Bug 91836 - [10 Regression] Speed regression of 525.x264_r with -Ofast -march=native since r275982
Summary: [10 Regression] Speed regression of 525.x264_r with -Ofast -march=native sinc...
Status: RESOLVED FIXED
Alias: None
Product: gcc
Classification: Unclassified
Component: ipa (show other bugs)
Version: 10.0
: P3 normal
Target Milestone: 10.0
Assignee: Martin Liška
URL:
Keywords:
Depends on:
Blocks: spec
  Show dependency tree / graph
 
Reported: 2019-09-20 11:59 UTC by Martin Liška
Modified: 2019-09-26 08:59 UTC (History)
1 user (show)

See Also:
Host:
Target:
Build:
Known to work: 9.2.0
Known to fail: 10.0
Last reconfirmed: 2019-09-21 00:00:00

Attachments
Add an attachment (proposed patch, testcase, etc.)

Note You need to log in before you can comment on or make changes to this bug.
Description Martin Liška 2019-09-20 11:59:09 UTC 
Seen on the graphs and I verified that locally with train run on znver1:
https://lnt.opensuse.org/db_default/v4/SPEC/graph?plot.0=35.407.0&plot.1=32.377.0
Comment 1 Richard Biener 2019-09-20 12:26:59 UTC 
It's likely been one of the vectorizer changes and fixed by

2019-09-20  Richard Biener  <rguenther@suse.de>

        PR tree-optimization/91821
        * tree-vect-loop.c (check_reduction_path): Check we can compute
        reduc_idx.
        (vect_is_simple_reduction): Set STMT_VINFO_REDUC_IDX.
        * tree-vect-patterns.c (vect_reassociating_reduction_p): Return
        operands in canonical order.
        * tree-vectorizer.c (vec_info::new_stmt_vec_info): Initialize
        STMT_VINFO_REDUC_IDX.
        * tree-vectorizer.h (_stmt_vec_info::reduc_idx): New.
        (STMT_VINFO_REDUC_IDX): Likewise.
Comment 2 Martin Liška 2019-09-20 12:34:44 UTC 
Ah, ok, let's wait for periodic testers over the weekend.
Comment 3 Martin Liška 2019-09-20 12:48:20 UTC 
One additional related can be:
https://lnt.opensuse.org/db_default/v4/SPEC/graph?plot.0=35.407.0
Comment 4 Martin Liška 2019-09-26 08:59:14 UTC 
I can confirm it's fixed.