From mboxrd@z Thu Jan  1 00:00:00 1970
Return-Path: <jvdelisle@charter.net>
Received: from impout004.msg.chrl.nc.charter.net
 (impout004aa.msg.chrl.nc.charter.net [47.43.20.28])
 by sourceware.org (Postfix) with ESMTPS id E04FF3857C73;
 Fri, 19 Mar 2021 19:50:00 +0000 (GMT)
DMARC-Filter: OpenDMARC Filter v1.3.2 sourceware.org E04FF3857C73
Authentication-Results: sourceware.org;
 dmarc=none (p=none dis=none) header.from=charter.net
Authentication-Results: sourceware.org;
 spf=pass smtp.mailfrom=jvdelisle@charter.net
Received: from [192.168.1.9] ([96.41.221.129]) by cmsmtp with ESMTPA
 id NL8AlXVTRxxx6NL8BlbDSQ; Fri, 19 Mar 2021 19:50:00 +0000
Authentication-Results: charter.net; none
X-Authority-Analysis: v=2.3 cv=TK7xtGta c=1 sm=1 tr=0
 a=07pILqX15KmGv9ZXTMmBNA==:117 a=07pILqX15KmGv9ZXTMmBNA==:17
 a=IkcTkHD0fZMA:10 a=AheCPF5wyFTC3HSU4g8A:9 a=QEXdDO2ut3YA:10
Subject: Re: [patch, fortran] Also use size estimate for vector-matrix matmul
To: Thomas Koenig <tkoenig@netcologne.de>,
 gcc-patches <gcc-patches@gcc.gnu.org>,
 "fortran@gcc.gnu.org" <fortran@gcc.gnu.org>
References: <01315155-a238-c1ff-c08a-b0319422f001@netcologne.de>
From: Jerry DeLisle <jvdelisle@charter.net>
Message-ID: <1273b3de-e65b-b5fa-9c73-6cc515e28989@charter.net>
Date: Fri, 19 Mar 2021 12:49:58 -0700
User-Agent: Mozilla/5.0 (X11; Linux x86_64; rv:78.0) Gecko/20100101
 Thunderbird/78.8.0
MIME-Version: 1.0
In-Reply-To: <01315155-a238-c1ff-c08a-b0319422f001@netcologne.de>
Content-Type: text/plain; charset=utf-8; format=flowed
Content-Transfer-Encoding: 8bit
Content-Language: en-US
X-CMAE-Envelope: MS4wfN0XqCNXBidVkeuWJ17n+rELv4gwGNj5myYx4jJfS6Qc1Md40svThcdcqhKS5b2Fy/voOZAXlYEyMR0MDnBZqNS+niQgvPN+mltWh12NpCFkbGlk2mp1
 lmLGHcJbraUllEivh1HXOqUHGW7wAIyBlsRu1FajYqrwJKrF2bTKVylkDo5obi9tuxaFDOnaq+IcpR+JRf27zyt8IVsAoGys+PELU5ZUATu4DKGQVvLQT/Rv
 R8xtaVTfMucCKghE3D5Hfg==
X-Spam-Status: No, score=-1.8 required=5.0 tests=BAYES_00, BODY_8BITS,
 FREEMAIL_FROM, KAM_DMARC_STATUS, NICE_REPLY_A, RCVD_IN_DNSWL_LOW,
 RCVD_IN_MSPIKE_H3, RCVD_IN_MSPIKE_WL, SPF_HELO_NONE, SPF_PASS,
 TXREP autolearn=ham autolearn_force=no version=3.4.2
X-Spam-Checker-Version: SpamAssassin 3.4.2 (2018-09-13) on
 server2.sourceware.org
X-BeenThere: fortran@gcc.gnu.org
X-Mailman-Version: 2.1.29
Precedence: list
List-Id: Fortran mailing list <fortran.gcc.gnu.org>
List-Unsubscribe: <https://gcc.gnu.org/mailman/options/fortran>,
 <mailto:fortran-request@gcc.gnu.org?subject=unsubscribe>
List-Archive: <https://gcc.gnu.org/pipermail/fortran/>
List-Help: <mailto:fortran-request@gcc.gnu.org?subject=help>
List-Subscribe: <https://gcc.gnu.org/mailman/listinfo/fortran>,
 <mailto:fortran-request@gcc.gnu.org?subject=subscribe>
X-List-Received-Date: Fri, 19 Mar 2021 19:50:02 -0000

Yes Ok for trunk.

Thanks much!

On 3/19/21 10:37 AM, Thomas Koenig via Fortran wrote:
> Hell world,
>
> here is the patch I talked about earlier.  It passes regression testing.
>
> OK for trunk?
>
> Best regards
>
>     Thomas
>
> Add size check to vector-matrix matmul.
>
> It turns out the library version is much faster for vector-matrix
> multiplications for large sizes than what inlining can produce.
> Use size checks for switching between this and inlining for
> that case to.
>
> gcc/fortran/ChangeLog:
>
>     * frontend-passes.c (inline_limit_check): Add rank_a
>     argument. If a is rank 1, set the second dimension to 1.
>     (inline_matmul_assign): Pass rank_a argument to inline_limit_check.
>     (call_external_blas): Likewise.
>
> gcc/testsuite/ChangeLog:
>
>     * gfortran.dg/inline_matmul_6.f90: Adjust count for _gfortran_matmul.