From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: from us-smtp-delivery-124.mimecast.com (us-smtp-delivery-124.mimecast.com [216.205.24.124]) by sourceware.org (Postfix) with ESMTP id C5ED83858D39 for ; Wed, 15 Sep 2021 09:25:43 +0000 (GMT) DMARC-Filter: OpenDMARC Filter v1.4.1 sourceware.org C5ED83858D39 Received: from mimecast-mx01.redhat.com (mimecast-mx01.redhat.com [209.132.183.4]) (Using TLS) by relay.mimecast.com with ESMTP id us-mta-401-h2CLwJiUMbGO9AttnufArA-1; Wed, 15 Sep 2021 05:25:40 -0400 X-MC-Unique: h2CLwJiUMbGO9AttnufArA-1 Received: from smtp.corp.redhat.com (int-mx03.intmail.prod.int.phx2.redhat.com [10.5.11.13]) (using TLSv1.2 with cipher AECDH-AES256-SHA (256/256 bits)) (No client certificate requested) by mimecast-mx01.redhat.com (Postfix) with ESMTPS id 2051D802C87; Wed, 15 Sep 2021 09:25:39 +0000 (UTC) Received: from tucnak.zalov.cz (unknown [10.39.192.10]) by smtp.corp.redhat.com (Postfix) with ESMTPS id 5728E60622; Wed, 15 Sep 2021 09:25:38 +0000 (UTC) Received: from tucnak.zalov.cz (localhost [127.0.0.1]) by tucnak.zalov.cz (8.16.1/8.16.1) with ESMTPS id 18F9PYg91744483 (version=TLSv1.3 cipher=TLS_AES_256_GCM_SHA384 bits=256 verify=NOT); Wed, 15 Sep 2021 11:25:34 +0200 Received: (from jakub@localhost) by tucnak.zalov.cz (8.16.1/8.16.1/Submit) id 18F9PVmV1744482; Wed, 15 Sep 2021 11:25:31 +0200 Date: Wed, 15 Sep 2021 11:25:31 +0200 From: Jakub Jelinek To: Richard Biener Cc: "Liu, Hongtao" , Thomas Schwinge , "gcc@gcc.gnu.org" , Tobias Burnus Subject: Re: GCC/OpenMP offloading for Intel GPUs? Message-ID: <20210915092531.GB304296@tucnak> Reply-To: Jakub Jelinek References: <87v933nlhn.fsf@dem-tschwing-1.ger.mentorg.com> MIME-Version: 1.0 In-Reply-To: X-Scanned-By: MIMEDefang 2.79 on 10.5.11.13 X-Mimecast-Spam-Score: 0 X-Mimecast-Originator: redhat.com Content-Type: text/plain; charset=us-ascii Content-Disposition: inline X-Spam-Status: No, score=-5.8 required=5.0 tests=BAYES_00, DKIMWL_WL_HIGH, DKIM_SIGNED, DKIM_VALID, DKIM_VALID_AU, DKIM_VALID_EF, RCVD_IN_DNSWL_LOW, RCVD_IN_MSPIKE_H2, SPF_HELO_NONE, SPF_NONE, TXREP autolearn=ham autolearn_force=no version=3.4.4 X-Spam-Checker-Version: SpamAssassin 3.4.4 (2020-01-24) on server2.sourceware.org X-BeenThere: gcc@gcc.gnu.org X-Mailman-Version: 2.1.29 Precedence: list List-Id: Gcc mailing list List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , X-List-Received-Date: Wed, 15 Sep 2021 09:25:44 -0000 On Wed, Sep 15, 2021 at 11:19:29AM +0200, Richard Biener wrote: > On Wed, Sep 15, 2021 at 4:02 AM Liu, Hongtao via Gcc wrote: > > > > I got some feedback from my colleague > > > > ----------------- > > What we need from GCC > > > > 1. generate SPIR-V > > But is SPIR-V powerful enough here, if wikipedia is right and it is > close to GLSL > then it likely has not the ability to perform calls? You'd need sth > like HSAIL then. I believe SPIR-V is essentially LLVM IR, except with some restrictions and perhaps small additions. > > 2. offload bundler to create FAT object We already produce FAT objects our way. > > -------------- > > > > If the answer is yes for both, they can hook it up with libomptarget library and our IGC back-end. We really can't use libomptarget, it clashes with what libgomp provides which is handling both the host OpenMP threads and OpenMP/OpenACC offloading. Is the IGC back-end open source? Does it use LLVM to compile SPIR-V into whatever ISA the Intel GPGPUs have? Jakub