[
https://issues.apache.org/jira/browse/ARROW-1588?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16175166#comment-16175166
]
Tim Armstrong commented on ARROW-1588:
--------------------------------------
Ah ok, makes sense. I don't know to what extent this applies but our experience
is that decimal operations are a lot faster on narrower 4-byte and 8-byte
representations. One reason is that the 4 byte and 8 byte decimal values fit in
registers and can be manipulated with normal integer operations. A more subtle
reason is that implementing some operations correctly (at least in Impala's
implementation) requires temporarily promoting to a wider type, e.g. 4 byte ->
8 byte or 8 byte -> 16 bytes. Emulated 128-bit and 256-bit operations are
pretty slow.
> [C++/Format] Harden Decimal Format
> ----------------------------------
>
> Key: ARROW-1588
> URL: https://issues.apache.org/jira/browse/ARROW-1588
> Project: Apache Arrow
> Issue Type: Improvement
> Components: C++, Format
> Affects Versions: 0.7.0
> Reporter: Phillip Cloud
> Assignee: Phillip Cloud
> Fix For: 0.8.0
>
>
> We should finalize and harden the decimal format. The remaining issues are
> officially writing down the choice of making every decimal value 16 bytes and
> byte order.
> For byte order we'll need to run some benchmarks to compare little endian vs
> big endian. I plan to work on this over the next week or two.
> [[email protected]] [~wesmckinn] If there are any additional items you'd
> like to see addressed here please chime in.
--
This message was sent by Atlassian JIRA
(v6.4.14#64029)
