From: Uros Bizjak <ubizjak@gmail.com>
To: "gcc-patches@gcc.gnu.org" <gcc-patches@gcc.gnu.org>
Subject: [PATCH] i386: Remove mulv2si emulated sequence for TARGET_SSE2 [PR109797]
Date: Fri, 12 May 2023 18:43:50 +0200 [thread overview]
Message-ID: <CAFULd4Z7q=6W_0Y2Xe0ZpWdHr+5-v4h+ZpDR7ekj_L3zDDXWOA@mail.gmail.com> (raw)
[-- Attachment #1: Type: text/plain, Size: 563 bytes --]
Remove mulv2si emulated sequence for TARGET_SSE2 and enable
only native PMULLD instruction for TARGET_SSE4_1. Ideally, the
vectorization for TARGET_SSE2 should depend on more precise cost
estimation (the PR contains patch for ix86_multiplication_cost),
but even with patched cost function the runtime regression
was not fixed.
PR target/109797
gcc/ChangeLog:
* config/i386/mmx.md (mulv2si3): Remove expander.
(mulv2si3): Rename insn pattern from *mulv2si.
Bootstrapped and regression tested on x86_64-linux-gnu {,-m32}.
Pushed to master.
Uros.
[-- Attachment #2: p.diff.txt --]
[-- Type: text/plain, Size: 1499 bytes --]
diff --git a/gcc/config/i386/mmx.md b/gcc/config/i386/mmx.md
index e7ca921dd2b..b2954fff8ae 100644
--- a/gcc/config/i386/mmx.md
+++ b/gcc/config/i386/mmx.md
@@ -2092,39 +2092,7 @@ (define_insn "*<insn><mode>3"
(set_attr "type" "sseadd")
(set_attr "mode" "TI")])
-(define_expand "mulv2si3"
- [(set (match_operand:V2SI 0 "register_operand")
- (mult:V2SI
- (match_operand:V2SI 1 "register_operand")
- (match_operand:V2SI 2 "register_operand")))]
- "TARGET_MMX_WITH_SSE"
-{
- if (!TARGET_SSE4_1)
- {
- rtx op1 = lowpart_subreg (V4SImode, force_reg (V2SImode, operands[1]),
- V2SImode);
- rtx op2 = lowpart_subreg (V4SImode, force_reg (V2SImode, operands[2]),
- V2SImode);
-
- rtx tmp1 = gen_reg_rtx (V4SImode);
- emit_insn (gen_vec_interleave_lowv4si (tmp1, op1, op1));
- rtx tmp2 = gen_reg_rtx (V4SImode);
- emit_insn (gen_vec_interleave_lowv4si (tmp2, op2, op2));
-
- rtx res = gen_reg_rtx (V2DImode);
- emit_insn (gen_vec_widen_umult_even_v4si (res, tmp1, tmp2));
-
- rtx op0 = gen_reg_rtx (V4SImode);
- emit_insn (gen_sse2_pshufd_1 (op0, gen_lowpart (V4SImode, res),
- const0_rtx, const2_rtx,
- const0_rtx, const2_rtx));
-
- emit_move_insn (operands[0], lowpart_subreg (V2SImode, op0, V4SImode));
- DONE;
- }
-})
-
-(define_insn "*mulv2si3"
+(define_insn "mulv2si3"
[(set (match_operand:V2SI 0 "register_operand" "=Yr,*x,v")
(mult:V2SI
(match_operand:V2SI 1 "register_operand" "%0,0,v")
reply other threads:[~2023-05-12 16:44 UTC|newest]
Thread overview: [no followups] expand[flat|nested] mbox.gz Atom feed
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to='CAFULd4Z7q=6W_0Y2Xe0ZpWdHr+5-v4h+ZpDR7ekj_L3zDDXWOA@mail.gmail.com' \
--to=ubizjak@gmail.com \
--cc=gcc-patches@gcc.gnu.org \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for read-only IMAP folder(s) and NNTP newsgroup(s).