cyber205の日記： UnixBench on Raspberry Pi Model3 B+

BYTEBENCH の派生タイプUNIXベンチマーク。新しくラズパイを入手したので動かしてみた。
もう初代の頃にあったような遅さは感じなくなって、普通にミニPCとして使える。
この書き込みも現在、ベンチマークに使ったマシン上のFirefoxから書き込み中

___BYTE_UNIX_Benchmarks_(Version_5.1.3)
___System:_raspberrypi:_GNU/Linux
___OS:_GNU/Linux_--_4.14.62-v7+_--_#1134_SMP_Tue_Aug_14_17:10:10_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)
___09:50:48_up__8:06,__1_user,__load_average:_0.50,_0.30,_0.21;_runlevel_5
------------------------------------------------------------------------
５千円ぐらいで買えるマイコンボードに、１GHzオーバーの高速でブン回るプロセッサ(1.4GHz)を４コアも積んでる。
21世紀に入るか入らないかの頃は1GHzを超えるプロセッサが出現しただけでビッグニュースだったのに。
GbEに2.4GHz/6GHzのWiFi、BTにGPIO、カメラポートも装備。
もう、遊びでUNIX使うだけならx86である必要ないな。これってスマホ用プロセッサだっけ？
------------------------------------------------------------------------
まずはシングルコア性能
------------------------------------------------------------------------
Benchmark_Run:_木__9月_06_2018_09:50:48_-_10:19:04
4_CPUs_in_system;_running_1_parallel_copy_of_tests
Dhrystone_2_using_register_variables________4736332.5_lps___(10.0_s,_7_samples)
Double-Precision_Whetstone_____________________1075.2_MWIPS_(10.9_s,_7_samples)
Execl_Throughput________________________________822.0_lps___(29.8_s,_2_samples)
File_Copy_1024_bufsize_2000_maxblocks________133582.5_KBps__(30.0_s,_2_samples)
File_Copy_256_bufsize_500_maxblocks___________41891.9_KBps__(30.0_s,_2_samples)
File_Copy_4096_bufsize_8000_maxblocks________317229.2_KBps__(30.0_s,_2_samples)
Pipe_Throughput______________________________294369.9_lps___(10.0_s,_7_samples)
Pipe-based_Context_Switching__________________51445.5_lps___(10.0_s,_7_samples)
Process_Creation_______________________________1624.1_lps___(30.0_s,_2_samples)
Shell_Scripts_(1_concurrent)___________________1913.2_lpm___(60.0_s,_2_samples)
Shell_Scripts_(8_concurrent)____________________556.8_lpm___(60.1_s,_2_samples)
System_Call_Overhead_________________________598853.3_lps___(10.0_s,_7_samples)

System_Benchmarks_Index_Values_______________BASELINE_______RESULT____INDEX
Dhrystone_2_using_register_variables_________116700.0____4736332.5____405.9
Double-Precision_Whetstone_______________________55.0_______1075.2____195.5
Execl_Throughput_________________________________43.0________822.0____191.2
File_Copy_1024_bufsize_2000_maxblocks__________3960.0_____133582.5____337.3
File_Copy_256_bufsize_500_maxblocks____________1655.0______41891.9____253.1
File_Copy_4096_bufsize_8000_maxblocks__________5800.0_____317229.2____546.9
Pipe_Throughput_______________________________12440.0_____294369.9____236.6
Pipe-based_Context_Switching___________________4000.0______51445.5____128.6
Process_Creation________________________________126.0_______1624.1____128.9
Shell_Scripts_(1_concurrent)_____________________42.4_______1913.2____451.2
Shell_Scripts_(8_concurrent)______________________6.0________556.8____928.0
System_Call_Overhead__________________________15000.0_____598853.3____399.2
___________________________________________________________________========
System_Benchmarks_Index_Score_________________________________________296.7

------------------------------------------------------------------------
そして４コア並列運転した時の性能
------------------------------------------------------------------------
Benchmark_Run:_木__9月_06_2018_10:19:04_-_10:47:24
4_CPUs_in_system;_running_4_parallel_copies_of_tests

Dhrystone_2_using_register_variables_______16506360.9_lps___(10.0_s,_7_samples)
Double-Precision_Whetstone_____________________4117.9_MWIPS_(9.9_s,_7_samples)
Execl_Throughput_______________________________2201.2_lps___(29.9_s,_2_samples)
File_Copy_1024_bufsize_2000_maxblocks________214451.0_KBps__(30.0_s,_2_samples)
File_Copy_256_bufsize_500_maxblocks___________58974.3_KBps__(30.0_s,_2_samples)
File_Copy_4096_bufsize_8000_maxblocks________526484.5_KBps__(30.0_s,_2_samples)
Pipe_Throughput_____________________________1130803.7_lps___(10.0_s,_7_samples)
Pipe-based_Context_Switching_________________209719.8_lps___(10.0_s,_7_samples)
Process_Creation_______________________________4304.1_lps___(30.0_s,_2_samples)
Shell_Scripts_(1_concurrent)___________________4287.3_lpm___(60.1_s,_2_samples)
Shell_Scripts_(8_concurrent)____________________545.6_lpm___(60.1_s,_2_samples)
System_Call_Overhead________________________2243881.7_lps___(10.0_s,_7_samples)

System_Benchmarks_Index_Values_______________BASELINE_______RESULT____INDEX
Dhrystone_2_using_register_variables_________116700.0___16506360.9___1414.4
Double-Precision_Whetstone_______________________55.0_______4117.9____748.7
Execl_Throughput_________________________________43.0_______2201.2____511.9
File_Copy_1024_bufsize_2000_maxblocks__________3960.0_____214451.0____541.5
File_Copy_256_bufsize_500_maxblocks____________1655.0______58974.3____356.3
File_Copy_4096_bufsize_8000_maxblocks__________5800.0_____526484.5____907.7
Pipe_Throughput_______________________________12440.0____1130803.7____909.0
Pipe-based_Context_Switching___________________4000.0_____209719.8____524.3
Process_Creation________________________________126.0_______4304.1____341.6
Shell_Scripts_(1_concurrent)_____________________42.4_______4287.3___1011.2
Shell_Scripts_(8_concurrent)______________________6.0________545.6____909.4
System_Call_Overhead__________________________15000.0____2243881.7___1495.9
___________________________________________________________________========
System_Benchmarks_Index_Score_________________________________________725.5

むちゃくちゃパワフルじゃないですか。
こうなると、I/Oまわりがパソコン並みの高速バス(PCI Express)じゃないのがもったいないね。
残念ながら有線GbE接続はUSBの下にブラ下がっているのだそうだ。