From patchwork Wed Aug 2 21:54:44 2017 Content-Type: text/plain; charset="utf-8" MIME-Version: 1.0 Content-Transfer-Encoding: 7bit X-Patchwork-Submitter: Jim Wilson X-Patchwork-Id: 109279 Delivered-To: patch@linaro.org Received: by 10.140.101.6 with SMTP id t6csp3304515qge; Wed, 2 Aug 2017 14:55:30 -0700 (PDT) X-Received: by 10.98.71.203 with SMTP id p72mr24626830pfi.106.1501710929937; Wed, 02 Aug 2017 14:55:29 -0700 (PDT) ARC-Seal: i=1; a=rsa-sha256; t=1501710929; cv=none; d=google.com; s=arc-20160816; b=O4vwNiGBw16OtKPZeK4lntO4SqPQSCKDOp1t7IP4xHtMQwvkea+AhYDWYvp430jvwc 3d07/+7R05VeAriplisoysPITfVGRFJ7kEVvy1EpnNJc3ZtPsXOSIdArx+Y+Ow6YoydP IpQS2vgKs2BiAdaD0kbmh8FHm4RKWBqqFUe/2UXo07pwvry0cV9MvccupZm1fqJuwxWH YmNbj3v8sgb+EyH+krQOOMC8uBFVDlr2TjmXcSCmK931CnHLl0iwLSu/pPB2Mgdso+G+ xTVBwglfU6T8uCDTl1UFwbkWGMv9K18FRT0E3zh0vxTOMwkUbWyLLjboPAPtK3U3G21c ogtQ== ARC-Message-Signature: i=1; a=rsa-sha256; c=relaxed/relaxed; d=google.com; s=arc-20160816; h=message-id:date:subject:cc:to:from:delivered-to:sender:list-help :list-post:list-archive:list-unsubscribe:list-id:precedence :mailing-list:dkim-signature:domainkey-signature :arc-authentication-results; bh=cQaQ2fGdjLBflQksXoiO0HWS4kXeD15IFBjUh1eMD/M=; b=b2dDa0N5nu6ufQVOZ/72kW2b92vuT+kdWpgrSyCdKvvmjOSoOQZWm/UnGx4rQmA3JY qyAzoGjjgmuzrZjHPGOL6gBmL76ueeZbL1pcm8lgioTOHhB2UrSaQaT8Ejo0NwpXqpBk ICibpEU48HXkkMGmaiFUCOKNQ4mPE+tigCmK6CiFuFap2rkX3EFFPqYy4V2F/lXDBa24 cM2IDFWd4bZKBznQo3wY0WCi4yei3WwFwuT8r5802ns8plSCEaYLCQHyDhTSRmzoKZnj c1jgU9jPNlaUeBPTyyWGVyp0fso22TXLE0BSymcXQiaGI2HbRuKvfZiv4Ya8Qb97WpHR kGGA== ARC-Authentication-Results: i=1; mx.google.com; dkim=pass header.i=@gcc.gnu.org header.b=W9ZzeGHl; spf=pass (google.com: domain of gcc-patches-return-459682-patch=linaro.org@gcc.gnu.org designates 209.132.180.131 as permitted sender) smtp.mailfrom=gcc-patches-return-459682-patch=linaro.org@gcc.gnu.org; dmarc=fail (p=NONE sp=NONE dis=NONE) header.from=linaro.org Return-Path: Received: from sourceware.org (server1.sourceware.org. [209.132.180.131]) by mx.google.com with ESMTPS id u22si21945038plk.573.2017.08.02.14.55.29 for (version=TLS1_2 cipher=ECDHE-RSA-AES128-GCM-SHA256 bits=128/128); Wed, 02 Aug 2017 14:55:29 -0700 (PDT) Received-SPF: pass (google.com: domain of gcc-patches-return-459682-patch=linaro.org@gcc.gnu.org designates 209.132.180.131 as permitted sender) client-ip=209.132.180.131; Authentication-Results: mx.google.com; dkim=pass header.i=@gcc.gnu.org header.b=W9ZzeGHl; spf=pass (google.com: domain of gcc-patches-return-459682-patch=linaro.org@gcc.gnu.org designates 209.132.180.131 as permitted sender) smtp.mailfrom=gcc-patches-return-459682-patch=linaro.org@gcc.gnu.org; dmarc=fail (p=NONE sp=NONE dis=NONE) header.from=linaro.org DomainKey-Signature: a=rsa-sha1; c=nofws; d=gcc.gnu.org; h=list-id :list-unsubscribe:list-archive:list-post:list-help:sender:from :to:cc:subject:date:message-id; q=dns; s=default; b=VHtqdRKB2TDT OcjiQXQo+WkUTvnPZXMI8fFLsnXH67FmeK6BkhrRLhXpnbUx8kbivw81NmlnfpQx Iv/hFKPlj6gHWPPf4h1DJrBA6HBh/dQsiv7TD2V6mGFjKMNEm5ts1JmAm4B+tPkp WybHD8IR18mrB8loEZqD/FvyQFPxL/s= 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 :to:cc:subject:date:message-id; s=default; bh=Kf5kx+lgtjoEh/Ekg+ m3PeLJGNI=; b=W9ZzeGHl85zrjhgpFhi7QtAQIYhcSNpdpKXyZnDZBdew0Jliva yWmTTMRAo9hKQR0ikU7I6wH7roRZatd4G6YsdViFUd2nK7pFQHdFATQa/HAKFW84 ZjN8qHUJ64gZkKS1iUWAV03gTQGr0BDmtbjXElywvG/cErvC3q3DN1Vyw= Received: (qmail 46392 invoked by alias); 2 Aug 2017 21:55:15 -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 45312 invoked by uid 89); 2 Aug 2017 21:55:14 -0000 Authentication-Results: sourceware.org; auth=none X-Virus-Found: No X-Spam-SWARE-Status: No, score=-26.4 required=5.0 tests=BAYES_00, GIT_PATCH_0, GIT_PATCH_1, GIT_PATCH_2, GIT_PATCH_3, RCVD_IN_DNSWL_NONE, RCVD_IN_SORBS_SPAM, SPF_PASS autolearn=ham version=3.3.2 spammy=Processing, H*r:201, adr, mrs X-HELO: mail-oi0-f46.google.com Received: from mail-oi0-f46.google.com (HELO mail-oi0-f46.google.com) (209.85.218.46) by sourceware.org (qpsmtpd/0.93/v0.84-503-g423c35a) with ESMTP; Wed, 02 Aug 2017 21:55:10 +0000 Received: by mail-oi0-f46.google.com with SMTP id g131so57137983oic.3 for ; Wed, 02 Aug 2017 14:55:10 -0700 (PDT) X-Google-DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=1e100.net; s=20161025; h=x-gm-message-state:from:to:cc:subject:date:message-id; bh=cQaQ2fGdjLBflQksXoiO0HWS4kXeD15IFBjUh1eMD/M=; b=hPYGBwfMU9jkRhftmC8UJmARLiDtMRewEOT+8R8rYnE1nsIufhoUX8muSvKD4RjBOO rQ36caLdUysTUHhq050WxA0X77Zj05RbsZScoZN3amp7otf1xyU84LtDIaepqlLTReGD GmB/rGJXoR4Fy7+bGAuPiPULHMuHoaC7SW145eztTBG/cfw3UWpg4grcHXlyQirtbLaO 2m8p1lSGryYRweGjS4ELDLeQY2p/+FGqGiEJ2bwrLVtBpy1KYmDSv6k3/WrIPQShI9xI BGzHODmOU3rF4nrGDYjX2Jflx8YO6qs2d0J3WRTUPzZiYx3A/9Floz6bo7fLW8FjiO1C KQBg== X-Gm-Message-State: AIVw111Pq+dfQPdFNsIHgqeDIBf+6nhdP+9D95jb4h5iqfvflHACXtRm WCuUZrPOTu/loboZLTZULw== X-Received: by 10.202.97.214 with SMTP id v205mr20230911oib.139.1501710907849; Wed, 02 Aug 2017 14:55:07 -0700 (PDT) Received: from weathertop.attlocal.net ([2602:306:80a3:c890:201:73ff:fe02:1650]) by smtp.gmail.com with ESMTPSA id o11sm3879827oig.23.2017.08.02.14.55.05 (version=TLS1_2 cipher=ECDHE-RSA-AES128-SHA bits=128/128); Wed, 02 Aug 2017 14:55:06 -0700 (PDT) From: Jim Wilson To: gcc-patches@gcc.gnu.org Cc: Jim Wilson Subject: [PATCH, AArch64] Add falkor pipeline description. Date: Wed, 2 Aug 2017 14:54:44 -0700 Message-Id: <1501710884-7378-1-git-send-email-jim.wilson@linaro.org> This adds a pipeline description for the Qualcomm Falkor core. This was tested with a bootstrap and make check. There were no regressions. This gives about 0.5% performance gain on SPEC CPU2006 on our internal tree, which has a few other patches that aren't in the FSF tree yet. OK? Jim gcc/ * config/aarch64/aarch64-cores.def (falkor): Use falkor pipeline. (qdf24xx): Likewise. * config/aarch64/aarch64.md: Include falkor.md. * config/aarch64/falkor.md: New. --- gcc/config/aarch64/aarch64-cores.def | 4 +- gcc/config/aarch64/aarch64.md | 1 + gcc/config/aarch64/falkor.md | 681 +++++++++++++++++++++++++++++++++++ 3 files changed, 684 insertions(+), 2 deletions(-) create mode 100644 gcc/config/aarch64/falkor.md -- 2.7.4 diff --git a/gcc/config/aarch64/aarch64-cores.def b/gcc/config/aarch64/aarch64-cores.def index b8d0ba6..1089332 100644 --- a/gcc/config/aarch64/aarch64-cores.def +++ b/gcc/config/aarch64/aarch64-cores.def @@ -65,8 +65,8 @@ AARCH64_CORE("thunderxt83", thunderxt83, thunderx, 8A, AARCH64_FL_FOR_ARCH AARCH64_CORE("xgene1", xgene1, xgene1, 8A, AARCH64_FL_FOR_ARCH8, xgene1, 0x50, 0x000, -1) /* Qualcomm ('Q') cores. */ -AARCH64_CORE("falkor", falkor, cortexa57, 8A, AARCH64_FL_FOR_ARCH8 | AARCH64_FL_CRC | AARCH64_FL_CRYPTO | AARCH64_FL_RDMA, qdf24xx, 0x51, 0xC00, -1) -AARCH64_CORE("qdf24xx", qdf24xx, cortexa57, 8A, AARCH64_FL_FOR_ARCH8 | AARCH64_FL_CRC | AARCH64_FL_CRYPTO | AARCH64_FL_RDMA, qdf24xx, 0x51, 0xC00, -1) +AARCH64_CORE("falkor", falkor, falkor, 8A, AARCH64_FL_FOR_ARCH8 | AARCH64_FL_CRC | AARCH64_FL_CRYPTO | AARCH64_FL_RDMA, qdf24xx, 0x51, 0xC00, -1) +AARCH64_CORE("qdf24xx", qdf24xx, falkor, 8A, AARCH64_FL_FOR_ARCH8 | AARCH64_FL_CRC | AARCH64_FL_CRYPTO | AARCH64_FL_RDMA, qdf24xx, 0x51, 0xC00, -1) /* Samsung ('S') cores. */ AARCH64_CORE("exynos-m1", exynosm1, exynosm1, 8A, AARCH64_FL_FOR_ARCH8 | AARCH64_FL_CRC | AARCH64_FL_CRYPTO, exynosm1, 0x53, 0x001, -1) diff --git a/gcc/config/aarch64/aarch64.md b/gcc/config/aarch64/aarch64.md index fc79947..51061b2 100644 --- a/gcc/config/aarch64/aarch64.md +++ b/gcc/config/aarch64/aarch64.md @@ -231,6 +231,7 @@ (include "../arm/cortex-a53.md") (include "../arm/cortex-a57.md") (include "../arm/exynos-m1.md") +(include "falkor.md") (include "thunderx.md") (include "../arm/xgene1.md") (include "thunderx2t99.md") diff --git a/gcc/config/aarch64/falkor.md b/gcc/config/aarch64/falkor.md new file mode 100644 index 0000000..b422ab3 --- /dev/null +++ b/gcc/config/aarch64/falkor.md @@ -0,0 +1,681 @@ +;; Falkor pipeline description +;; Copyright (C) 2017 Free Software Foundation, Inc. +;; +;; This file is part of GCC. +;; +;; GCC is free software; you can redistribute it and/or modify it +;; under the terms of the GNU General Public License as published by +;; the Free Software Foundation; either version 3, or (at your option) +;; any later version. +;; +;; GCC is distributed in the hope that it will be useful, but +;; WITHOUT ANY WARRANTY; without even the implied warranty of +;; MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU +;; General Public License for more details. +;; +;; You should have received a copy of the GNU General Public License +;; along with GCC; see the file COPYING3. If not see +;; . + +(define_automaton "falkor") + +;; Complex int instructions (e.g. multiply and divide) execute in the X +;; pipeline. Simple int instructions execute in the X, Y, and Z pipelines. + +(define_cpu_unit "falkor_x" "falkor") +(define_cpu_unit "falkor_y" "falkor") +(define_cpu_unit "falkor_z" "falkor") + +;; Branches execute in the B pipeline or in one of the int pipelines depending +;; on how complex it is. Simple int insns (like movz) can also execute here. + +(define_cpu_unit "falkor_b" "falkor") + +;; Vector and FP insns execute in the VX and VY pipelines. + +(define_automaton "falkor_vfp") + +(define_cpu_unit "falkor_vx" "falkor_vfp") +(define_cpu_unit "falkor_vy" "falkor_vfp") + +;; Loads execute in the LD pipeline. +;; Stores execute in the ST, SD, and VSD pipelines, for address, data, and +;; vector data. + +(define_automaton "falkor_mem") + +(define_cpu_unit "falkor_ld" "falkor_mem") +(define_cpu_unit "falkor_st" "falkor_mem") +(define_cpu_unit "falkor_sd" "falkor_mem") +(define_cpu_unit "falkor_vsd" "falkor_mem") + +;; The GTOV and VTOG pipelines are for general to vector reg moves, and vice +;; versa. + +(define_cpu_unit "falkor_gtov" "falkor") +(define_cpu_unit "falkor_vtog" "falkor") + +;; Common reservation combinations. + +(define_reservation "falkor_vxvy" "falkor_vx|falkor_vy") +(define_reservation "falkor_zb" "falkor_z|falkor_b") +(define_reservation "falkor_xyz" "falkor_x|falkor_y|falkor_z") +(define_reservation "falkor_xyzb" "falkor_x|falkor_y|falkor_z|falkor_b") + +;; SIMD Floating-Point Instructions + +(define_insn_reservation "falkor_afp_1_vxvy" 1 + (and (eq_attr "tune" "falkor") + (eq_attr "type" "neon_fp_neg_s,neon_fp_neg_d,neon_fp_abs_s,neon_fp_abs_d")) + "falkor_vxvy") + +(define_insn_reservation "falkor_afp_1_vxvy_vxvy" 1 + (and (eq_attr "tune" "falkor") + (eq_attr "type" "neon_fp_neg_s_q,neon_fp_neg_d_q,neon_fp_abs_s_q,neon_fp_abs_d_q")) + "falkor_vxvy+falkor_vxvy") + +(define_insn_reservation "falkor_afp_2_vxvy" 2 + (and (eq_attr "tune" "falkor") + (eq_attr "type" "neon_fp_minmax_s,neon_fp_minmax_d,neon_fp_reduc_minmax_s,neon_fp_reduc_minmax_d,neon_fp_compare_s,neon_fp_compare_d,neon_fp_round_s,neon_fp_round_d")) + "falkor_vxvy") + +(define_insn_reservation "falkor_afp_2_vxvy_vxvy" 2 + (and (eq_attr "tune" "falkor") + (eq_attr "type" "neon_fp_minmax_s_q,neon_fp_minmax_d_q,neon_fp_compare_s_q,neon_fp_compare_d_q,neon_fp_round_s_q,neon_fp_round_d_q")) + "falkor_vxvy+falkor_vxvy") + +(define_insn_reservation "falkor_afp_3_vxvy" 3 + (and (eq_attr "tune" "falkor") + (eq_attr "type" "neon_fp_reduc_minmax_s_q,neon_fp_reduc_minmax_d_q,neon_fp_abd_s,neon_fp_abd_d,neon_fp_addsub_s,neon_fp_addsub_d,neon_fp_reduc_add_s,neon_fp_reduc_add_d")) + "falkor_vxvy") + +(define_insn_reservation "falkor_afp_3_vxvy_vxvy" 3 + (and (eq_attr "tune" "falkor") + (eq_attr "type" "neon_fp_abd_s_q,neon_fp_abd_d_q,neon_fp_addsub_s_q,neon_fp_addsub_d_q,neon_fp_reduc_add_s_q,neon_fp_reduc_add_d_q")) + "falkor_vxvy+falkor_vxvy") + +(define_insn_reservation "falkor_afp_4_vxvy" 4 + (and (eq_attr "tune" "falkor") + (eq_attr "type" "neon_fp_to_int_s,neon_fp_to_int_d,neon_int_to_fp_s,neon_int_to_fp_d,neon_fp_cvt_widen_h,neon_fp_cvt_widen_s")) + "falkor_vxvy") + +(define_insn_reservation "falkor_afp_4_vxvy_vxvy" 4 + (and (eq_attr "tune" "falkor") + (eq_attr "type" "neon_fp_to_int_s_q,neon_fp_to_int_d_q,neon_int_to_fp_s_q,neon_int_to_fp_d_q")) + "falkor_vxvy+falkor_vxvy") + +(define_insn_reservation "falkor_afp_5_vxvy_mul" 5 + (and (eq_attr "tune" "falkor") + (eq_attr "type" "neon_fp_mul_s,neon_fp_mul_s_scalar")) + "falkor_vxvy") + +(define_insn_reservation "falkor_afp_5_vxvy_mla" 5 + (and (eq_attr "tune" "falkor") + (eq_attr "type" "neon_fp_mla_s,neon_fp_mla_s_scalar")) + "falkor_vxvy") + +(define_insn_reservation "falkor_afp_5_vxvy_vxvy_mul" 5 + (and (eq_attr "tune" "falkor") + (eq_attr "type" "neon_fp_mul_s_q,neon_fp_mul_s_scalar_q")) + "falkor_vxvy") + +(define_insn_reservation "falkor_afp_5_vxvy_vxvy_mla" 5 + (and (eq_attr "tune" "falkor") + (eq_attr "type" "neon_fp_mla_s_q,neon_fp_mla_s_scalar_q")) + "falkor_vxvy") + +(define_insn_reservation "falkor_afp_6_vxvy_mul" 6 + (and (eq_attr "tune" "falkor") + (eq_attr "type" "neon_fp_mul_d")) + "falkor_vxvy") + +(define_insn_reservation "falkor_afp_6_vxvy_mla" 6 + (and (eq_attr "tune" "falkor") + (eq_attr "type" "neon_fp_mla_d")) + "falkor_vxvy") + +(define_insn_reservation "falkor_afp_6_vxvy_vxvy_mul" 6 + (and (eq_attr "tune" "falkor") + (eq_attr "type" "neon_fp_mul_d_q,neon_fp_mul_d_scalar_q")) + "falkor_vxvy+falkor_vxvy") + +(define_insn_reservation "falkor_afp_6_vxvy_vxvy_mla" 6 + (and (eq_attr "tune" "falkor") + (eq_attr "type" "neon_fp_mla_d_q,neon_fp_mla_d_scalar_q")) + "falkor_vxvy+falkor_vxvy") + +(define_insn_reservation "falkor_afp_4_vxvy_vxvy_vxvy" 4 + (and (eq_attr "tune" "falkor") + (eq_attr "type" "neon_fp_cvt_narrow_s_q,neon_fp_cvt_narrow_d_q")) + "falkor_vxvy+falkor_vxvy,falkor_vxvy") + +(define_insn_reservation "falkor_afp_6_vx_vy" 6 + (and (eq_attr "tune" "falkor") + (eq_attr "type" "neon_fp_div_s")) + "falkor_vx+falkor_vy") + +(define_insn_reservation "falkor_afp_11_vx_vy" 11 + (and (eq_attr "tune" "falkor") + (eq_attr "type" "neon_fp_div_d")) + "falkor_vx+falkor_vy") + +(define_insn_reservation "falkor_afp_6_vx_vy_vx_vy" 6 + (and (eq_attr "tune" "falkor") + (eq_attr "type" "neon_fp_div_s_q")) + "(falkor_vx+falkor_vy),(falkor_vx+falkor_vy)") + +(define_insn_reservation "falkor_afp_11_vx_vy_vx_vy" 11 + (and (eq_attr "tune" "falkor") + (eq_attr "type" "neon_fp_div_d_q")) + "(falkor_vx+falkor_vy),(falkor_vx+falkor_vy)") + +(define_insn_reservation "falkor_afp_12_vx_vy" 12 + (and (eq_attr "tune" "falkor") + (eq_attr "type" "neon_fp_sqrt_s")) + "falkor_vx+falkor_vy") + +(define_insn_reservation "falkor_afp_22_vx_vy" 22 + (and (eq_attr "tune" "falkor") + (eq_attr "type" "neon_fp_sqrt_d")) + "falkor_vx+falkor_vy") + +(define_insn_reservation "falkor_afp_12_vx_vy_vx_vy" 12 + (and (eq_attr "tune" "falkor") + (eq_attr "type" "neon_fp_sqrt_s_q")) + "(falkor_vx+falkor_vy),(falkor_vx+falkor_vy)") + +(define_insn_reservation "falkor_afp_22_vx_vy_vx_vy" 22 + (and (eq_attr "tune" "falkor") + (eq_attr "type" "neon_fp_sqrt_d_q")) + "(falkor_vx+falkor_vy),(falkor_vx+falkor_vy)") + +;; SIMD Integer Instructions + +(define_insn_reservation "falkor_ai_1_vxvy" 1 + (and (eq_attr "tune" "falkor") + (eq_attr "type" "neon_add,neon_reduc_add,neon_logic,neon_neg,neon_sub")) + "falkor_vxvy") + +(define_insn_reservation "falkor_ai_1_vxvy_vxvy" 1 + (and (eq_attr "tune" "falkor") + (eq_attr "type" "neon_shift_imm_long,neon_add_q,neon_reduc_add_q,neon_logic_q,neon_neg_q,neon_sub_q")) + "falkor_vxvy+falkor_vxvy") + +(define_insn_reservation "falkor_ai_2_vxvy" 2 + (and (eq_attr "tune" "falkor") + (eq_attr "type" "neon_add_long,neon_sub_long,neon_add_halve,neon_sub_halve,neon_shift_imm,neon_shift_reg,neon_minmax,neon_abs,neon_compare,neon_compare_zero,neon_tst")) + "falkor_vxvy") + +(define_insn_reservation "falkor_ai_2_vxvy_vxvy" 2 + (and (eq_attr "tune" "falkor") + (eq_attr "type" "neon_add_halve_q,neon_sub_halve_q,neon_shift_imm_q,neon_shift_reg_q,neon_minmax_q,neon_abs_q,neon_compare_q,neon_compare_zero_q,neon_tst_q,neon_reduc_add_long")) + "falkor_vxvy+falkor_vxvy") + +(define_insn_reservation "falkor_ai_3_vxvy" 3 + (and (eq_attr "tune" "falkor") + (eq_attr "type" "neon_shift_acc,neon_reduc_add_acc,neon_abd,neon_qadd,neon_qsub,neon_qabs,neon_qneg,neon_sat_shift_imm,neon_sat_shift_imm_narrow_q,neon_sat_shift_reg,neon_reduc_minmax")) + "falkor_vxvy") + +(define_insn_reservation "falkor_ai_4_vxvy" 4 + (and (eq_attr "tune" "falkor") + (eq_attr "type" "neon_reduc_minmax_q")) + "falkor_vxvy") + +(define_insn_reservation "falkor_ai_3_vxvy_vxvy" 3 + (and (eq_attr "tune" "falkor") + (eq_attr "type" "neon_shift_acc_q,neon_reduc_add_acc_q,neon_abd_q,neon_abd_long,neon_qadd_q,neon_qsub_q,neon_qabs_q,neon_qneg_q,neon_sat_shift_imm_q,neon_sat_shift_reg_q")) + "falkor_vxvy+falkor_vxvy") + +(define_insn_reservation "falkor_ai_4_vxvy_mul" 4 + (and (eq_attr "tune" "falkor") + (eq_attr "type" "neon_mul_b,neon_mul_h,neon_mul_s,neon_mul_h_scalar,neon_mul_s_scalar,neon_sat_mul_b,neon_sat_mul_h,neon_sat_mul_s,neon_sat_mul_h_scalar,neon_sat_mul_s_scalar")) + "falkor_vxvy") + +(define_insn_reservation "falkor_ai_4_vxvy_mla" 4 + (and (eq_attr "tune" "falkor") + (eq_attr "type" "neon_mla_b,neon_mla_h,neon_mla_s,neon_mla_h_scalar,neon_mla_s_scalar")) + "falkor_vxvy") + +(define_insn_reservation "falkor_ai_4_vxvy_vxvy_mul" 4 + (and (eq_attr "tune" "falkor") + (eq_attr "type" "neon_mul_b_q,neon_mul_h_q,neon_mul_s_q,neon_mul_h_scalar_q,neon_mul_s_scalar_q,neon_sat_mul_b_q,neon_sat_mul_h_q,neon_sat_mul_s_q,neon_mul_b_long,neon_mul_h_long,neon_mul_s_long,neon_mul_d_long,neon_mul_h_scalar_long,neon_mul_s_scalar_long,neon_sat_mul_b_long,neon_sat_mul_h_long,neon_sat_mul_s_long,neon_sat_mul_h_scalar_q,neon_sat_mul_s_scalar_q,neon_sat_mul_h_scalar_long,neon_sat_mul_s_scalar_long")) + "falkor_vxvy+falkor_vxvy") + +(define_insn_reservation "falkor_ai_4_vxvy_vxvy_mla" 4 + (and (eq_attr "tune" "falkor") + (eq_attr "type" "neon_mla_b_q,neon_mla_h_q,neon_mla_s_q,neon_mla_h_scalar_q,neon_mla_s_scalar_q,neon_mla_b_long,neon_mla_h_long,neon_mla_s_long,neon_mla_h_scalar_long,neon_mla_s_scalar_long,neon_sat_mla_b_long,neon_sat_mla_h_long,neon_sat_mla_s_long,neon_sat_mla_h_scalar_long,neon_sat_mla_s_scalar_long")) + "falkor_vxvy+falkor_vxvy") + +(define_insn_reservation "falkor_ai_4_vxvy_vxvy" 4 + (and (eq_attr "tune" "falkor") + (eq_attr "type" "neon_add_halve_narrow_q,neon_sub_halve_narrow_q,neon_arith_acc")) + "falkor_vxvy+falkor_vxvy") + +(define_insn_reservation "falkor_2_ai_vxvy_vxvy_vxvy_vxvy" 2 + (and (eq_attr "tune" "falkor") + (eq_attr "type" "neon_add_widen,neon_sub_widen")) + "(falkor_vxvy+falkor_vxvy),(falkor_vxvy+falkor_vxvy)") + +(define_insn_reservation "falkor_4_ai_vxvy_vxvy_vxvy_vxvy" 4 + (and (eq_attr "tune" "falkor") + (eq_attr "type" "neon_arith_acc_q")) + "(falkor_vxvy+falkor_vxvy),(falkor_vxvy+falkor_vxvy)") + +;; SIMD Load Instructions + +(define_insn_reservation "falkor_ald_4_ld" 4 + (and (eq_attr "tune" "falkor") + (eq_attr "type" "neon_load1_1reg,neon_load1_1reg_q,neon_load1_all_lanes,neon_load2_one_lane")) + "falkor_ld") + +(define_insn_reservation "falkor_ald_4_ld_none" 4 + (and (eq_attr "tune" "falkor") + (eq_attr "type" "neon_load1_2reg,neon_load2_2reg,neon_load2_all_lanes")) + "falkor_ld") + +(define_insn_reservation "falkor_ald_4_ld_ld" 4 + (and (eq_attr "tune" "falkor") + (eq_attr "type" "neon_load1_2reg_q,neon_load2_2reg_q,neon_load2_all_lanes_q,neon_load3_one_lane,neon_load4_one_lane,neon_ldp,neon_ldp_q")) + "falkor_ld,falkor_ld") + +(define_insn_reservation "falkor_ald_4_ld_ld_none" 4 + (and (eq_attr "tune" "falkor") + (eq_attr "type" "neon_load1_3reg,neon_load3_3reg,neon_load3_all_lanes")) + "falkor_ld,falkor_ld") + +(define_insn_reservation "falkor_ald_4_ld_ld_ld" 4 + (and (eq_attr "tune" "falkor") + (eq_attr "type" "neon_load1_3reg_q,neon_load3_3reg_q,neon_load3_all_lanes_q")) + "falkor_ld,falkor_ld,falkor_ld") + +(define_insn_reservation "falkor_ald_4_ld_ld_none_none" 4 + (and (eq_attr "tune" "falkor") + (eq_attr "type" "neon_load1_4reg,neon_load4_4reg")) + "falkor_ld,falkor_ld") + +(define_insn_reservation "falkor_ald_4_ld_ld_ld_ld" 4 + (and (eq_attr "tune" "falkor") + (eq_attr "type" "neon_load1_4reg_q,neon_load4_4reg_q,neon_load4_all_lanes,neon_load4_all_lanes_q")) + "falkor_ld,falkor_ld,falkor_ld,falkor_ld") + +;; Arithmetic and Logical Instructions + +(define_insn_reservation "falkor_alu_1_xyz" 1 + (and (eq_attr "tune" "falkor") + (eq_attr "type" "alus_sreg,alus_imm,alus_shift_imm,csel,adc_reg,alu_imm,alu_sreg,alu_shift_imm,alu_ext,alus_ext,logic_imm,logic_reg,logic_shift_imm,logics_imm,logics_reg,logics_shift_imm,mov_reg")) + "falkor_xyz") + +;; SIMD Miscellaneous Instructions + +;; No separate type for ins and dup. But this is correct for both. + +(define_insn_reservation "falkor_am_3_gtov" 3 + (and (eq_attr "tune" "falkor") + (eq_attr "type" "neon_from_gp")) + "falkor_gtov") + +;; No separate type for ins and dup. Assuming dup is more common. Ins is +;; gtov+vxvy and latency of 4. + +(define_insn_reservation "falkor_am_3_gtov_gtov" 3 + (and (eq_attr "tune" "falkor") + (eq_attr "type" "neon_from_gp_q")) + "falkor_gtov,falkor_gtov") + +;; neon_to_gp_q is used for 32-bit ARM instructions that move 64-bits of data +;; so no use needed here. + +(define_insn_reservation "falkor_am_3_vtog" 3 + (and (eq_attr "tune" "falkor") + (eq_attr "type" "neon_to_gp")) + "falkor_vtog") + +(define_insn_reservation "falkor_am_1_vxvy" 1 + (and (eq_attr "tune" "falkor") + (eq_attr "type" "neon_bsl,neon_dup,neon_ext,neon_ins,neon_ins_q,neon_move,neon_rev,neon_tbl1,neon_permute,neon_shift_imm_narrow_q")) + "falkor_vxvy") + +(define_insn_reservation "falkor_am_1_vxvy_vxvy" 1 + (and (eq_attr "tune" "falkor") + (eq_attr "type" "neon_bsl_q,neon_dup_q,neon_ext_q,neon_move_q,neon_rev_q,neon_tbl1_q,neon_permute_q")) + "falkor_vxvy+falkor_vxvy") + +(define_insn_reservation "falkor_am_2_vxvy" 2 + (and (eq_attr "tune" "falkor") + (eq_attr "type" "neon_cls,neon_cnt,neon_rbit")) + "falkor_vxvy") + +(define_insn_reservation "falkor_am_4_vxvy_vxvy" 4 + (and (eq_attr "tune" "falkor") + (eq_attr "type" "neon_cls_q,neon_cnt_q,neon_rbit_q,neon_tbl2")) + "falkor_vxvy+falkor_vxvy") + +(define_insn_reservation "falkor_am_3_vxvy" 3 + (and (eq_attr "tune" "falkor") + (eq_attr "type" "neon_fp_recpe_s,neon_fp_recpe_d,neon_fp_rsqrte_s,neon_fp_rsqrte_d,neon_fp_recpx_s,neon_fp_recpx_d")) + "falkor_vxvy") + +(define_insn_reservation "falkor_am_3_vxvy_vxvy" 3 + (and (eq_attr "tune" "falkor") + (eq_attr "type" "neon_fp_recpe_s_q,neon_fp_recpe_d_q,neon_fp_rsqrte_s_q,neon_fp_rsqrte_d_q")) + "falkor_vxvy+falkor_vxvy") + +(define_insn_reservation "falkor_am_5_vxvy" 5 + (and (eq_attr "tune" "falkor") + (eq_attr "type" "neon_fp_recps_s")) + "falkor_vxvy") + +(define_insn_reservation "falkor_am_5_vxvy_vxvy" 5 + (and (eq_attr "tune" "falkor") + (eq_attr "type" "neon_fp_recps_s_q")) + "falkor_vxvy+falkor_vxvy") + +(define_insn_reservation "falkor_am_6_vxvy" 6 + (and (eq_attr "tune" "falkor") + (eq_attr "type" "neon_fp_recps_d,neon_fp_rsqrts_d")) + "falkor_vxvy") + +(define_insn_reservation "falkor_am_6_vxvy_vxvy" 6 + (and (eq_attr "tune" "falkor") + (eq_attr "type" "neon_fp_recps_d_q,neon_fp_rsqrts_d_q")) + "falkor_vxvy+falkor_vxvy") + +(define_insn_reservation "falkor_am_5_vxvy_vxvy_vxvy" 5 + (and (eq_attr "tune" "falkor") + (eq_attr "type" "neon_tbl2_q,neon_tbl3")) + "(falkor_vxvy+falkor_vxvy),falkor_vxvy") + +(define_insn_reservation "falkor_am_6_vxvy_vxvy_vxvy_vxvy" 6 + (and (eq_attr "tune" "falkor") + (eq_attr "type" "neon_tbl3_q,neon_tbl4")) + "(falkor_vxvy+falkor_vxvy),(falkor_vxvy+falkor_vxvy)") + +(define_insn_reservation "falkor_am_7_vxvy_vxvy_vxvy_vxvy_vxvy" 7 + (and (eq_attr "tune" "falkor") + (eq_attr "type" "neon_tbl4_q")) + "(falkor_vxvy+falkor_vxvy),(falkor_vxvy+falkor_vxvy),falkor_vxvy") + +;; SIMD Store Instructions + +;; ??? stp is neon_store1_2reg in aarch64.md, but neon_stp in aarch64-simd.md. +;; Similarly with ldp. + +(define_insn_reservation "falkor_ast_st_vsd" 0 + (and (eq_attr "tune" "falkor") + (eq_attr "type" "neon_store1_1reg,neon_store1_1reg_q,neon_store1_one_lane,neon_store1_one_lane_q,neon_store1_2reg,neon_store2_2reg,neon_store2_one_lane,neon_store2_one_lane_q,neon_stp")) + "falkor_st+falkor_vsd") + +(define_insn_reservation "falkor_as_0_st_vsd_st_vsd" 0 + (and (eq_attr "tune" "falkor") + (eq_attr "type" "neon_store1_2reg_q,neon_store1_3reg,neon_store1_4reg,neon_store2_2reg_q,neon_store3_3reg,neon_store4_4reg,neon_store3_one_lane,neon_store3_one_lane_q,neon_store4_one_lane,neon_store4_one_lane_q,neon_stp_q")) + "(falkor_st+falkor_vsd),(falkor_st+falkor_vsd)") + +(define_insn_reservation "falkor_as_0_st_vsd_st_vsd_st_vsd" 0 + (and (eq_attr "tune" "falkor") + (eq_attr "type" "neon_store1_3reg_q,neon_store3_3reg_q")) + "(falkor_st+falkor_vsd),(falkor_st+falkor_vsd),(falkor_st+falkor_vsd)") + +(define_insn_reservation "falkor_as_0_st_vsd_st_vsd_st_vsd_st_vsd" 0 + (and (eq_attr "tune" "falkor") + (eq_attr "type" "neon_store1_4reg_q,neon_store4_4reg_q")) + "(falkor_st+falkor_vsd),(falkor_st+falkor_vsd),(falkor_st+falkor_vsd),(falkor_st+falkor_vsd)") + +;; Branch Instructions + +(define_insn_reservation "falkor_branch_0_zb" 0 + (and (eq_attr "tune" "falkor") + (eq_attr "type" "branch")) + "falkor_zb") + +(define_insn_reservation "falkor_call_0_xyzb" 0 + (and (eq_attr "tune" "falkor") + (eq_attr "type" "call")) + "falkor_xyzb") + +;; Cryptography Extensions + +(define_insn_reservation "falkor_cry_1_vxvy" 1 + (and (eq_attr "tune" "falkor") + (eq_attr "type" "crypto_sha1_fast")) + "falkor_vxvy") + +(define_insn_reservation "falkor_cry_2_vxvy" 2 + (and (eq_attr "tune" "falkor") + (eq_attr "type" "crypto_aesmc")) + "falkor_vxvy") + +(define_insn_reservation "falkor_cry_2_vxvy_vxvy" 2 + (and (eq_attr "tune" "falkor") + (eq_attr "type" "crypto_sha1_xor,crypto_sha256_fast,crypto_pmull")) + "falkor_vxvy+falkor_vxvy") + +(define_insn_reservation "falkor_cry_4_vy_vx" 4 + (and (eq_attr "tune" "falkor") + (eq_attr "type" "crypto_sha1_slow")) + "falkor_vy+falkor_vx") + +(define_insn_reservation "falkor_cry_6_vy_vx" 6 + (and (eq_attr "tune" "falkor") + (eq_attr "type" "crypto_sha256_slow")) + "falkor_vy+falkor_vx") + +(define_insn_reservation "falkor_cry_3_vxvy_vxvy" 3 + (and (eq_attr "tune" "falkor") + (eq_attr "type" "crypto_aese")) + "falkor_vxvy+falkor_vxvy") + +;; FP Load Instructions + +(define_insn_reservation "falkor_fld_4_ld" 4 + (and (eq_attr "tune" "falkor") + (eq_attr "type" "f_loads,f_loadd")) + "falkor_ld") + +;; No separate FP store section, these are found in the SIMD store section. + +(define_insn_reservation "falkor_fld_0_st_vsd" 0 + (and (eq_attr "tune" "falkor") + (eq_attr "type" "f_stores,f_stored")) + "falkor_st+falkor_vsd") + +;; FP Data Processing Instructions + +(define_insn_reservation "falkor_fpdt_0_vxvy" 0 + (and (eq_attr "tune" "falkor") + (eq_attr "type" "fcmps,fcmpd,fccmps,fccmpd")) + "falkor_vxvy") + +(define_insn_reservation "falkor_fpdt_5_vtog" 5 + (and (eq_attr "tune" "falkor") + (eq_attr "type" "f_cvtf2i")) + "falkor_vtog") + +(define_insn_reservation "falkor_fpdt_1_vxvy" 1 + (and (eq_attr "tune" "falkor") + (eq_attr "type" "ffariths,ffarithd,fcsel")) + "falkor_vxvy") + +(define_insn_reservation "falkor_fpdt_2_vxvy" 2 + (and (eq_attr "tune" "falkor") + (eq_attr "type" "f_minmaxd,f_minmaxs,f_rintd,f_rints")) + "falkor_vxvy") + +;; Scalar FP ABD is handled same as vector FP ABD. + +(define_insn_reservation "falkor_fpdt_3_vxvy" 3 + (and (eq_attr "tune" "falkor") + (eq_attr "type" "faddd,fadds")) + "falkor_vxvy") + +(define_insn_reservation "falkor_fpdt_4_vxvy" 4 + (and (eq_attr "tune" "falkor") + (eq_attr "type" "f_cvt")) + "falkor_vxvy") + +(define_insn_reservation "falkor_fpdt_5_vxvy_mul" 5 + (and (eq_attr "tune" "falkor") + (eq_attr "type" "fmuls")) + "falkor_vxvy") + +(define_insn_reservation "falkor_fpdt_5_vxvy_mla" 5 + (and (eq_attr "tune" "falkor") + (eq_attr "type" "fmacs,ffmas")) + "falkor_vxvy") + +(define_insn_reservation "falkor_fpdt_6_vxvy_mul" 6 + (and (eq_attr "tune" "falkor") + (eq_attr "type" "fmuld")) + "falkor_vxvy") + +(define_insn_reservation "falkor_fpdt_6_vxvy_mla" 6 + (and (eq_attr "tune" "falkor") + (eq_attr "type" "fmacd,ffmad")) + "falkor_vxvy") + +(define_insn_reservation "falkor_fpdt_6_vx_vy" 6 + (and (eq_attr "tune" "falkor") + (eq_attr "type" "fdivs")) + "falkor_vx+falkor_vy") + +(define_insn_reservation "falkor_fpdt_11_vx_vy" 11 + (and (eq_attr "tune" "falkor") + (eq_attr "type" "fdivd")) + "falkor_vx+falkor_vy") + +(define_insn_reservation "falkor_fpdt_12_vx_vy" 12 + (and (eq_attr "tune" "falkor") + (eq_attr "type" "fsqrts")) + "falkor_vxvy") + +(define_insn_reservation "falkor_fpdt_22_vx_vy" 22 + (and (eq_attr "tune" "falkor") + (eq_attr "type" "fsqrtd")) + "falkor_vxvy") + +;; FP Miscellaneous Instructions + +(define_insn_reservation "falkor_fpmsc_3_vtog" 3 + (and (eq_attr "tune" "falkor") + (eq_attr "type" "f_mrc")) + "falkor_vtog") + +(define_insn_reservation "falkor_fpmsc_3_gtov" 3 + (and (eq_attr "tune" "falkor") + (eq_attr "type" "f_mcr")) + "falkor_gtov") + +(define_insn_reservation "falkor_fpmsc_1_vxvy" 1 + (and (eq_attr "tune" "falkor") + (eq_attr "type" "fmov,fconsts,fconstd")) + "falkor_vxvy") + +;; No separate type for float-to-fixed conversions. Same type as +;; float-to-int conversions. They schedule the same though, so no problem. + +(define_insn_reservation "falkor_fpmsc_6_gtov" 6 + (and (eq_attr "tune" "falkor") + (eq_attr "type" "f_cvti2f")) + "falkor_gtov") + +;; Load Instructions + +(define_insn_reservation "falkor_ld_3_ld" 3 + (and (eq_attr "tune" "falkor") + (eq_attr "type" "load1,load2")) + "falkor_ld") + +;; Miscellaneous Data-Processing Instructions + +(define_insn_reservation "falkor_misc_1_xyz" 1 + (and (eq_attr "tune" "falkor") + (eq_attr "type" "bfx,bfm,extend,rotate_imm,shift_imm")) + "falkor_xyz") + +(define_insn_reservation "falkor_misc_2_x" 2 + (and (eq_attr "tune" "falkor") + (eq_attr "type" "crc")) + "falkor_x") + +(define_insn_reservation "falkor_misc_2_xyz" 2 + (and (eq_attr "tune" "falkor") + (eq_attr "type" "clz,rbit,rev")) + "falkor_xyz") + +;; Divide and Multiply Instructions + +(define_insn_reservation "falkor_muldiv_4_x_mul" 4 + (and (eq_attr "tune" "falkor") + (eq_attr "type" "mul")) + "falkor_x") + +(define_insn_reservation "falkor_muldiv_4_x_mla" 4 + (and (eq_attr "tune" "falkor") + (eq_attr "type" "mla,smlal,umlal")) + "falkor_x") + +(define_insn_reservation "falkor_muldiv_5_x_mul" 5 + (and (eq_attr "tune" "falkor") + (eq_attr "type" "smull,umull")) + "falkor_x") + +(define_insn_reservation "falkor_md_11_x_z" 11 + (and (eq_attr "tune" "falkor") + (eq_attr "type" "sdiv,udiv")) + "falkor_x+falkor_z") + +;; Move and Shift Instructions + +(define_insn_reservation "falkor_mvs_1_xyz" 1 + (and (eq_attr "tune" "falkor") + (eq_attr "type" "mov_imm,shift_reg")) + "falkor_xyz") + +(define_insn_reservation "falkor_mvs_1_xyzb" 1 + (and (eq_attr "tune" "falkor") + (eq_attr "type" "adr")) + "falkor_xyzb") + +;; Other Instructions + +;; Block is for instruction scheduling blockage insns in RTL. There are no +;; hardware instructions emitted for them, so don't use any resources. + +(define_insn_reservation "falkor_other_0_nothing" 0 + (and (eq_attr "tune" "falkor") + (eq_attr "type" "no_insn,trap,block")) + "nothing") + +(define_insn_reservation "falkor_other_2_z" 2 + (and (eq_attr "tune" "falkor") + (eq_attr "type" "mrs")) + "falkor_z") + +;; Assume multiple instructions use all pipes. + +(define_insn_reservation "falkor_extra" 1 + (and (eq_attr "tune" "falkor") + (eq_attr "type" "multiple")) + "falkor_x+falkor_y+falkor_z+falkor_b+falkor_vx+falkor_vy+falkor_ld+falkor_st+falkor_sd+falkor_vsd+falkor_gtov+falkor_vtog") + +;; Store Instructions + +;; No use of store_rel, store3, or store4 in aarch64. + +(define_insn_reservation "falkor_st_0_st_sd" 0 + (and (eq_attr "tune" "falkor") + (eq_attr "type" "store1,store2")) + "falkor_st+falkor_sd") + +;; Muliply bypasses. + +;; 1 cycle latency (0 bubble) for an integer mul or mac feeding into a mac. + +(define_bypass 1 + "falkor_ai_4_vxvy_mul,falkor_ai_4_vxvy_mla,falkor_ai_4_vxvy_vxvy_mul,falkor_ai_4_vxvy_vxvy_mla,falkor_muldiv_4_x_mul,falkor_muldiv_4_x_mla,falkor_muldiv_5_x_mul" + "falkor_ai_4_vxvy_mla,falkor_ai_4_vxvy_vxvy_mla,falkor_muldiv_4_x_mla") + +;; 3 cycle latency (2 bubbles) for an FP mul or mac feeding into a mac. + +(define_bypass 3 + "falkor_afp_5_vxvy_mul,falkor_afp_5_vxvy_mla,falkor_afp_5_vxvy_vxvy_mul,falkor_afp_5_vxvy_vxvy_mla,falkor_afp_6_vxvy_mul,falkor_afp_6_vxvy_mla,falkor_afp_6_vxvy_vxvy_mul,falkor_afp_6_vxvy_vxvy_mla,falkor_fpdt_5_vxvy_mul,falkor_fpdt_5_vxvy_mla,falkor_fpdt_6_vxvy_mul,falkor_fpdt_6_vxvy_mla" + "falkor_afp_5_vxvy_mla,falkor_afp_5_vxvy_vxvy_mla,falkor_afp_6_vxvy_mla,falkor_afp_6_vxvy_vxvy_mla,falkor_fpdt_5_vxvy_mla,falkor_fpdt_6_vxvy_mla")