Raspberry Pi 4 Model B をやっとセットアップできたので、とりあえず UnixBench でベンチマークをしてみた。
UnixBench のダウンロードから実行までの流れ
UnixBench をダウンロードして、
$ curl -O https://storage.googleapis.com/google-code-archive-downloads/v2/code.google.com/byte-unixbench/UnixBench5.1.3.tgz展開して、
$ tar zxvf UnixBench5.1.3.tgzディレクトリを移動して、
$ cd UnixBench/実行する。
$ ./RunUnixBench の実行結果
UnixBench の実行結果は下記の通り。とりあえず、 System Benchmarks Index Score だけを抜粋。（ UnixBench の全部の結果は最後に貼り付ける。）
System Benchmarks Index Score 863.3Raspberry Pi 4 Model B は CPU が結構、発熱するらしいので、 UnixBench 中に CPU の温度が上がるかを、下記のような簡単な Ruby スクリプトを実行しながら UnixBench を動かして確認してみた。（ちなみに、 Raspbian v10 には ruby は入っていなかったので、 rbenv で ruby 2.7.0 をインストールした。）
$ vi check_cpu_temp.rb
# ctl + c などで止めない限り動き続ける
while true do
# get_throttled -> 主に CPU（？）が電力不足になっていないかの確認
puts "throttled: #{`vcgencmd get_throttled`}"
# measure_temp -> CPU の温度
puts "measure_temp: #{`vcgencmd measure_temp`}"
# measure_volts -> CPU の動作電圧
puts "measure_volts: #{`vcgencmd measure_volts`}"
# measure_clock -> CPU の動作周波数
puts "measure_clock arm: #{`vcgencmd measure_clock arm`}"
sleep 1
end実行方法は下記の通り。
$ ruby check_cpu_temp.rb結果は、 UnixBench を動かす前の、特に何もしていないときの CPU の温度が「57.0℃」で、一番温度が高くなるときが「83.0℃」だった。
ちなみに、ヒートシンクをつける前とつけたあとで CPU の温度に違いが出るか確認したけど、 CPU の最高温度はヒートシンクをつけてもつけなくても、変わらず「83.0℃」だった。
あと、ヒートシンクをつけてもつけなくても、 UnixBench の結果はほとんど変わらなかった。 System Benchmarks Index Score だけを見ると、ヒートシンクをつけない場合が 863.3 で、ヒートシンクをつけた場合が 849.9 だった。 （微妙につけない場合の方がスコアがいいけど、誤差の範囲だと思う。）
以下、 CPU 温度の計測の出力結果を参考に貼り付けておく。
CPU の温度が低いとき
throttled: throttled=0x20000
measure_temp: temp=57.0'C
measure_volts: volt=0.8595V
measure_clock arm: frequency(48)=600169920CPU の温度が高いとき
throttled: throttled=0x20002
measure_temp: temp=83.0'C
measure_volts: volt=0.8595V
measure_clock arm: frequency(48)=1000265600ヒートシンクをつけた状態で、 CPU の温度が高いとき
throttled: throttled=0x20002
measure_temp: temp=83.0'C
measure_volts: volt=0.8595V
measure_clock arm: frequency(48)=1000265600UnixBench の実行結果（全部）
ヒートシンクをつけていない状態
========================================================================
BYTE UNIX Benchmarks (Version 5.1.3)
System: raspberrypi: GNU/Linux
OS: GNU/Linux -- 4.19.93-v7l+ -- #1290 SMP Fri Jan 10 16:45:11 GMT 2020
Machine: armv7l (unknown)
Language: en_US.utf8 (charmap="ANSI_X3.4-1968", collate="ANSI_X3.4-1968")
CPU 0: ARMv7 Processor rev 3 (v7l) (0.0 bogomips)
CPU 1: ARMv7 Processor rev 3 (v7l) (0.0 bogomips)
CPU 2: ARMv7 Processor rev 3 (v7l) (0.0 bogomips)
CPU 3: ARMv7 Processor rev 3 (v7l) (0.0 bogomips)
11:49:05 up 0 min, 1 user, load average: 0.26, 0.06, 0.02; runlevel 5
------------------------------------------------------------------------
Benchmark Run: Wed Feb 05 2020 11:49:05 - 12:17:02
4 CPUs in system; running 1 parallel copy of tests
Dhrystone 2 using register variables 10124314.6 lps (10.0 s, 7 samples)
Double-Precision Whetstone 2393.4 MWIPS (9.6 s, 7 samples)
Execl Throughput 912.5 lps (29.7 s, 2 samples)
File Copy 1024 bufsize 2000 maxblocks 115490.9 KBps (30.0 s, 2 samples)
File Copy 256 bufsize 500 maxblocks 31533.0 KBps (30.0 s, 2 samples)
File Copy 4096 bufsize 8000 maxblocks 325053.8 KBps (30.0 s, 2 samples)
Pipe Throughput 160911.4 lps (10.0 s, 7 samples)
Pipe-based Context Switching 48996.4 lps (10.0 s, 7 samples)
Process Creation 1860.8 lps (30.0 s, 2 samples)
Shell Scripts (1 concurrent) 2704.0 lpm (60.0 s, 2 samples)
Shell Scripts (8 concurrent) 708.5 lpm (60.0 s, 2 samples)
System Call Overhead 492917.7 lps (10.0 s, 7 samples)
System Benchmarks Index Values BASELINE RESULT INDEX
Dhrystone 2 using register variables 116700.0 10124314.6 867.6
Double-Precision Whetstone 55.0 2393.4 435.2
Execl Throughput 43.0 912.5 212.2
File Copy 1024 bufsize 2000 maxblocks 3960.0 115490.9 291.6
File Copy 256 bufsize 500 maxblocks 1655.0 31533.0 190.5
File Copy 4096 bufsize 8000 maxblocks 5800.0 325053.8 560.4
Pipe Throughput 12440.0 160911.4 129.4
Pipe-based Context Switching 4000.0 48996.4 122.5
Process Creation 126.0 1860.8 147.7
Shell Scripts (1 concurrent) 42.4 2704.0 637.7
Shell Scripts (8 concurrent) 6.0 708.5 1180.8
System Call Overhead 15000.0 492917.7 328.6
========
System Benchmarks Index Score 326.1
------------------------------------------------------------------------
Benchmark Run: Wed Feb 05 2020 12:17:02 - 12:45:05
4 CPUs in system; running 4 parallel copies of tests
Dhrystone 2 using register variables 40271820.3 lps (10.0 s, 7 samples)
Double-Precision Whetstone 9603.7 MWIPS (9.6 s, 7 samples)
Execl Throughput 2756.2 lps (29.9 s, 2 samples)
File Copy 1024 bufsize 2000 maxblocks 218366.8 KBps (30.0 s, 2 samples)
File Copy 256 bufsize 500 maxblocks 59831.0 KBps (30.0 s, 2 samples)
File Copy 4096 bufsize 8000 maxblocks 609570.0 KBps (30.0 s, 2 samples)
Pipe Throughput 616968.5 lps (10.0 s, 7 samples)
Pipe-based Context Switching 197090.1 lps (10.0 s, 7 samples)
Process Creation 4722.9 lps (30.0 s, 2 samples)
Shell Scripts (1 concurrent) 5962.8 lpm (60.0 s, 2 samples)
Shell Scripts (8 concurrent) 792.7 lpm (60.2 s, 2 samples)
System Call Overhead 1867015.5 lps (10.0 s, 7 samples)
System Benchmarks Index Values BASELINE RESULT INDEX
Dhrystone 2 using register variables 116700.0 40271820.3 3450.9
Double-Precision Whetstone 55.0 9603.7 1746.1
Execl Throughput 43.0 2756.2 641.0
File Copy 1024 bufsize 2000 maxblocks 3960.0 218366.8 551.4
File Copy 256 bufsize 500 maxblocks 1655.0 59831.0 361.5
File Copy 4096 bufsize 8000 maxblocks 5800.0 609570.0 1051.0
Pipe Throughput 12440.0 616968.5 496.0
Pipe-based Context Switching 4000.0 197090.1 492.7
Process Creation 126.0 4722.9 374.8
Shell Scripts (1 concurrent) 42.4 5962.8 1406.3
Shell Scripts (8 concurrent) 6.0 792.7 1321.1
System Call Overhead 15000.0 1867015.5 1244.7
========
System Benchmarks Index Score 863.3ヒートシンクをつけた状態
========================================================================
BYTE UNIX Benchmarks (Version 5.1.3)
System: raspberrypi: GNU/Linux
OS: GNU/Linux -- 4.19.93-v7l+ -- #1290 SMP Fri Jan 10 16:45:11 GMT 2020
Machine: armv7l (unknown)
Language: en_US.utf8 (charmap="ANSI_X3.4-1968", collate="ANSI_X3.4-1968")
CPU 0: ARMv7 Processor rev 3 (v7l) (0.0 bogomips)
CPU 1: ARMv7 Processor rev 3 (v7l) (0.0 bogomips)
CPU 2: ARMv7 Processor rev 3 (v7l) (0.0 bogomips)
CPU 3: ARMv7 Processor rev 3 (v7l) (0.0 bogomips)
11:50:43 up 14 min, 2 users, load average: 0.00, 0.00, 0.00; runlevel 5
------------------------------------------------------------------------
Benchmark Run: Mon Feb 03 2020 11:50:43 - 12:18:40
4 CPUs in system; running 1 parallel copy of tests
Dhrystone 2 using register variables 10207174.9 lps (10.0 s, 7 samples)
Double-Precision Whetstone 2393.4 MWIPS (9.6 s, 7 samples)
Execl Throughput 912.8 lps (30.0 s, 2 samples)
File Copy 1024 bufsize 2000 maxblocks 114212.8 KBps (30.0 s, 2 samples)
File Copy 256 bufsize 500 maxblocks 31633.8 KBps (30.0 s, 2 samples)
File Copy 4096 bufsize 8000 maxblocks 323503.6 KBps (30.0 s, 2 samples)
Pipe Throughput 159788.2 lps (10.0 s, 7 samples)
Pipe-based Context Switching 45280.8 lps (10.0 s, 7 samples)
Process Creation 1865.6 lps (30.0 s, 2 samples)
Shell Scripts (1 concurrent) 2671.1 lpm (60.0 s, 2 samples)
Shell Scripts (8 concurrent) 749.7 lpm (60.0 s, 2 samples)
System Call Overhead 493415.8 lps (10.0 s, 7 samples)
System Benchmarks Index Values BASELINE RESULT INDEX
Dhrystone 2 using register variables 116700.0 10207174.9 874.7
Double-Precision Whetstone 55.0 2393.4 435.2
Execl Throughput 43.0 912.8 212.3
File Copy 1024 bufsize 2000 maxblocks 3960.0 114212.8 288.4
File Copy 256 bufsize 500 maxblocks 1655.0 31633.8 191.1
File Copy 4096 bufsize 8000 maxblocks 5800.0 323503.6 557.8
Pipe Throughput 12440.0 159788.2 128.4
Pipe-based Context Switching 4000.0 45280.8 113.2
Process Creation 126.0 1865.6 148.1
Shell Scripts (1 concurrent) 42.4 2671.1 630.0
Shell Scripts (8 concurrent) 6.0 749.7 1249.5
System Call Overhead 15000.0 493415.8 328.9
========
System Benchmarks Index Score 324.9
------------------------------------------------------------------------
Benchmark Run: Mon Feb 03 2020 12:18:40 - 12:46:45
4 CPUs in system; running 4 parallel copies of tests
Dhrystone 2 using register variables 40809123.9 lps (10.0 s, 7 samples)
Double-Precision Whetstone 9579.5 MWIPS (9.7 s, 7 samples)
Execl Throughput 2734.0 lps (29.9 s, 2 samples)
File Copy 1024 bufsize 2000 maxblocks 218005.5 KBps (30.0 s, 2 samples)
File Copy 256 bufsize 500 maxblocks 59396.4 KBps (30.0 s, 2 samples)
File Copy 4096 bufsize 8000 maxblocks 596932.0 KBps (30.0 s, 2 samples)
Pipe Throughput 612995.6 lps (10.0 s, 7 samples)
Pipe-based Context Switching 185175.4 lps (10.0 s, 7 samples)
Process Creation 4742.7 lps (30.0 s, 2 samples)
Shell Scripts (1 concurrent) 5824.8 lpm (60.0 s, 2 samples)
Shell Scripts (8 concurrent) 738.0 lpm (60.2 s, 2 samples)
System Call Overhead 1864851.8 lps (10.0 s, 7 samples)
System Benchmarks Index Values BASELINE RESULT INDEX
Dhrystone 2 using register variables 116700.0 40809123.9 3496.9
Double-Precision Whetstone 55.0 9579.5 1741.7
Execl Throughput 43.0 2734.0 635.8
File Copy 1024 bufsize 2000 maxblocks 3960.0 218005.5 550.5
File Copy 256 bufsize 500 maxblocks 1655.0 59396.4 358.9
File Copy 4096 bufsize 8000 maxblocks 5800.0 596932.0 1029.2
Pipe Throughput 12440.0 612995.6 492.8
Pipe-based Context Switching 4000.0 185175.4 462.9
Process Creation 126.0 4742.7 376.4
Shell Scripts (1 concurrent) 42.4 5824.8 1373.8
Shell Scripts (8 concurrent) 6.0 738.0 1229.9
System Call Overhead 15000.0 1864851.8 1243.2
========
System Benchmarks Index Score 849.9以上、 Raspberry Pi 4 Model B で UnixBench でベンチマークをしてみた話でした。