[
https://issues.apache.org/jira/browse/IMPALA-14729?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Zoltán Borók-Nagy updated IMPALA-14729:
---------------------------------------
Description:
At the time of writing, XXH3 is considered to be the "best hash algorithm" in
terms of speed, quality, and it works well on different input sizes.
It has 64-bit and 128-bit variants.
We should asses the Impala code, and see if we should replace
* Murmur
* FNV
* FastHash
* ? CRC
Related tickets are
* IMPALA-2281
* IMPALA-3255
Those all pre-date XXH3 which was initially released in 2019/2020.
We could also add XXH3_HASH() as a built-in function to Impala. Just like we
have MURMUR_HASH() and FNV_HASH().
was:
At the time of writing, XXH3 is considered to be the "best hash algorithm" in
terms of speed, quality, and it works well on different input sizes.
It has 64-bit and 128-bit variants.
We should asses the Impala code, and see if we should replace
* Murmur
* FNV
* FastHash
* ? CRC
Related tickets are
* IMPALA-2281
* IMPALA-3255
Those all pre-date XXH3 which was initially released in 2019/2020.
> Investigate XXH3
> ----------------
>
> Key: IMPALA-14729
> URL: https://issues.apache.org/jira/browse/IMPALA-14729
> Project: IMPALA
> Issue Type: Improvement
> Components: Backend, Frontend
> Reporter: Zoltán Borók-Nagy
> Priority: Major
> Labels: performance, ramp-up
>
> At the time of writing, XXH3 is considered to be the "best hash algorithm" in
> terms of speed, quality, and it works well on different input sizes.
> It has 64-bit and 128-bit variants.
> We should asses the Impala code, and see if we should replace
> * Murmur
> * FNV
> * FastHash
> * ? CRC
> Related tickets are
> * IMPALA-2281
> * IMPALA-3255
> Those all pre-date XXH3 which was initially released in 2019/2020.
> We could also add XXH3_HASH() as a built-in function to Impala. Just like we
> have MURMUR_HASH() and FNV_HASH().
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
