Rossi Sun

I think in tidb 4.0, arbitrary column type modification is an experimental feature, meaning it may not be fully implemented or is problematic. So I suggest pre-creating tables with decimal...

Hi Bobby, thank you for the comments. That's very helpful! Yes I believe a pinned memory pool could definitely be a big win. I'll keep watching the progress of rmm's...

> I am not sure how deep you can get your pipelined calls to be. Since our pipeline model will aggressively pipe as many cuDF calls as possible, a pipeline...

> Yeah, I did that. > > I used `mysqldump` to make a copy of the uploaded database, and then replaced the `prepare` stage with flashing TiDB with the cached...

Meanwhile we'll try to reproduce your issue in our local. Please give us some time. Thanks.

Hi Yujin, We have some findings: 1. The very first crash, i.e. `free(): invalid pointer`, was caused by the defect in tidb branch https://github.com/csm-yujinkim/tidb/tree/tidb_cura. Using the tidb binary I gave...

> 这个有点那个 嗯？你能那个的再那个一点么？

+1 The improvement is very nice and clear.

Normally this projection can be done by inserting a `Projection` node between `TableScan` and `HashJoin` nodes during the planning phase. So it is not necessary for a hash join algorithm...

Will the swiss table currently being used in acero's hash join (aka. swiss join) help on this?