0 s, 7 samples)
Pipe-based Context Switching 71096.9 lps (10.0 s, 7 samples)
Process Creation 2454.5 lps (30.0 s, 2 samples)
Shell Scripts (1 concurrent) 2137.7 lpm (60.0 s, 2 samples)
Shell Scripts (8 concurrent) 641.6 lpm (60.1 s, 2 samples)
System Call Overhead 672695.8 lps (10.0 s, 7 samples)
System Benchmarks Index Values BASELINE RESULT INDEX
Dhrystone 2 using register variables 116700.0 5061681.5 433.7
Double-Precision Whetstone 55.0 1238.6 225.2
Execl Throughput 43.0 1073.1 249.6
File Copy 1024 bufsize 2000 maxblocks 3960.0 158305.2 399.8
File Copy 256 bufsize 500 maxblocks 1655.0 47425.0 286.6
File Copy 4096 bufsize 8000 maxblocks 5800.0 371694.3 640.9
Pipe Throughput 12440.0 342347.0 275.2
Pipe-based Context Switching 4000.0 71096.9 177.7
Process Creation 126.0 2454.5 194.8
Shell Scripts (1 concurrent) 42.4 2137.7 504.2
Shell Scripts (8 concurrent) 6.0 641.6 1069.4
System Call Overhead 15000.0 672695.8 448.5
========
System Benchmarks Index Score 355.8
------------------------------------------------------------------------
Benchmark Run: Mon Jun 18 2018 11:15:41 - 11:44:06
4 CPUs in system; running 4 parallel copies of tests
Dhrystone 2 using register variables 17551748.2 lps (10.0 s, 7 samples)
Double-Precision Whetstone 4309.6 MWIPS (11.3 s, 7 samples)
Execl Throughput 2256.3 lps (29.9 s, 2 samples)
File Copy 1024 bufsize 2000 maxblocks 236474.4 KBps (30.0 s, 2 samples)
File Copy 256 bufsize 500 maxblocks 63637.9 KBps (30.0 s, 2 samples)
File Copy 4096 bufsize 8000 maxblocks 587143.3 KBps (30.0 s, 2 samples)
Pipe Throughput 1173153.0 lps (10.0 s, 7 samples)
Pipe-based Context Switching 239052.0 lps (10.0 s, 7 samples)
Process Creation 4902.9 lps (30.0 s, 2 samples)
Shell Scripts (1 concurrent) 4627.9 lpm (60.0 s, 2 samples)
Shell Scripts (8 concurrent) 634.2 lpm (60.2 s, 2 samples)
System Call Overhead 2218194.7 lps (10.0 s, 7 samples)
System Benchmarks Index Values BASELINE RESULT INDEX
Dhrystone 2 using register variables 116700.0 17551748.2 1504.0
Double-Precision Whetstone 55.0 4309.6 783.6
Execl Throughput 43.0 2256.3 524.7
File Copy 1024 bufsize 2000 maxblocks 3960.0 236474.4 597.2
File Copy 256 bufsize 500 maxblocks 1655.0 63637.9 384.5
File Copy 4096 bufsize 8000 maxblocks 5800.0 587143.3 1012.3
Pipe Throughput 12440.0 1173153.0 943.0
Pipe-based Context Switching 4000.0 239052.0 597.6
Process Creation 126.0 4902.9 389.1
Shell Scripts (1 concurrent) 42.4 4627.9 1091.5
Shell Scripts (8 concurrent) 6.0 634.2 1057.1
System Call Overhead 15000.0 2218194.7 1478.8
========
System Benchmarks Index Score 783.8
Raspberry Pi 3 Model B 【電源 5V, 2.0Aの場合】
電流値が足りなくて、コンソール(HDMI出力)で警告がたくさん出てました。
温度警告の表示も点灯してたり。
========================================================================
BYTE UNIX Benchmarks (Version 5.1.3)
System: raspberrypi: GNU/Linux
OS: GNU/Linux -- 4.14.34-v7+ -- #1110 SMP Mon Apr 16 15:18:51 BST 2018
Machine: armv7l (unknown)
Language: en_US.utf8 (charmap="ANSI_X3.4-1968", collate="ANSI_X3.4-1968")
CPU 0: ARMv7 Processor rev 4 (v7l) (0.0 bogomips)
CPU 1: ARMv7 Processor rev 4 (v7l) (0.0 bogomips)
CPU 2: ARMv7 Processor rev 4 (v7l) (0.0 bogomips)
CPU 3: ARMv7 Processor rev 4 (v7l) (0.0 bogomips)
13:11:22 up 0 min, 1 user, load average: 0.17, 0.06, 0.02; runlevel Jun
------------------------------------------------------------------------
Benchmark Run: Mon Jun 18 2018 13:11:22 - 13:39:30
4 CPUs in system; running 1 parallel copy of tests
Dhrystone 2 using register variables 4338436.2 lps (10.0 s, 7 samples)
Double-Precision Whetstone 1060.4 MWIPS (9.9 s, 7 samples)
Execl Throughput 951.5 lps (29.9 s, 2 samples)
File Copy 1024 bufsize 2000 maxblocks 136781.0 KBps (30.0 s, 2 samples)
File Copy 256 bufsize 500 maxblocks 40424.7 KBps (30.0 s, 2 samples)
File Copy 4096 bufsize 8000 maxblocks 343526.3 KBps (30.0 s, 2 samples)
Pipe Throughput 294266.3 lps (10.0 s, 7 samples)
Pipe-based Context Switching 60753.8 lps (10.0 s, 7 samples)
Process Creation 1410.2 lps (30.0 s, 2 samples)
Shell Scripts (1 concurrent) 1194.0 lpm (60.1 s, 2 samples)
Shell Scripts (8 concurrent) 345.1 lpm (60.2 s, 2 samples)
System Call Overhead 579737.8 lps (10.0 s, 7 samples)
System Benchmarks Index Values BASELINE RESULT INDEX
Dhrystone 2 using register variables 116700.0 4338436.2 371.8
Double-Precision Whetstone 55.0 1060.4 192.8
Execl Throughput 43.0 951.5 221.3
File Copy 1024 bufsize 2000 maxblocks 3960.0 136781.0 345.4
File Copy 256 bufsize 500 maxblocks 1655.0 40424.7 244.3
File Copy 4096 bufsize 8000 maxblocks 5800.0 343526.3 592.3
Pipe Throughput 12440.0 294266.3 236.5
Pipe-based Context Switching 4000.0 60753.8 151.9
Process Creation 126.0 1410.2 111.9
Shell Scripts (1 concurrent) 42.4 1194.0 281.6
Shell Scripts (8 concurrent) 6.0 345.1 575.1
System Call Overhead 15000.0 579737.8 386.5
========
System Benchmarks Index Score 276.5
------------------------------------------------------------------------
Benchmark Run: Mon Jun 18 2018 13:39:30 - 14:07:50
4 CPUs in system; running 4 parallel copies of tests
Dhrystone 2 using register variables 14122797.6 lps (10.0 s, 7 samples)
Double-Precision Whetstone 2550.9 MWIPS (8.2 s, 7 samples)
Execl Throughput 1346.9 lps (29.8 s, 2 samples)
File Copy 1024 bufsize 2000 maxblocks 210083.5 KBps (30.0 s, 2 samples)
File Copy 256 bufsize 500 maxblocks 59418.5 KBps (30.0 s, 2 samples)
File Copy 4096 bufsize 8000 maxblocks 449648.7 KBps (30.0 s, 2 samples)
Pipe Throughput 754827.0 lps (10.0 s, 7 samples)
Pipe-based Context Switching 226921.3 lps (10.0 s, 7 samples)
Process Creation 2911.5 lps (30.0 s, 2 samples)
Shell Scripts (1 concurrent) 2762.0 lpm (60.1 s, 2 samples)
Shell Scripts (8 concurrent) 431.9 lpm (60.4 s, 2 samples)
System Call Overhead 1261392.3 lps (10.0 s, 7 samples)
System Benchmarks Index Values BASELINE RESULT INDEX
Dhrystone 2 using register variables 116700.0 14122797.6 1210.2
Double-Precision Whetstone 55.0 2550.9 463.8
Execl Throughput 43.0 1346.9 313.2
File Copy 1024 bufsize 2000 maxblocks 3960.0 210083.5 530.5
File Copy 256 bufsize 500 maxblocks 1655.0 59418.5 359.0
File Copy 4096 bufsize 8000 maxblocks 5800.0 449648.7 775.3
Pipe Throughput 12440.0 754827.0 606.8
Pipe-based Context Switching 4000.0 226921.3 567.3
Process Creation 126.0 2911.5 231.1
Shell Scripts (1 concurrent) 42.4 2762.0 651.4
Shell Scripts (8 concurrent) 6.0 431.9 719.9
System Call Overhead 15000.0 1261392.3 840.9
========
System Benchmarks Index Score 552.8
Raspberry Pi 3 Model B 【電源 5V, 3.0Aの場合】
========================================================================
BYTE UNIX Benchmarks (Version 5.1.3)
System: raspberrypi: GNU/Linux
OS: GNU/Linux -- 4.14.34-v7+ -- #1110 SMP Mon Apr 16 15:18:51 BST 2018
Machine: armv7l (unknown)
Language: en_US.utf8 (charmap="ANSI_X3.4-1968", collate="ANSI_X3.4-1968")
CPU 0: ARMv7 Processor rev 4 (v7l) (0.0 bogomips)
CPU 1: ARMv7 Processor rev 4 (v7l) (0.0 bogomips)
CPU 2: ARMv7 Processor rev 4 (v7l) (0.0 bogomips)
CPU 3: ARMv7 Processor rev 4 (v7l) (0.0 bogomips)
12:09:22 up 1 min, 1 user, load average: 0.25, 0.07, 0.02; runlevel Jun
------------------------------------------------------------------------
Benchmark Run: Mon Jun 18 2018 12:09:22 - 12:37:32
4 CPUs in system; running 1 parallel copy of tests
Dhrystone 2 using register variables 4340520.8 lps (10.0 s, 7 samples)
Double-Precision Whetstone 1061.4 MWIPS (9.9 s, 7 samples)
Execl Throughput 954.9 lps (29.9 s, 2 samples)
File Copy 1024 bufsize 2000 maxblocks 138503.9 KBps (30.0 s, 2 samples)
File Copy 256 bufsize 500 maxblocks 40566.5 KBps (30.0 s, 2 samples)
File Copy 4096 bufsize 8000 maxblocks 345448.5 KBps (30.0 s, 2 samples)
Pipe Throughput 294819.0 lps (10.0 s, 7 samples)
Pipe-based Context Switching 61330.5 lps (10.0 s, 7 samples)
Process Creation 2070.0 lps (30.0 s, 2 samples)
Shell Scripts (1 concurrent) 1926.8 lpm (60.0 s, 2 samples)
Shell Scripts (8 concurrent) 604.0 lpm (60.1 s, 2 samples)
System Call Overhead 578260.6 lps (10.0 s, 7 samples)
System Benchmarks Index Values BASELINE RESULT INDEX
Dhrystone 2 using register variables 116700.0 4340520.8 371.9
Double-Precision Whetstone 55.0 1061.4 193.0
Execl Throughput 43.0 954.9 222.1
File Copy 1024 bufsize 2000 maxblocks 3960.0 138503.9 349.8
File Copy 256 bufsize 500 maxblocks 1655.0 40566.5 245.1
File Copy 4096 bufsize 8000 maxblocks 5800.0 345448.5 595.6
Pipe Throughput 12440.0 294819.0 237.0
Pipe-based Context Switching 4000.0 61330.5 153.3
Process Creation 126.0 2070.0 164.3
Shell Scripts (1 concurrent) 42.4 1926.8 454.4
Shell Scripts (8 concurrent) 6.0 604.0 1006.6
System Call Overhead 15000.0 578260.6 385.5
========
System Benchmarks Index Score 312.2
------------------------------------------------------------------------
Benchmark Run: Mon Jun 18 2018 12:37:32 - 13:05:46
4 CPUs in system; running 4 parallel copies of tests
Dhrystone 2 using register variables 17247074.6 lps (10.0 s, 7 samples)
Double-Precision Whetstone 4234.9 MWIPS (9.9 s, 7 samples)
Execl Throughput 2266.3 lps (29.9 s, 2 samples)
File Copy 1024 bufsize 2000 maxblocks 233294.6 KBps (30.0 s, 2 samples)
File Copy 256 bufsize 500 maxblocks 63137.4 KBps (30.0 s, 2 samples)
File Copy 4096 bufsize 8000 maxblocks 543352.4 KBps (30.0 s, 2 samples)
Pipe Throughput 1178818.7 lps (10.0 s, 7 samples)
Pipe-based Context Switching 242635.0 lps (10.0 s, 7 samples)
Process Creation 4907.9 lps (30.0 s, 2 samples)
Shell Scripts (1 concurrent) 4698.9 lpm (60.1 s, 2 samples)
Shell Scripts (8 concurrent) 646.7 lpm (60.2 s, 2 samples)
System Call Overhead 2223012.4 lps (10.0 s, 7 samples)
System Benchmarks Index Values BASELINE RESULT INDEX
Dhrystone 2 using register variables 116700.0 17247074.6 1477.9
Double-Precision Whetstone 55.0 4234.9 770.0
Execl Throughput 43.0 2266.3 527.0
File Copy 1024 bufsize 2000 maxblocks 3960.0 233294.6 589.1
File Copy 256 bufsize 500 maxblocks 1655.0 63137.4 381.5
File Copy 4096 bufsize 8000 maxblocks 5800.0 543352.4 936.8
Pipe Throughput 12440.0 1178818.7 947.6
Pipe-based Context Switching 4000.0 242635.0 606.6
Process Creation 126.0 4907.9 389.5
Shell Scripts (1 concurrent) 42.4 4698.9 1108.2
Shell Scripts (8 concurrent) 6.0 646.7 1077.9
System Call Overhead 15000.0 2223012.4 1482.0
========
System Benchmarks Index Score 779.1
最後に
不思議なスコアになってしまったのが気になる。