概要
- BYTE UNIX Benchmarksで、ベンチマーク.
- 普通にやると、1回1時間のテスト。
環境
- Raspberry Pi 3 +
2017-03-02-raspbian-jessie
(4.4.50) + ファン付ケース - Tinker Board + TinkerOS v1.4 (4.4.16)
- Tinker Board + CustomBuildOS (4.4.52)
手順
キホン、こんな感じ。もうちょっとなんかパッケージ入れたかも..
sudo apt-get install git-core build-essential
git clone https://github.com/kdlucas/byte-unixbench.git
cd byte-unixbench/UnixBench
./Run
-
RPi3では、Makefileを編集 (
-march=native -mtune=native
があるとエラー出る)diff --git a/UnixBench/Makefile b/UnixBench/Makefile index d6868ba..1dd7e42 100644 --- a/UnixBench/Makefile +++ b/UnixBench/Makefile @@ -87,7 +87,7 @@ OPTON = -O3 -ffast-math OSNAME:=$(shell uname -s) ifeq ($(OSNAME),Linux) -OPTON += -march=native -mtune=native +#OPTON += -march=native -mtune=native endif ifeq ($(OSNAME),Darwin) # (adjust flags or comment out this section for older versions of XCode or OS X)
結果
とりあえず、表に埋めただけ...
RESULT | running 1 parallel copy of tests | running 4 parallel copies of tests | |||||
---|---|---|---|---|---|---|---|
System Benchmarks Index Values | BASELINE | RPi3 4.4.50-v7+ | TinkerOS v1.4 4.4.16 | CustomBuild 4.4.52 | RPi3 4.4.50-v7+ | TinkerOS v1.4 4.4.16 | CustomBuild 4.4.52 |
Dhrystone 2 using register variables | 116700.0 | 4518310.1 | 8866656.0 | 10094986.0 | 18041530.0 | 28069515.5 | 30422699.0 |
Double-Precision Whetstone | 55.0 | 780.2 | 1221.3 | 1494.8 | 3118.2 | 4191.8 | 5071.9 |
Execl Throughput | 43.0 | 615.3 | 791.4 | 974.9 | 2258.6 | 4046.7 | 4141.1 |
File Copy 1024 bufsize 2000 maxblocks | 3960.0 | 152828.9 | 171877.4 | 173252.4 | 245301.5 | 269298.5 | 218831.1 |
File Copy 256 bufsize 500 maxblocks | 1655.0 | 44207.5 | 65857.3 | 57547.3 | 66563.1 | 84120.9 | 62352.0 |
File Copy 4096 bufsize 8000 maxblocks | 5800.0 | 360868.5 | 471021.1 | 421426.1 | 624393.6 | 778903.3 | 683966.7 |
Pipe Throughput | 12440.0 | 286377.4 | 414751.2 | 362544.6 | 1139462.8 | 1509070.2 | 1253660.1 |
Pipe-based Context Switching | 4000.0 | 58794.8 | 25190.0 | 48032.5 | 196303.1 | 302571.7 | 148544.1 |
Process Creation | 126.0 | 2496.4 | 3192.2 | 369.6 | 5085.2 | 9703.9 | 8500.8 |
Shell Scripts (1 concurrent) | 42.4 | 2215.7 | 2812.4 | 1352.4 | 4944.6 | 8245.1 | 7484.8 |
Shell Scripts (8 concurrent) | 6.0 | 610.3 | 1114.1 | 1020.5 | 661.0 | 1077.0 | 1025.5 |
System Call Overhead | 15000.0 | 700103.4 | 871505.5 | 783442.5 | 2688464.6 | 2939861.8 | 2394917.0 |
INDEX | running 1 parallel copy of tests | running 4 parallel copies of tests | |||||
System Benchmarks Index Values | RPi3 4.4.50-v7+ | TinkerOS v1.4 4.4.16 | CustomBuild 4.4.52 | RPi3 4.4.50-v7+ | TinkerOS v1.4 4.4.16 | CustomBuild 4.4.52 | |
Dhrystone 2 using register variables | 387.2 | 759.8 | 865.0 | 1546.0 | 2405.3 | 2606.9 | |
Double-Precision Whetstone | 141.8 | 222.1 | 271.8 | 566.9 | 762.1 | 922.2 | |
Execl Throughput | 143.1 | 184.0 | 226.7 | 525.3 | 941.1 | 963 | |
File Copy 1024 bufsize 2000 maxblocks | 385.9 | 434.0 | 437.5 | 619.4 | 680.0 | 552.6 | |
File Copy 256 bufsize 500 maxblocks | 267.1 | 397.9 | 347.7 | 402.2 | 508.3 | 376.7 | |
File Copy 4096 bufsize 8000 maxblocks | 622.2 | 812.1 | 726.6 | 1076.5 | 1342.9 | 1179.3 | |
Pipe Throughput | 230.2 | 333.4 | 291.4 | 916.0 | 1213.1 | 1007.8 | |
Pipe-based Context Switching | 147.0 | 63.0 | 120.1 | 490.8 | 756.4 | 371.4 | |
Process Creation | 198.1 | 253.3 | 29.3 | 403.6 | 770.1 | 674.7 | |
Shell Scripts (1 concurrent) | 522.6 | 663.3 | 319.0 | 1166.2 | 1944.6 | 1765.3 | |
Shell Scripts (8 concurrent) | 1017.2 | 1856.9 | 1700.9 | 1101.6 | 1795.0 | 1709.2 | |
System Call Overhead | 466.7 | 581.0 | 522.3 | 1792.3 | 1959.9 | 1596.6 | |
System Benchmarks Index Score | 311.6 | 399.6 | 330.7 | 781.1 | 1119.4 | 969.7 |