From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: by sourceware.org (Postfix, from userid 48) id 3CDDE3857734; Thu, 11 May 2023 08:06:22 +0000 (GMT) DKIM-Filter: OpenDKIM Filter v2.11.0 sourceware.org 3CDDE3857734 DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=gcc.gnu.org; s=default; t=1683792382; bh=nYZ1r6NYl90gzK+5xavm0Ul5Hqj0r0TEbJ9SVRGNp4I=; h=From:To:Subject:Date:In-Reply-To:References:From; b=rpoY9CEAOSS8EZI0JNbDsFQ/SKG8+Mv9Dpgc7EUmUQs4bq6w3uY67+6B/p2wkkLt8 zbcoYzmHjZUe1BKaAi1vAVbcQUq+NUerDKufPaTaABqfuM85HXggvF3ebEaFCBH9Dm 5ZdWLvarV2fS0umCtZXMAOuiLjxPI5vvO/YS6w0A= From: "ubizjak at gmail dot com" To: gcc-bugs@gcc.gnu.org Subject: [Bug target/109807] [14 Regression] sse2-mmx-pmaddwd.c met ICE after commit r14-666-g608e7f3ab47 with march=cascadelake Date: Thu, 11 May 2023 08:06:22 +0000 X-Bugzilla-Reason: CC X-Bugzilla-Type: changed X-Bugzilla-Watch-Reason: None X-Bugzilla-Product: gcc X-Bugzilla-Component: target X-Bugzilla-Version: 14.0 X-Bugzilla-Keywords: X-Bugzilla-Severity: normal X-Bugzilla-Who: ubizjak at gmail dot com X-Bugzilla-Status: UNCONFIRMED X-Bugzilla-Resolution: X-Bugzilla-Priority: P3 X-Bugzilla-Assigned-To: unassigned at gcc dot gnu.org X-Bugzilla-Target-Milestone: --- X-Bugzilla-Flags: X-Bugzilla-Changed-Fields: Message-ID: In-Reply-To: References: Content-Type: text/plain; charset="UTF-8" Content-Transfer-Encoding: quoted-printable X-Bugzilla-URL: http://gcc.gnu.org/bugzilla/ Auto-Submitted: auto-generated MIME-Version: 1.0 List-Id: https://gcc.gnu.org/bugzilla/show_bug.cgi?id=3D109807 --- Comment #5 from Uro=C5=A1 Bizjak --- (In reply to Haochen Jiang from comment #4) > (In reply to Uro=C5=A1 Bizjak from comment #2) > > (In reply to Haochen Jiang from comment #1) > > > I further checked the reason, V2SI should never dropped into that fun= ction > > > because we have no pattern under V2SI. > > >=20 > > > I suppose it is because -march=3Dcascadelake will open SSE4.1, with t= he new > > > pattern, it wrongly dropped into that. > > >=20 > > > -m32 will not ICE since TARGET_MMX_WITH_SSE need 64 bit and won't ena= ble the > > > new pattern. > >=20 > > V2SI mul was introduced in r14-493 (AKA partial fix for PR109690). >=20 > I see. So we might need to add cost for that right? Adding "case V2SImode:" above "case V4SImode:" in ix86_widen_mult_cost will= fix this issue. Probably V4HImode is also needed above V8HImode, we have had a mulv4hi3 pattern for some time now.=