refactor(hashtable): adaptive string hash table

[usamoi]

I hereby agree to the terms of the CLA available at: https://databend.rs/dev/policies/cla/

Summary

It's the pull request about OSPP 2022 Adaptive String Hash Table.

[vercel]

The latest updates on your projects. Learn more about Vercel for Git ↗︎

1 Ignored Deployment

Name	Status	Preview	Updated
databend	⬜️ Ignored (Inspect)		Oct 27, 2022 at 1:38PM (UTC)

[FANNG1]

cool, is there any performance stats if using adaptive string hash table ?

[PsiACE]

cool, is there any performance stats if using adaptive string hash table ?

https://github.com/usamoi/saha/blob/main/doc/benchmark.md

No performance evaluation of databend with saha is available at this time, but for the performance of the saha implementation, see the link above.

[Xuanwo]

Hi, any updates on this PR? The conflict files list is longer day by day...

Is it possible to split it into multiple so that we don't need to pending on a large PR?

[sundy-li]

I just bench it with the hits dataset. It shows 20%~50% performance improvement in group aggregation queries.

Query	Main	new hashtable	ratio（old/new)	comments
Q1	1.05	0.94	1.117021277
Q2	2.9	1.98	1.464646465	distinct int
Q3	1.5	1.3	1.153846154	distinct string
Q4	0.05	0.05	1
Q5	2.62	1.78	1.471910112	group by int
Q6	2.79	1.97	1.416243655	group by int
Q7	0.38	0.3	1.266666667	group by string
Q8	0.43	0.34	1.264705882	group by string
Q9	3.1	2.83	1.09540636
Q10	3.42	3.42	1
Q11	20.72	20.83	0.9947191551
Q12	4.56	4.74	0.9620253165
Q13	6.62	6.02	1.099667774
Q14	16.98	15.83	1.072646873

https://docs.google.com/spreadsheets/d/1BnNEoaA37XxhCQwzwMxp2gRK04tp8hqnwbY6m1pVT1o/edit#gid=0

[PsiACE]

@usamoi Great job, glad to see you did it.

cc @sundy-li , Once the version is released, we can consider updating the version in ClickBench.

[Xuanwo]

@mergify update

[mergify]

update

✅ Branch has been successfully updated

[sundy-li]

Awesome job!

[BohuTANG]

Great job!