lidavidm commented on PR #569:
URL: https://github.com/apache/arrow-adbc/pull/569#issuecomment-1492078860
Some numbers, but they are highly variable + I'm running against Dockerized
postgres on localhost which is likely not representative at all.
Some thoughts:
- asyncpg does impressively well despite being at a disadvantage here, we
should look into it more
- DuckDB's parallelization doesn't quite seem to help here
- Performance is too variable in the multi-column benchmark
<details>
```
[ 56.25%] ··· =========== =========== =========== ===========
==================
-- data_type
-----------
------------------------------------------------------
row_count INT BIGINT FLOAT DOUBLE
PRECISION
=========== =========== =========== ===========
==================
10000 6.88±2ms 6.77±2ms 6.91±2ms 6.73±2ms
100000 55.6±20ms 62.8±20ms 58.8±20ms 60.4±20ms
1000000 563±200ms 571±300ms 576±200ms 590±200ms
=========== =========== =========== ===========
==================
[ 62.50%] ··· benchmarks.MultiColumnSuite.time_pandas_asyncpg
ok
[ 62.50%] ··· =========== =========== =========== ===========
==================
-- data_type
-----------
------------------------------------------------------
row_count INT BIGINT FLOAT DOUBLE
PRECISION
=========== =========== =========== ===========
==================
10000 8.35±3ms 8.28±3ms 6.03±2ms 5.94±2ms
100000 80.4±30ms 79.9±30ms 59.3±20ms 62.5±20ms
1000000 792±300ms 810±400ms 607±200ms 616±200ms
=========== =========== =========== ===========
==================
[ 68.75%] ··· benchmarks.MultiColumnSuite.time_pandas_duckdb
ok
[ 68.75%] ··· =========== =========== =========== ===========
==================
-- data_type
-----------
------------------------------------------------------
row_count INT BIGINT FLOAT DOUBLE
PRECISION
=========== =========== =========== ===========
==================
10000 12.7±1ms 12.9±1ms 13.1±2ms 13.2±1ms
100000 42.4±10ms 44.1±10ms 44.4±10ms 46.5±10ms
1000000 573±100ms 640±200ms 644±200ms 616±200ms
=========== =========== =========== ===========
==================
[ 75.00%] ··· benchmarks.MultiColumnSuite.time_pandas_psycopg2
ok
[ 75.00%] ··· =========== =========== =========== ===========
==================
-- data_type
-----------
------------------------------------------------------
row_count INT BIGINT FLOAT DOUBLE
PRECISION
=========== =========== =========== ===========
==================
10000 19.0±4ms 19.3±4ms 18.0±3ms 18.1±3ms
100000 111±40ms 115±40ms 103±30ms 103±30ms
1000000 1.08±0.4s 1.10±0.4s 998±300ms 994±300ms
=========== =========== =========== ===========
==================
[ 81.25%] ··· benchmarks.OneColumnSuite.time_pandas_adbc
ok
[ 81.25%] ··· =========== ============= ============ =============
==================
-- data_type
-----------
-----------------------------------------------------------
row_count INT BIGINT FLOAT DOUBLE
PRECISION
=========== ============= ============ =============
==================
10000 4.06±0.07ms 4.32±0.2ms 4.20±0.06ms
4.19±0.03ms
100000 32.7±0.2ms 33.6±0.2ms 34.6±0.4ms
34.4±0.6ms
1000000 329±5ms 327±1ms 338±9ms
332±0.4ms
=========== ============= ============ =============
==================
[ 87.50%] ··· benchmarks.OneColumnSuite.time_pandas_asyncpg
ok
[ 87.50%] ··· =========== ============= ============= =============
==================
-- data_type
-----------
------------------------------------------------------------
row_count INT BIGINT FLOAT DOUBLE
PRECISION
=========== ============= ============= =============
==================
10000 4.60±0.03ms 4.60±0.03ms 3.79±0.01ms
3.82±0.01ms
100000 43.2±0.3ms 43.6±0.3ms 36.1±4ms
35.6±0.4ms
1000000 448±2ms 446±1ms 370±2ms
369±2ms
=========== ============= ============= =============
==================
[ 93.75%] ··· benchmarks.OneColumnSuite.time_pandas_duckdb
ok
[ 93.75%] ··· =========== ============ ============ ============
==================
-- data_type
-----------
---------------------------------------------------------
row_count INT BIGINT FLOAT DOUBLE
PRECISION
=========== ============ ============ ============
==================
10000 11.2±0.3ms 11.3±0.3ms 11.3±0.2ms
11.2±0.2ms
100000 29.6±0.2ms 30.1±0.5ms 30.2±0.8ms
30.2±0.6ms
1000000 422±5ms 414±6ms 417±2ms
414±3ms
=========== ============ ============ ============
==================
[100.00%] ··· benchmarks.OneColumnSuite.time_pandas_psycopg2
ok
[100.00%] ··· =========== ============ ============ ============
==================
-- data_type
-----------
---------------------------------------------------------
row_count INT BIGINT FLOAT DOUBLE
PRECISION
=========== ============ ============ ============
==================
10000 15.0±0.1ms 15.3±0.1ms 15.0±0.3ms
14.8±0.1ms
100000 72.3±0.8ms 73.7±0.4ms 69.2±2ms
68.7±0.6ms
1000000 681±9ms 682±2ms 644±6ms
643±4ms
=========== ============ ============ ============
==================
```
</details>
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: [email protected]
For queries about this service, please contact Infrastructure at:
[email protected]
