clickhouse / ch-bench Goto Github PK

Hi @ernado, it seems not all benchmarks considered de-serialization(not only reading blocks but also converting bytes into uint64). Taking ch-bench-official as an example, rows.Scan should be called before count++ according to the example below:

https://github.com/ClickHouse/clickhouse-go/blob/a22f61818dd9c2d23e457c792b8dca44ab875060/benchmark/v2/read/main.go#L43-L53

Thanks for benchmarking.

I've noticed that not all libraries are in the same condition. For instance, loyd/clickhouse.rs uses LZ4 by default, that's a reasonable default, but the best library uses no compression by default AFAIK. Disabling it increases performance x2 on my machine.

I would like to see more attention to such details for all other libraries, not only for your favorite one.

mt - multi thread

select * from system.numbers limit 1000000000 format Null;
0 rows in set. Elapsed: 0.475 sec. Processed 1.00 billion rows, 8.00 GB (2.10 billion rows/s., 16.83 GB/s.)

select * from system.numbers_mt limit 1000000000 format Null;
0 rows in set. Elapsed: 0.305 sec. Processed 1.00 billion rows, 8.00 GB (3.28 billion rows/s., 26.21 GB/s.)

In readme: The main goal is to measure minimal client overhead (CPU, RAM) to **read data**, i.e. data blocks deserialization and transfer.

Maybe adding benchmarks to measure inserting performance will help people who are in heavy insert use cases to choose clickhouse client.