Bug 84467 - Choosing between Integer and NEON for 64-bit operations
Summary: Choosing between Integer and NEON for 64-bit operations
Status: RESOLVED FIXED
Alias: None
Product: gcc
Classification: Unclassified
Component: target (show other bugs)
Version: 8.0.1
: P3 normal
Target Milestone: 9.5
Assignee: Not yet assigned to anyone
URL:
Keywords: missed-optimization
Depends on:
Blocks:
 
Reported: 2018-02-19 15:49 UTC by sudi
Modified: 2021-06-01 11:22 UTC (History)
0 users

See Also:
Host:
Target: arm
Build:
Known to work:
Known to fail:
Last reconfirmed: 2018-02-19 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 sudi 2018-02-19 15:49:11 UTC 
This is a follow up report to bug 82989
The comment bug 82989, comment 12 details about the need for early decisions to be made about choosing to take either NEON code or ARM code. This means that at the expand phase, we should be able to make a clear choice and avoid mixing the two.
Comment 1 ktkachov 2018-02-19 15:53:03 UTC 
Confirmed.
Comment 2 Jakub Jelinek 2019-05-03 09:16:07 UTC 
GCC 9.1 has been released.
Comment 3 Jakub Jelinek 2019-08-12 08:55:04 UTC 
GCC 9.2 has been released.
Comment 4 Jakub Jelinek 2020-03-12 11:58:59 UTC 
GCC 9.3.0 has been released, adjusting target milestone.
Comment 5 Richard Biener 2021-06-01 08:10:23 UTC 
GCC 9.4 is being released, retargeting bugs to GCC 9.5.
Comment 6 Richard Earnshaw 2021-06-01 11:22:26 UTC 
This was ultimately addressed by deciding not to use neon at all for 64-bit operations.  Too many cases lead to very poor performance and with better optimization on the general regs expansions (which is only possible if a hard choice is made early) we generally end up with better code overall.