Benchs

[euri10]

wip, trying to figure out if reads are indeed weirdly slow

[euri10]

im wondering if the sqlalchemy write part does something it feels weird that it would be faster than the base driver

❯ python scripts/bench.py --driver asyncpg --driver sqlite
Running benchmark for driver: asyncpg
Running benchmark for driver: sqlite
                          Benchmark Results
┏━━━━━━━━━┳━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━┳━━━━━━━━━━┳━━━━━━━━━━━━━━━━━┓
┃ Driver  ┃ Library    ┃ Scenario       ┃ Time (s) ┃ % Slower vs Raw ┃
┡━━━━━━━━━╇━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━╇━━━━━━━━━━╇━━━━━━━━━━━━━━━━━┩
│ asyncpg │ raw        │ initialization │   0.0497 │               — │
│ asyncpg │ sqlspec    │ initialization │   0.0977 │           96.6% │
│ asyncpg │ sqlalchemy │ initialization │   0.0329 │          -33.7% │
│ asyncpg │ raw        │ write_heavy    │  51.8678 │               — │
│ asyncpg │ sqlspec    │ write_heavy    │  56.8252 │            9.6% │
│ asyncpg │ sqlalchemy │ write_heavy    │   4.5475 │          -91.2% │
│ asyncpg │ raw        │ read_heavy     │   0.0721 │               — │
│ asyncpg │ sqlspec    │ read_heavy     │   0.1858 │          157.8% │
│ asyncpg │ sqlalchemy │ read_heavy     │   0.1080 │           49.9% │
│ sqlite  │ raw        │ initialization │   0.0007 │               — │
│ sqlite  │ sqlspec    │ initialization │   0.0029 │          343.8% │
│ sqlite  │ sqlalchemy │ initialization │   0.0043 │          555.7% │
│ sqlite  │ raw        │ write_heavy    │   0.0594 │               — │
│ sqlite  │ sqlspec    │ write_heavy    │   2.5934 │         4269.5% │
│ sqlite  │ sqlalchemy │ write_heavy    │   0.7772 │         1209.4% │
│ sqlite  │ raw        │ read_heavy     │   0.0804 │               — │
│ sqlite  │ sqlspec    │ read_heavy     │   2.5633 │         3087.9% │
│ sqlite  │ sqlalchemy │ read_heavy     │   0.7890 │          881.2% │
└─────────┴────────────┴────────────────┴──────────┴─────────────────┘
Benchmarks complete for drivers: asyncpg, sqlite