Closed wangkuiyi closed 5 years ago
The benchmark shows that this PR improves the reading speed for about 1000 times!
The benchmark on develop branch at this moment:
yi@WangYis-iMac:/go/src/github.com/wangkuiyi/recordio (develop)$ go test -bench=. goos: darwin goarch: amd64 pkg: github.com/wangkuiyi/recordio BenchmarkRead/reading_records_00000_to_00050-4 5 297171427 ns/op BenchmarkRead/reading_records_00050_to_00100-4 3 486892504 ns/op BenchmarkRead/reading_records_00100_to_00150-4 5 293632646 ns/op PASS ok github.com/wangkuiyi/recordio 8.859s
The benchmark of this PR:
yi@WangYis-iMac:/go/src/github.com/wangkuiyi/recordio (use_pipeline_in_io)$ go test -bench=. goos: darwin goarch: amd64 pkg: github.com/wangkuiyi/recordio BenchmarkRead/reading_records_00000_to_00050-4 5000 414011 ns/op BenchmarkRead/reading_records_00050_to_00100-4 3000 535504 ns/op BenchmarkRead/reading_records_00100_to_00150-4 5000 473778 ns/op PASS ok github.com/wangkuiyi/recordio 6.589s
The performance improvement hopefully fixes https://github.com/wangkuiyi/recordio/issues/50.
The benchmark shows that this PR improves the reading speed for about 1000 times!
The benchmark on develop branch at this moment:
The benchmark of this PR:
The performance improvement hopefully fixes https://github.com/wangkuiyi/recordio/issues/50.