スピード測定のためBYTEBENCH取ったんだが@Raspberry Pi | cyber205の日記

cyber205の日記：スピード測定のためBYTEBENCH取ったんだが@Raspberry Pi 0

日記 by cyber205 2014年11月03日 21時26分

最初に書いておくと結果は３つアップした。
ノーマル時とオーバークロック時と、手元のマシンの参考データ。
Raspberry Piに装備しているのは32MBのクラス10なSDカード。

まずノーマル動作をしているRaspberry Pi の結果がこれ。

BYTE UNIX Benchmarks (Version 5.1.3)

System: raspberrypi: GNU/Linux OS: GNU/Linux -- 3.12.28+ -- #709 PREEMPT Mon Sep 8 15:28:00 BST 2014 Machine: armv6l (unknown) Language: en_US.utf8 (charmap="UTF-8", collate="UTF-8") CPU 0: ARMv6-compatible processor rev 7 (v6l) (0.0 bogomips)

08:30:01 up 3 min, 2 users, load average: 0.42, 0.35, 0.15; runlevel 2

------------------------------------------------------------------------ Benchmark Run: 月 11月 03 2014 08:30:01 - 08:58:15 1 CPU in system; running 1 parallel copy of tests

Dhrystone 2 using register variables________1650136.8 lps___(10.0 s, 7 samples) Double-Precision Whetstone______________________264.4 MWIPS (10.0 s, 7 samples) Execl Throughput________________________________214.1 lps___(29.9 s, 2 samples) File Copy 1024 bufsize 2000 maxblocks_________36122.1 KBps__(30.0 s, 2 samples) File Copy 256 bufsize 500 maxblocks___________11125.8 KBps__(30.0 s, 2 samples) File Copy 4096 bufsize 8000 maxblocks_________85486.5 KBps__(30.0 s, 2 samples) Pipe Throughput______________________________146460.7 lps___(10.0 s, 7 samples) Pipe-based Context Switching__________________18083.4 lps___(10.0 s, 7 samples) Process Creation________________________________673.2 lps___(30.0 s, 2 samples) Shell Scripts (1 concurrent)____________________375.2 lpm___(60.1 s, 2 samples) Shell Scripts (8 concurrent)_____________________47.4 lpm___(60.7 s, 2 samples) System Call Overhead_________________________270055.7 lps___(10.0 s, 7 samples)

System Benchmarks Index Values_______________BASELINE_______RESULT____INDEX Dhrystone 2 using register variables_________116700.0____1650136.8____141.4 Double-Precision Whetstone_______________________55.0________264.4_____48.1 Execl Throughput_________________________________43.0________214.1_____49.8 File Copy 1024 bufsize 2000 maxblocks__________3960.0______36122.1_____91.2 File Copy 256 bufsize 500 maxblocks____________1655.0______11125.8_____67.2 File Copy 4096 bufsize 8000 maxblocks__________5800.0______85486.5____147.4 Pipe Throughput_______________________________12440.0_____146460.7____117.7 Pipe-based Context Switching___________________4000.0______18083.4_____45.2 Process Creation________________________________126.0________673.2_____53.4 Shell Scripts (1 concurrent)_____________________42.4________375.2_____88.5 Shell Scripts (8 concurrent)______________________6.0_________47.4_____79.1 System Call Overhead__________________________15000.0_____270055.7____180.0 ___________________________________________________________________======== System Benchmarks Index Score__________________________________________83.2

------------------------------------------------------------------------ ------------------------------------------------------------------------ 多少はスコアが上がることもあるらしいが、だいたい総合スコアは80ぐらいかな。もう一度測定してみた。

今度はオーバークロックで限度いっぱいの1GHzに上げてみる。 Raspberry PiにはNone(700MHz)/Modeest(800MHz)/Medium(900MHz) High(950MHz)/Turbo(1GHz)とオーバークロックできるメニューがある。

BYTE UNIX Benchmarks (Version 5.1.3)

17:53:25 up 1 min,__1 user,__load average: 0.72, 0.29, 0.11; runlevel 2

------------------------------------------------------------------------ Benchmark Run: 月 11月 03 2014 17:53:25 - 18:21:28 1 CPU in system; running 1 parallel copy of tests

Dhrystone 2 using register variables________2364480.8 lps___(10.0 s, 7 samples) Double-Precision Whetstone______________________382.9 MWIPS (9.4 s, 7 samples) Execl Throughput________________________________353.9 lps___(29.9 s, 2 samples) File Copy 1024 bufsize 2000 maxblocks_________60669.0 KBps__(30.0 s, 2 samples) File Copy 256 bufsize 500 maxblocks___________18994.0 KBps__(30.0 s, 2 samples) File Copy 4096 bufsize 8000 maxblocks________145062.9 KBps__(30.0 s, 2 samples) Pipe Throughput______________________________237935.9 lps___(10.0 s, 7 samples) Pipe-based Context Switching__________________30581.6 lps___(10.0 s, 7 samples) Process Creation_______________________________1100.5 lps___(30.0 s, 2 samples) Shell Scripts (1 concurrent)____________________602.1 lpm___(60.1 s, 2 samples) Shell Scripts (8 concurrent)_____________________76.6 lpm___(60.3 s, 2 samples) System Call Overhead_________________________388572.1 lps___(10.0 s, 7 samples)

System Benchmarks Index Values_______________BASELINE_______RESULT____INDEX Dhrystone 2 using register variables_________116700.0____2364480.8____202.6 Double-Precision Whetstone_______________________55.0________382.9_____69.6 Execl Throughput_________________________________43.0________353.9_____82.3 File Copy 1024 bufsize 2000 maxblocks__________3960.0______60669.0____153.2 File Copy 256 bufsize 500 maxblocks____________1655.0______18994.0____114.8 File Copy 4096 bufsize 8000 maxblocks__________5800.0_____145062.9____250.1 Pipe Throughput_______________________________12440.0_____237935.9____191.3 Pipe-based Context Switching___________________4000.0______30581.6_____76.5 Process Creation________________________________126.0_______1100.5_____87.3 Shell Scripts (1 concurrent)_____________________42.4________602.1____142.0 Shell Scripts (8 concurrent)______________________6.0_________76.6____127.6 System Call Overhead__________________________15000.0_____388572.1____259.0 ___________________________________________________________________======== System Benchmarks Index Score_________________________________________133.0

------------------------------------------------------------------------ ------------------------------------------------------------------------ ぬぉっ、これは凄い。性能が1.6倍になる。自動的に電圧設定も変更されて、不可逆的にチップに(オーバークロック済みの)刻印が入るそうで、一度でも電圧変更をやらかしたRaspberry Piは修理保証が効かないそうだが、こんだけ性能が上がるなら使う価値はある。五千円もしないボードにしてはなかなかやるな。

参考までに、手元のマシン、NEC LaVie LL370/E をいってみよう。 BYTE UNIX Benchmarks (Version 5.1.3) System: lavie: GNU/Linux OS: GNU/Linux -- 3.16-3-686-pae -- #1 SMP Debian 3.16.5-1 (2014-10-10) Machine: i686 (unknown) Language: en_US.utf8 (charmap="ANSI_X3.4-1968", collate="ANSI_X3.4-1968") CPU 0: mobile AMD Sempron(tm) Processor 3000+ (1600.1 bogomips) MMX,AMD_MMX,Physical_Address_Ext,SYSENTER/SYSEXIT,SYSCALL/SYSRET 18:54:04 up 2 min, 1 user, load average: 0.63, 0.70, 0.31; runlevel 1 ------------------------------------------------------------------------ Benchmark Run: 月 11月 03 2014 18:54:04 - 19:22:24 1 CPU in system; running 1 parallel copy of tests

Dhrystone 2 using register variables________7347843.3 lps___(10.0 s, 7 samples) Double-Precision Whetstone_____________________1962.9 MWIPS_(10.0 s, 7 samples) Execl Throughput_______________________________1876.1 lps___(30.0 s, 2 samples) File Copy 1024 bufsize 2000 maxblocks________269729.0 KBps__(30.0 s, 2 samples) File Copy 256 bufsize 500 maxblocks___________95704.7 KBps__(30.0 s, 2 samples) File Copy 4096 bufsize 8000 maxblocks________444024.0 KBps__(30.0 s, 2 samples) Pipe Throughput______________________________742994.0 lps __(10.0 s, 7 samples) Pipe-based Context Switching_________________170994.2 lps___(10.0 s, 7 samples) Process Creation_______________________________5298.6 lps___(30.0 s, 2 samples) Shell Scripts (1 concurrent)___________________3287.2 lpm___(60.0 s, 2 samples) Shell Scripts (8 concurrent)____________________429.1 lpm___(60.1 s, 2 samples) System Call Overhead________________________1560181.6 lps___(10.0 s, 7 samples)

System Benchmarks Index Values_______________BASELINE_______RESULT____INDEX Dhrystone 2 using register variables_________116700.0____7347843.3____629.6 Double-Precision Whetstone_______________________55.0_______1962.9____356.9 Execl Throughput_________________________________43.0_______1876.1____436.3 File Copy 1024 bufsize 2000 maxblocks__________3960.0_____269729.0____681.1 File Copy 256 bufsize 500 maxblocks____________1655.0______95704.7____578.3 File Copy 4096 bufsize 8000 maxblocks__________5800.0_____444024.0____765.6 Pipe Throughput_______________________________12440.0_____742994.0____597.3 Pipe-based Context Switching___________________4000.0_____170994.2____427.5 Process Creation________________________________126.0_______5298.6____420.5 Shell Scripts (1 concurrent)_____________________42.4_______3287.2____775.3 Shell Scripts (8 concurrent)______________________6.0________429.1____715.1 System Call Overhead__________________________15000.0____1560181.6___1040.1 ___________________________________________________________________======== System Benchmarks Index Score_________________________________________591.4

------------------------------------------------------------------------ ------------------------------------------------------------------------ SPARC Station20-61/128MB-RAM+SPARC Storage Array with Solaris 2.3比較で 60倍速となるこのマシンも、KDEを使うにはちょっと遅い。(GNOMEは論外) 8倍速程度では、正直なところiceweasel(firefox)のようなブラウザすらマトモに動かない。

cyber205の日記：スピード測定のためBYTEBENCH取ったんだが@Raspberry Pi 0

スピード測定のためBYTEBENCH取ったんだが@Raspberry Pi More ログイン

スラド