← Back to index | Original Bugzilla link

Bug 16914 – Add -mavx switch to generate AVX instructions instead of SSE instructions

Status: RESOLVED
Resolution: WONTFIX
Severity: enhancement
Priority: P1
Component: dmd
Product: D
Version: D2
Platform: All
OS: All
Creation time: 2016-12-02T06:57:00Z
Last change time: 2016-12-08T05:22:25Z
Keywords: SIMD
Assigned to: nobody
Creator: bugzilla
See also: https://issues.dlang.org/show_bug.cgi?id=16958

Comments

Comment #0 by bugzilla — 2016-12-02T06:57:26Z

Because they're smaller & faster, and they have 3 operands instead of 2. Note that this is not necessarily supporting 32 byte vectors.

Comment #1 by bugzilla — 2016-12-02T06:58:28Z

A start: https://github.com/dlang/dmd/pull/6293

Comment #2 by bugzilla — 2016-12-03T05:00:28Z

https://github.com/dlang/dlang.org/pull/1521

Comment #3 by bugzilla — 2016-12-08T05:22:25Z

Superceded by https://issues.dlang.org/show_bug.cgi?id=16958