You are viewing a plain text version of this content. The canonical link for it is here.

Posted to issues@arrow.apache.org by "Antoine Pitrou (Jira)" <ji...@apache.org> on 2021/08/12 17:26:00 UTC

[jira] [Created] (ARROW-13619) [C++] Hash aggregation seems to ignore ARROW_USER_SIMD_LEVEL

Antoine Pitrou created ARROW-13619:
--------------------------------------

             Summary: [C++] Hash aggregation seems to ignore ARROW_USER_SIMD_LEVEL
                 Key: ARROW-13619
                 URL: https://issues.apache.org/jira/browse/ARROW-13619
             Project: Apache Arrow
          Issue Type: Bug
          Components: C++
    Affects Versions: 5.0.0
            Reporter: Antoine Pitrou
             Fix For: 6.0.0


The {{ARROW_USER_SIMD_LEVEL}} environment variable allows to override the SIMD optimization level at runtime. For example {{ARROW_USER_SIMD_LEVEL=NONE}} is useful to benchmark a non-SIMD path.

However, it seems the hash aggregation code doesn't react to this. Hence, on an AVX2-enabled machine (which is really the majority of high-end x86 computers), it's cumbersome to run benchmarks for these code paths.



--
This message was sent by Atlassian Jira
(v8.3.4#803005)