From patchwork Thu Nov 10 10:44:52 2016 Content-Type: text/plain; charset="utf-8" MIME-Version: 1.0 Content-Transfer-Encoding: 7bit X-Patchwork-Submitter: Chung-Lin Tang X-Patchwork-Id: 693138 Return-Path: X-Original-To: incoming@patchwork.ozlabs.org Delivered-To: patchwork-incoming@bilbo.ozlabs.org Received: from sourceware.org (server1.sourceware.org [209.132.180.131]) (using TLSv1.2 with cipher ECDHE-RSA-AES256-GCM-SHA384 (256/256 bits)) (No client certificate requested) by ozlabs.org (Postfix) with ESMTPS id 3tF0806gBkz9t25 for ; Thu, 10 Nov 2016 21:45:52 +1100 (AEDT) Authentication-Results: ozlabs.org; dkim=pass (1024-bit key; unprotected) header.d=gcc.gnu.org header.i=@gcc.gnu.org header.b="pHJQ0at1"; dkim-atps=neutral DomainKey-Signature: a=rsa-sha1; c=nofws; d=gcc.gnu.org; h=list-id :list-unsubscribe:list-archive:list-post:list-help:sender:from :subject:to:cc:message-id:date:mime-version:content-type; q=dns; s=default; b=Oowr+7h+d2yWIIPvLAVhU/Zo+nUP/CpiD6EdbpWrp8BavLAUB7 SXo2SBubum3FpiRS3RwOqdN/ZT4Akh3+IvdMHrWwCl7FbdFIIylxlCeZk5yki+Fd LP//diYVde/M/wZbHZoEFYPgAFDNpyTeZ9lF5SkxnHaFsAN954bZAt+1g= DKIM-Signature: v=1; a=rsa-sha1; c=relaxed; d=gcc.gnu.org; h=list-id :list-unsubscribe:list-archive:list-post:list-help:sender:from :subject:to:cc:message-id:date:mime-version:content-type; s= default; bh=g0tb16t/HOOehCGhCiEsioqyfbA=; b=pHJQ0at1z/wojXy3JqBP v10ZbUm+ahscaYMxQQuyQfqxoJwIUH+Fkv7hkGT28tYAmOYKgDEu37MiLZ7EbDpz U1HRqus/bOc0CVjnNrSWnvOecfVvDm7Jqy6PD+Y1CFSHwEhoAWTkLH+qEcCrfexX aaHB3G5bcA5A5A84UIkADfg= Received: (qmail 29143 invoked by alias); 10 Nov 2016 10:45:43 -0000 Mailing-List: contact gcc-patches-help@gcc.gnu.org; run by ezmlm Precedence: bulk List-Id: List-Unsubscribe: List-Archive: List-Post: List-Help: Sender: gcc-patches-owner@gcc.gnu.org Delivered-To: mailing list gcc-patches@gcc.gnu.org Received: (qmail 28973 invoked by uid 89); 10 Nov 2016 10:45:35 -0000 Authentication-Results: sourceware.org; auth=none X-Virus-Found: No X-Spam-SWARE-Status: No, score=-1.1 required=5.0 tests=AWL, BAYES_00, KAM_ASCII_DIVIDERS, RCVD_IN_DNSWL_NONE, SPF_PASS, URIBL_RED autolearn=no version=3.3.2 spammy=tree-nested.c, treeh, UD:tree.c, UD:omp-low.c X-HELO: relay1.mentorg.com Received: from relay1.mentorg.com (HELO relay1.mentorg.com) (192.94.38.131) by sourceware.org (qpsmtpd/0.93/v0.84-503-g423c35a) with ESMTP; Thu, 10 Nov 2016 10:45:13 +0000 Received: from svr-orw-mbx-04.mgc.mentorg.com ([147.34.90.204]) by relay1.mentorg.com with esmtp id 1c4mqt-0002Qt-7S from ChungLin_Tang@mentor.com ; Thu, 10 Nov 2016 02:45:03 -0800 Received: from svr-orw-mbx-04.mgc.mentorg.com (147.34.90.204) by SVR-ORW-MBX-04.mgc.mentorg.com (147.34.90.204) with Microsoft SMTP Server (TLS) id 15.0.1210.3; Thu, 10 Nov 2016 02:45:00 -0800 Received: from [0.0.0.0] (147.34.91.1) by svr-orw-mbx-04.mgc.mentorg.com (147.34.90.204) with Microsoft SMTP Server (TLS) id 15.0.1210.3 via Frontend Transport; Thu, 10 Nov 2016 02:44:56 -0800 From: Chung-Lin Tang Subject: [Patch 1/5] OpenACC tile clause support, OMP_CLAUSE_TILE adjustments To: gcc-patches , Jakub Jelinek CC: Nathan Sidwell , Cesar Philippidis , Chung-Lin Tang Message-ID: <9f4c0d8a-b26f-573e-0746-395dba321c8f@mentor.com> Date: Thu, 10 Nov 2016 18:44:52 +0800 User-Agent: Mozilla/5.0 (Macintosh; Intel Mac OS X 10.11; rv:45.0) Gecko/20100101 Thunderbird/45.4.0 MIME-Version: 1.0 This patch contains a few supporting changes that adjusts how OMP_CLAUSE_TILE is handled. This is in support of the more elaborate omp-low.c changes in another patch. Thanks, Chung-Lin 2016-XX-XX Nathan Sidwell * tree.h (OMP_CLAUSE_TILE_ITERVAR, OMP_CLAUSE_TILE_COUNT): New. * tree.c (omp_clause_num_ops): Adjust TILE ops. * tree-nested.c (convert_nonlocal_omp_clauses): Allow OMP_CLAUSE_TILE. * gimplify.c (gimplify_scan_omp_clauses): No special handling for OMP_CLAUSE_TILE. (gomplify_adjust_omp_clauses): Don't delete TILE. (gimplify_omp_for): Deal with TILE. Index: tree.c =================================================================== --- tree.c (revision 241809) +++ tree.c (working copy) @@ -327,7 +327,7 @@ unsigned const char omp_clause_num_ops[] = 1, /* OMP_CLAUSE_NUM_GANGS */ 1, /* OMP_CLAUSE_NUM_WORKERS */ 1, /* OMP_CLAUSE_VECTOR_LENGTH */ - 1, /* OMP_CLAUSE_TILE */ + 3, /* OMP_CLAUSE_TILE */ 2, /* OMP_CLAUSE__GRIDDIM_ */ }; Index: tree.h =================================================================== --- tree.h (revision 241809) +++ tree.h (working copy) @@ -1654,6 +1654,10 @@ extern void protected_set_expr_location (tree, loc #define OMP_CLAUSE_TILE_LIST(NODE) \ OMP_CLAUSE_OPERAND (OMP_CLAUSE_SUBCODE_CHECK (NODE, OMP_CLAUSE_TILE), 0) +#define OMP_CLAUSE_TILE_ITERVAR(NODE) \ + OMP_CLAUSE_OPERAND (OMP_CLAUSE_SUBCODE_CHECK (NODE, OMP_CLAUSE_TILE), 1) +#define OMP_CLAUSE_TILE_COUNT(NODE) \ + OMP_CLAUSE_OPERAND (OMP_CLAUSE_SUBCODE_CHECK (NODE, OMP_CLAUSE_TILE), 2) #define OMP_CLAUSE__GRIDDIM__DIMENSION(NODE) \ (OMP_CLAUSE_SUBCODE_CHECK (NODE, OMP_CLAUSE__GRIDDIM_)\ Index: tree-nested.c =================================================================== --- tree-nested.c (revision 241809) +++ tree-nested.c (working copy) @@ -1274,6 +1274,7 @@ convert_nonlocal_omp_clauses (tree *pclauses, stru case OMP_CLAUSE_DEFAULT: case OMP_CLAUSE_COPYIN: case OMP_CLAUSE_COLLAPSE: + case OMP_CLAUSE_TILE: case OMP_CLAUSE_UNTIED: case OMP_CLAUSE_MERGEABLE: case OMP_CLAUSE_PROC_BIND: @@ -1286,8 +1287,6 @@ convert_nonlocal_omp_clauses (tree *pclauses, stru case OMP_CLAUSE_AUTO: break; - /* OpenACC tile clauses are discarded during gimplification. */ - case OMP_CLAUSE_TILE: /* The following clause belongs to the OpenACC cache directive, which is discarded during gimplification. */ case OMP_CLAUSE__CACHE_: @@ -1982,6 +1981,7 @@ convert_local_omp_clauses (tree *pclauses, struct case OMP_CLAUSE_DEFAULT: case OMP_CLAUSE_COPYIN: case OMP_CLAUSE_COLLAPSE: + case OMP_CLAUSE_TILE: case OMP_CLAUSE_UNTIED: case OMP_CLAUSE_MERGEABLE: case OMP_CLAUSE_PROC_BIND: @@ -1994,8 +1994,6 @@ convert_local_omp_clauses (tree *pclauses, struct case OMP_CLAUSE_AUTO: break; - /* OpenACC tile clauses are discarded during gimplification. */ - case OMP_CLAUSE_TILE: /* The following clause belongs to the OpenACC cache directive, which is discarded during gimplification. */ case OMP_CLAUSE__CACHE_: Index: gimplify.c =================================================================== --- gimplify.c (revision 241809) +++ gimplify.c (working copy) @@ -8138,20 +8138,11 @@ gimplify_scan_omp_clauses (tree *list_p, gimple_se remove = true; break; - case OMP_CLAUSE_TILE: - for (tree list = OMP_CLAUSE_TILE_LIST (c); !remove && list; - list = TREE_CHAIN (list)) - { - if (gimplify_expr (&TREE_VALUE (list), pre_p, NULL, - is_gimple_val, fb_rvalue) == GS_ERROR) - remove = true; - } - break; - case OMP_CLAUSE_NOWAIT: case OMP_CLAUSE_ORDERED: case OMP_CLAUSE_UNTIED: case OMP_CLAUSE_COLLAPSE: + case OMP_CLAUSE_TILE: case OMP_CLAUSE_AUTO: case OMP_CLAUSE_SEQ: case OMP_CLAUSE_INDEPENDENT: @@ -8927,13 +8918,7 @@ gimplify_adjust_omp_clauses (gimple_seq *pre_p, gi case OMP_CLAUSE_VECTOR: case OMP_CLAUSE_AUTO: case OMP_CLAUSE_SEQ: - break; - case OMP_CLAUSE_TILE: - /* We're not yet making use of the information provided by OpenACC - tile clauses. Discard these here, to simplify later middle end - processing. */ - remove = true; break; default: @@ -9388,10 +9373,23 @@ gimplify_omp_for (tree *expr_p, gimple_seq *pre_p) (OMP_FOR_INIT (for_stmt)) * 2); } - int collapse = 1; - c = find_omp_clause (OMP_FOR_CLAUSES (for_stmt), OMP_CLAUSE_COLLAPSE); - if (c) - collapse = tree_to_shwi (OMP_CLAUSE_COLLAPSE_EXPR (c)); + int collapse = 0; + /* Find the first of COLLAPSE or TILE. */ + for (c = OMP_FOR_CLAUSES (for_stmt); c; c = TREE_CHAIN (c)) + if (OMP_CLAUSE_CODE (c) == OMP_CLAUSE_COLLAPSE) + { + collapse = tree_to_shwi (OMP_CLAUSE_COLLAPSE_EXPR (c)); + if (collapse == 1) + /* Not really collapsing. */ + collapse = 0; + break; + } + else if (OMP_CLAUSE_CODE (c) == OMP_CLAUSE_TILE) + { + collapse = list_length (OMP_CLAUSE_TILE_LIST (c)); + break; + } + for (i = 0; i < TREE_VEC_LENGTH (OMP_FOR_INIT (for_stmt)); i++) { t = TREE_VEC_ELT (OMP_FOR_INIT (for_stmt), i); @@ -9807,7 +9805,7 @@ gimplify_omp_for (tree *expr_p, gimple_seq *pre_p) OMP_CLAUSE_LINEAR_STEP (c2) = OMP_CLAUSE_LINEAR_STEP (c); } - if ((var != decl || collapse > 1) && orig_for_stmt == for_stmt) + if ((var != decl || collapse) && orig_for_stmt == for_stmt) { for (c = OMP_FOR_CLAUSES (for_stmt); c ; c = OMP_CLAUSE_CHAIN (c)) if (((OMP_CLAUSE_CODE (c) == OMP_CLAUSE_LASTPRIVATE @@ -9817,7 +9815,7 @@ gimplify_omp_for (tree *expr_p, gimple_seq *pre_p) && OMP_CLAUSE_LINEAR_GIMPLE_SEQ (c) == NULL)) && OMP_CLAUSE_DECL (c) == decl) { - if (is_doacross && (collapse == 1 || i >= collapse)) + if (is_doacross && (!collapse || i >= collapse)) t = var; else {