Sean Smith
Sean Smith
c6a.metal: ``` glaredb> SELECT RegionID, SUM(AdvEngineID), COUNT(*) AS c, AVG(ResolutionWidth), COUNT(DISTINCT UserID) FROM hits GROUP BY RegionID ORDER BY c DESC LIMIT 10; ┌──────────┬─────────┬──────────┬────────────────────┬─────────┐ │ RegionID │ sum │ c...
Various partition settings on c6a.metal, perf regresses between 32 and 64 partitions. ``` glaredb> set partitions to 1; ┌─────────────────────┐ │ Query success │ │ No columns returned │ └─────────────────────┘ Execution...
~and also a ton of time on parquet metadata~ jk that's from mem allocation
Better after removing partitioned agg table locks: ``` glaredb> SELECT COUNT(DISTINCT UserID) FROM hits; ┌──────────┐ │ count │ │ Int64 │ ├──────────┤ │ 17630976 │ └──────────┘ Execution duration (s): 1.28571...
After moving hash table allocation to inside execution: ``` glaredb> SELECT COUNT(DISTINCT UserID) FROM hits; ┌──────────┐ │ count │ │ Int64 │ ├──────────┤ │ 17630976 │ └──────────┘ Execution duration (s):...
Do we have this?
How are you computing the right results?
Also how did you generate the data set? `revenue` for me a decimal, not a float.
We can definitely look into this. Is there a particular database you'd want to see supported through odbc?