mbutrovich commented on PR #12212:
URL: https://github.com/apache/arrow-rs/pull/12212#issuecomment-6865628209

   @Rich-T-kid I kicked off that run to check the AMD64 BinaryView slowdown at 
1% selectivity. On both aarch64 and x86_64, the `StringViewType` copies of 
`append_indexed_views_and_copy_strings` and 
`append_indexed_views_and_copy_strings_split` compile to aliases of the 
`BinaryViewType` copies. So the BinaryView cases run the same machine code as 
the Utf8View cases that got 7-18% faster, and the difference should be noise or 
code layout.
   
   The run fixes function and block alignment on both sides to take layout out 
of the comparison. The `primitive` take cases don't reach this code, so they 
show the noise floor. The `mixed_*view (max_string_len=128)` cases take the 
same path with mostly out-of-line values, and no earlier run included them.
   


-- 
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]

Reply via email to