cx.read_sql almost 2.5x slower than pandas.read_sql

[gsakkis]

I was excited to come across connector-x for improving some queries but for some reason the performance is nowhere close to the benchmarks shown at the README. For example for the same query:

• pandas: 22s, mem usage 650MB
• connector-x-: 52s, mem usage 500MB

Memory usage was computed from psutil.virtual_memory().used

This is against a MS SQL Server with pyodb 5.1.0 and ODBC Driver 18 for SQL Server. Any idea why it's 2.5x slower than pandas?

[wangxiaoying]

Hi @gsakkis , to answer your question we may need more information about your workload. But in general you can check on this discussion here.