You are viewing a plain text version of this content. The canonical link for it is here.
Posted to jira@arrow.apache.org by "Antoine Pitrou (Jira)" <ji...@apache.org> on 2021/11/03 18:09:00 UTC
[jira] [Commented] (ARROW-14481) [C++] Investigate recent
regressions in some utf8 kernel benchmarks
[ https://issues.apache.org/jira/browse/ARROW-14481?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17438224#comment-17438224 ]
Antoine Pitrou commented on ARROW-14481:
----------------------------------------
I do not see any regression locally between git master and a random commit from October 21st.
{code}
------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
Non-regressions: (26)
------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
benchmark baseline contender change % counters
Utf8Lower 644.618 MiB/sec 814.359 MiB/sec 26.332 {'family_index': 11, 'per_family_instance_index': 0, 'run_name': 'Utf8Lower', 'repetitions': 1, 'repetition_index': 0, 'threads': 1, 'iterations': 29}
Utf8Upper 647.102 MiB/sec 779.122 MiB/sec 20.402 {'family_index': 12, 'per_family_instance_index': 0, 'run_name': 'Utf8Upper', 'repetitions': 1, 'repetition_index': 0, 'threads': 1, 'iterations': 29}
IsAlphaNumericAscii 506.334 MiB/sec 577.621 MiB/sec 14.079 {'family_index': 2, 'per_family_instance_index': 0, 'run_name': 'IsAlphaNumericAscii', 'repetitions': 1, 'repetition_index': 0, 'threads': 1, 'iterations': 22}
MatchLike 808.991 MiB/sec 863.400 MiB/sec 6.725 {'family_index': 7, 'per_family_instance_index': 0, 'run_name': 'MatchLike', 'repetitions': 1, 'repetition_index': 0, 'threads': 1, 'iterations': 38}
TrimManyUtf8 669.685 MiB/sec 714.090 MiB/sec 6.631 {'family_index': 15, 'per_family_instance_index': 0, 'run_name': 'TrimManyUtf8', 'repetitions': 1, 'repetition_index': 0, 'threads': 1, 'iterations': 29}
BinaryJoinArrayArray 1.105 GiB/sec 1.131 GiB/sec 2.336 {'family_index': 17, 'per_family_instance_index': 0, 'run_name': 'BinaryJoinArrayArray', 'repetitions': 1, 'repetition_index': 0, 'threads': 1, 'iterations': 6817}
BinaryJoinElementWiseArrayScalar/64 1.018 GiB/sec 1.038 GiB/sec 1.894 {'family_index': 18, 'per_family_instance_index': 2, 'run_name': 'BinaryJoinElementWiseArrayScalar/64', 'repetitions': 1, 'repetition_index': 0, 'threads': 1, 'iterations': 102}
BinaryJoinElementWiseArrayArray/8 754.305 MiB/sec 768.563 MiB/sec 1.890 {'family_index': 19, 'per_family_instance_index': 1, 'run_name': 'BinaryJoinElementWiseArrayArray/8', 'repetitions': 1, 'repetition_index': 0, 'threads': 1, 'iterations': 564}
MatchLikePrefix 3.605 GiB/sec 3.665 GiB/sec 1.663 {'family_index': 9, 'per_family_instance_index': 0, 'run_name': 'MatchLikePrefix', 'repetitions': 1, 'repetition_index': 0, 'threads': 1, 'iterations': 161}
AsciiUpper 7.936 GiB/sec 8.051 GiB/sec 1.441 {'family_index': 1, 'per_family_instance_index': 0, 'run_name': 'AsciiUpper', 'repetitions': 1, 'repetition_index': 0, 'threads': 1, 'iterations': 359}
MatchLikeSuffix 3.637 GiB/sec 3.678 GiB/sec 1.132 {'family_index': 10, 'per_family_instance_index': 0, 'run_name': 'MatchLikeSuffix', 'repetitions': 1, 'repetition_index': 0, 'threads': 1, 'iterations': 165}
TrimSingleAscii 1.233 GiB/sec 1.245 GiB/sec 0.970 {'family_index': 5, 'per_family_instance_index': 0, 'run_name': 'TrimSingleAscii', 'repetitions': 1, 'repetition_index': 0, 'threads': 1, 'iterations': 57}
AsciiLower 7.933 GiB/sec 8.009 GiB/sec 0.957 {'family_index': 0, 'per_family_instance_index': 0, 'run_name': 'AsciiLower', 'repetitions': 1, 'repetition_index': 0, 'threads': 1, 'iterations': 365}
TrimManyAscii 955.260 MiB/sec 962.251 MiB/sec 0.732 {'family_index': 6, 'per_family_instance_index': 0, 'run_name': 'TrimManyAscii', 'repetitions': 1, 'repetition_index': 0, 'threads': 1, 'iterations': 42}
BinaryJoinElementWiseArrayScalar/8 816.534 MiB/sec 822.348 MiB/sec 0.712 {'family_index': 18, 'per_family_instance_index': 1, 'run_name': 'BinaryJoinElementWiseArrayScalar/8', 'repetitions': 1, 'repetition_index': 0, 'threads': 1, 'iterations': 612}
BinaryJoinElementWiseArrayArray/2 591.531 MiB/sec 594.633 MiB/sec 0.525 {'family_index': 19, 'per_family_instance_index': 0, 'run_name': 'BinaryJoinElementWiseArrayArray/2', 'repetitions': 1, 'repetition_index': 0, 'threads': 1, 'iterations': 1767}
MatchSubstring 577.765 MiB/sec 579.194 MiB/sec 0.247 {'family_index': 3, 'per_family_instance_index': 0, 'run_name': 'MatchSubstring', 'repetitions': 1, 'repetition_index': 0, 'threads': 1, 'iterations': 25}
BinaryJoinElementWiseArrayScalar/2 747.787 MiB/sec 748.576 MiB/sec 0.106 {'family_index': 18, 'per_family_instance_index': 0, 'run_name': 'BinaryJoinElementWiseArrayScalar/2', 'repetitions': 1, 'repetition_index': 0, 'threads': 1, 'iterations': 2232}
IsAlphaNumericUnicode 1005.992 MiB/sec 1005.034 MiB/sec -0.095 {'family_index': 13, 'per_family_instance_index': 0, 'run_name': 'IsAlphaNumericUnicode', 'repetitions': 1, 'repetition_index': 0, 'threads': 1, 'iterations': 44}
BinaryJoinArrayScalar 1.238 GiB/sec 1.236 GiB/sec -0.195 {'family_index': 16, 'per_family_instance_index': 0, 'run_name': 'BinaryJoinArrayScalar', 'repetitions': 1, 'repetition_index': 0, 'threads': 1, 'iterations': 7191}
BinaryJoinElementWiseArrayArray/128 1.225 GiB/sec 1.222 GiB/sec -0.246 {'family_index': 19, 'per_family_instance_index': 3, 'run_name': 'BinaryJoinElementWiseArrayArray/128', 'repetitions': 1, 'repetition_index': 0, 'threads': 1, 'iterations': 58}
BinaryJoinElementWiseArrayArray/64 1.014 GiB/sec 1.008 GiB/sec -0.594 {'family_index': 19, 'per_family_instance_index': 2, 'run_name': 'BinaryJoinElementWiseArrayArray/64', 'repetitions': 1, 'repetition_index': 0, 'threads': 1, 'iterations': 94}
BinaryJoinElementWiseArrayScalar/128 1.244 GiB/sec 1.232 GiB/sec -1.014 {'family_index': 18, 'per_family_instance_index': 3, 'run_name': 'BinaryJoinElementWiseArrayScalar/128', 'repetitions': 1, 'repetition_index': 0, 'threads': 1, 'iterations': 60}
SplitPattern 458.727 MiB/sec 451.400 MiB/sec -1.597 {'family_index': 4, 'per_family_instance_index': 0, 'run_name': 'SplitPattern', 'repetitions': 1, 'repetition_index': 0, 'threads': 1, 'iterations': 20}
MatchLikeSubstring 566.577 MiB/sec 548.685 MiB/sec -3.158 {'family_index': 8, 'per_family_instance_index': 0, 'run_name': 'MatchLikeSubstring', 'repetitions': 1, 'repetition_index': 0, 'threads': 1, 'iterations': 25}
TrimSingleUtf8 982.468 MiB/sec 942.122 MiB/sec -4.107 {'family_index': 14, 'per_family_instance_index': 0, 'run_name': 'TrimSingleUtf8', 'repetitions': 1, 'repetition_index': 0, 'threads': 1, 'iterations': 44}
{code}
> [C++] Investigate recent regressions in some utf8 kernel benchmarks
> -------------------------------------------------------------------
>
> Key: ARROW-14481
> URL: https://issues.apache.org/jira/browse/ARROW-14481
> Project: Apache Arrow
> Issue Type: Improvement
> Components: C++
> Reporter: David Li
> Priority: Major
> Labels: kernel
>
> See [https://conbench.ursa.dev/benchmarks/6ccff6887e7c47148a09fe46f18c8688/]
> Some (on the surface) unrelated commits have caused performance for a few string kernels to plummet. We should try to replicate locally.
--
This message was sent by Atlassian Jira
(v8.3.4#803005)