SubhamSinghal commented on PR #21730:
URL: https://github.com/apache/datafusion/pull/21730#issuecomment-4276014128
@Dandandan I have tested performance locally with Parquet file is 1.1 GB
with 20 million rows across ~2000 row groups (10K rows each).
Query looks like: `SELECT count(*), sum(v1) FROM bench WHERE id IN
list[]`
this is the result:
| Query | Main (ms) | Branch (ms) | Speedup | Expected |
|-------|-----------|-------------|---------|----------|
| Q1: narrow 50-vals [1..50] | 63 | 56 | 1.12x faster | Yes — prunes ~99%
of RGs |
| Q2: narrow 30-vals [5000..5029] | 61 | 60 | ~1.02x | Marginal — data
cached, I/O not dominant |
| Q3: wide 50-vals [1..9999] | 62 | 61 | ~1.02x | Expected — range covers
full span, no pruning |
| Q4: small 10-vals (control) | 57 | 57 | 1.00x | Expected — below
threshold, same path |
--
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]
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
