性能测试

Linux 测试环境

• go1.6.2 linux/amd64
• Ubuntu 16.04 amd64
• 虚拟机
• 2 个处理器核心
• 2 GB 内存

并行测试

BenchmarkParallelHprose2-2    	  200000	      9239 ns/op	     789 B/op	      25 allocs/op
BenchmarkParallelHprose2Unix-2	  200000	      6400 ns/op	     789 B/op	      25 allocs/op
BenchmarkParallelHprose-2     	  100000	     12541 ns/op	    4405 B/op	      63 allocs/op
BenchmarkParallelHproseUnix-2 	  200000	      9907 ns/op	    4405 B/op	      63 allocs/op
BenchmarkParallelGobRPC-2     	  100000	     17895 ns/op	     360 B/op	      12 allocs/op
BenchmarkParallelGobRPCUnix-2 	  100000	     12373 ns/op	     360 B/op	      12 allocs/op
BenchmarkParallelJSONRPC-2    	  100000	     21693 ns/op	    1218 B/op	      30 allocs/op
BenchmarkParallelJSONRPCUnix-2	  100000	     15956 ns/op	    1218 B/op	      30 allocs/op

非并行测试

BenchmarkHprose2-2            	  100000	     14252 ns/op	     789 B/op	      25 allocs/op
BenchmarkHprose2Unix-2        	  200000	      9885 ns/op	     789 B/op	      25 allocs/op
BenchmarkHprose-2             	  100000	     18234 ns/op	    4405 B/op	      63 allocs/op
BenchmarkHproseUnix-2         	  100000	     13060 ns/op	    4405 B/op	      63 allocs/op
BenchmarkGobRPC-2             	  100000	     20534 ns/op	     360 B/op	      12 allocs/op
BenchmarkGobRPCUnix-2         	  100000	     14351 ns/op	     360 B/op	      12 allocs/op
BenchmarkJSONRPC-2            	   50000	     24350 ns/op	    1224 B/op	      30 allocs/op
BenchmarkJSONRPCUnix-2        	  100000	     18321 ns/op	    1224 B/op	      30 allocs/op

Mac 测试环境

• go1.7.1 darwin/amd64
• macOS Sierra
• iMac (Retina 5K, 27-inch, Late 2015)
• 处理器 4GHz Intel Core i7
• 内存 32 GB 1867 MHz DDR3

并行测试

BenchmarkParallelHprose2-8       	  200000	     11080 ns/op	     791 B/op	      25 allocs/op
BenchmarkParallelHprose2Unix-8   	  300000	      5238 ns/op	     791 B/op	      25 allocs/op
BenchmarkParallelHprose-8        	  200000	     11243 ns/op	    4405 B/op	      63 allocs/op
BenchmarkParallelHproseUnix-8    	  300000	      5702 ns/op	    4405 B/op	      63 allocs/op
BenchmarkParallelGobRPC-8        	  100000	     16815 ns/op	     360 B/op	      12 allocs/op
BenchmarkParallelGobRPCUnix-8    	  200000	      6707 ns/op	     360 B/op	      12 allocs/op
BenchmarkParallelJSONRPC-8       	  100000	     17649 ns/op	    1216 B/op	      30 allocs/op
BenchmarkParallelJSONRPCUnix-8   	  200000	      7809 ns/op	    1216 B/op	      30 allocs/op

非并行测试

BenchmarkHprose2-8               	   50000	     34164 ns/op	     791 B/op	      25 allocs/op
BenchmarkHprose2Unix-8           	  100000	     12012 ns/op	     791 B/op	      25 allocs/op
BenchmarkHprose-8                	   50000	     35938 ns/op	    4405 B/op	      63 allocs/op
BenchmarkHproseUnix-8            	  100000	     13122 ns/op	    4405 B/op	      63 allocs/op
BenchmarkGobRPC-8                	   30000	     47610 ns/op	     360 B/op	      12 allocs/op
BenchmarkGobRPCUnix-8            	   50000	     24507 ns/op	     360 B/op	      12 allocs/op
BenchmarkJSONRPC-8               	   30000	     51826 ns/op	    1224 B/op	      30 allocs/op
BenchmarkJSONRPCUnix-8           	   50000	     29195 ns/op	    1224 B/op	      30 allocs/op

---

• Hprose2 表示 Hprose 2.0
• Hprose 表示 Hprose 1.x
• GobRPC 表示 go 内置的 net/rpc
• JSONRPC 表示 go 内置的 net/rpc/jsonrpc
• 带 Unix 表示 Unix Socket 通讯，不带 Unix 表示 TCP 通讯