[
https://issues.apache.org/jira/browse/IMPALA-3345?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17177327#comment-17177327
]
Tim Armstrong commented on IMPALA-3345:
---------------------------------------
I did try this with newer version of gcc and clang in godbolt and the code
generated seemed fine for int64_t
> Investigate issues with __builtin_*_overflow in LLVM
> ----------------------------------------------------
>
> Key: IMPALA-3345
> URL: https://issues.apache.org/jira/browse/IMPALA-3345
> Project: IMPALA
> Issue Type: Improvement
> Components: Backend
> Affects Versions: Impala 2.6.0
> Reporter: Tim Armstrong
> Priority: Minor
> Labels: codegen
>
> LLVM 3.8 added support for __builtin_add_overflow and __builtin_mul_overflow.
> We can use them in decimal-value-inline.h but there are various issues:
> * __builtin_mul_overflow for 128-bit ints (16-byte decimals) generates a
> reference to a runtime function implemented by LLVM's runtime library but not
> GCC's (we use the GCC runtime for all code).
> * __builtin_add_overflow seems extremely slow: I saw TPC-H query 1 regress by
> 3-4x compared to the current code.
> This JIRA is to investigate these issues and see if the functions can be
> beneficial.
--
This message was sent by Atlassian Jira
(v8.3.4#803005)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
