IPC benchmark on Linux which is inspired by APUE
Just Make style as the following :D
make && make testThese are kinds of IPC in Linux as the following:
| type | feature |
|---|---|
| pipe | unnamed pipe |
| fifo | named pipe |
| socketpair | unnamed unix domain socket |
| unix domain socket | named unix domain socket |
| TCP | remote domain socket |
| UDP | loopback interface |
#Benchmark
In my personal MacBook Pro (Retina, 13-inch, Late 2013) Intel(R) Core(TM) i5-4258U CPU @ 2.40GHz, the benchmark is as the following:
pipe
128 512 1024 4096
1319Mb/s 5110Mb/s 8932Mb/s 20297Mb/s
1288233msg/s 1247449msg/s 1090370msg/s 619407msg/s
fifo
128 512 1024 4096
1358Mb/s 5491Mb/s 8440Mb/s 22018Mb/s
1326016msg/s 1340502msg/s 1030215msg/s 671929msg/s
socketpair
128 512 1024 4096
1117Mb/s 4401Mb/s 8869Mb/s 17207Mb/s
1090370msg/s 1074548msg/s 1082674msg/s 525121msg/s
uds
128 512 1024 4096
100Mb/s 102Mb/s 109Mb/s 97Mb/s
97600msg/s 24986msg/s 13289msg/s 2968msg/s
tcp
128 512 1024 4096
106Mb/s 143Mb/s 135Mb/s 128Mb/s
103508msg/s 34852msg/s 16529msg/s 3897msg/sIn Hasee Notebook Intel(R) Core(TM) i7-4710MQ @ 2.5GHz, with DDR3 2400 8GB x1, the benchmark is as the following:
pipe
128 512 1024 4096
1808Mb/s 6899Mb/s 10836Mb/s 33358Mb/s
1766023msg/s 1684301msg/s 1322749msg/s 1018006msg/s
fifo
128 512 1024 4096
884Mb/s 3758Mb/s 7333Mb/s 27825Mb/s
863310msg/s 917532msg/s 895158msg/s 849143msg/s
socketpair
128 512 1024 4096
752Mb/s 3591Mb/s 7418Mb/s 20385Mb/s
734559msg/s 876703msg/s 905538msg/s 622098msg/s
uds
128 512 1024 4096
70Mb/s 522Mb/s 843Mb/s 1175Mb/s
68744msg/s 127364msg/s 102957msg/s 35857msg/s
tcp
128 512 1024 4096
197Mb/s 498Mb/s 951Mb/s 1174Mb/s
192263msg/s 121701msg/s 116046msg/s 35828msg/s
udp
128 512 1024 4096
155Mb/s 497Mb/s 691Mb/s 2907Mb/s
151295msg/s 121241msg/s 84371msg/s 88712msg/sAlso we make the benchmark on travis-ci
To contribute to ipc_benchmark, clone this repo locally and commit your code on a separate branch.
GitHub @detailyang
ipc_benchmark is licensed under the MIT license.
..