I've included more benchmarks in the project to compare GraphSON vs GraphBinary side by side, with more detail about different uses/types. The results from my machine are:
``` Benchmark Mode Cnt Score Error Units SerializationBenchmark.testReadMessage1Binary thrpt 20 4933443.595 ± 108335.893 ops/s SerializationBenchmark.testReadMessage1GraphSON thrpt 20 34318.763 ± 7782.995 ops/s SerializationBenchmark.testReadMessage2Binary thrpt 20 974269.201 ± 27497.059 ops/s SerializationBenchmark.testReadMessage2GraphSON thrpt 20 29928.189 ± 1187.881 ops/s SerializationBenchmark.testWriteBytecodeBinary thrpt 20 932400.254 ± 24727.644 ops/s SerializationBenchmark.testWriteBytecodeGraphSON thrpt 20 690590.285 ± 19508.398 ops/s SerializationBenchmark.testWriteResponseBinary thrpt 20 2629038.176 ± 67510.162 ops/s SerializationBenchmark.testWriteResponseGraphSON thrpt 20 652272.129 ± 20014.138 ops/s ``` These new results should be compared with the next group: ``` Benchmark Mode Cnt Score Error Units GraphBinaryReaderWriterBenchmark.readBytecode1 thrpt 20 1400024.496 ± 64637.425 ops/s GraphBinaryReaderWriterBenchmark.readBytecode2 thrpt 20 343833.159 ± 10609.128 ops/s GraphBinaryReaderWriterBenchmark.writeBytecode1 thrpt 20 1857617.437 ± 100685.512 ops/s GraphBinaryReaderWriterBenchmark.writeBytecode2 thrpt 20 431804.339 ± 24291.109 ops/s ``` ``` Benchmark Mode Cnt Score Error Units GraphSONMapperBenchmark.readBytecode1 thrpt 20 592331.268 ± 71423.780 ops/s GraphSONMapperBenchmark.readBytecode2 thrpt 20 123595.490 ± 9640.191 ops/s GraphSONMapperBenchmark.writeBytecode1 thrpt 20 1352647.895 ± 348521.676 ops/s GraphSONMapperBenchmark.writeBytecode2 thrpt 20 371473.035 ± 35986.231 ops/s ``` In the process, I've found that `P` deserialization can be optimized on both GraphSON and GraphBinary, I've filed TINKERPOP-2166 for it. [ Full content available at: https://github.com/apache/tinkerpop/pull/1065 ] This message was relayed via gitbox.apache.org for [email protected]
