From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: from mail-lj1-x232.google.com (mail-lj1-x232.google.com [IPv6:2a00:1450:4864:20::232]) by sourceware.org (Postfix) with ESMTPS id 2FC463858D28 for ; Tue, 29 Aug 2023 08:10:58 +0000 (GMT) DMARC-Filter: OpenDMARC Filter v1.4.2 sourceware.org 2FC463858D28 Authentication-Results: sourceware.org; dmarc=pass (p=none dis=none) header.from=gmail.com Authentication-Results: sourceware.org; spf=pass smtp.mailfrom=gmail.com Received: by mail-lj1-x232.google.com with SMTP id 38308e7fff4ca-2bcc331f942so48103171fa.0 for ; Tue, 29 Aug 2023 01:10:58 -0700 (PDT) DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=gmail.com; s=20221208; t=1693296656; x=1693901456; darn=gcc.gnu.org; h=content-transfer-encoding:cc:to:subject:message-id:date:from :in-reply-to:references:mime-version:from:to:cc:subject:date :message-id:reply-to; bh=8P59HUzNXSgasd8AoaJ8wcC/WtCQ/iS0qc/9yaoZhC0=; b=imdO/0oiA0oQ+TKCUkVooUxq8qNahTNHOW5Uy2Kz1i2bewhHuamGrlRMaTaTu+mHWo DrWozmC8gRdLdFCq2nJVxUmvtk6SemhRq7Y79HbUVr5mrLA2sRcgMdJMdYehJ9o+lL9L uYdbhXuJXt1ncYMDvCTNR4c6CnobhcBnxBTEh66nsJ1lYJb0rkbxZvgC2YRLVBA+vlpD aYJWRB6+1k/LLFa3KsZpx1yqZa1BsLsLAdrqr8plNsbjvkDf4KdMMgTYribRJPVmto7v rRfC9pbfwOiTUjM+KV0mVGGSKs73qAxQmnBWyPdBUaGB5bpx5+eCdNj+J0Zt/vh4xrN8 XYYA== X-Google-DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=1e100.net; s=20221208; t=1693296656; x=1693901456; h=content-transfer-encoding:cc:to:subject:message-id:date:from :in-reply-to:references:mime-version:x-gm-message-state:from:to:cc :subject:date:message-id:reply-to; bh=8P59HUzNXSgasd8AoaJ8wcC/WtCQ/iS0qc/9yaoZhC0=; b=k8pbYLiBdzKDZLpk7AAhjShiK4YkaLoCzEXE8GGYb4gnaB6MJ4wM9qFAcoXOKbsqvc 9jXrYvXs1F9frOEt0vaWYZlr3gJr7fCmhil7zEFPt6bRNE0xsQ340QRVEMDEZp+CTCAE jq0t6CWjuTskTOHJKgoj1PCwpC4HHsjUI9WrutNqngj9EhfAWRfPLSiFkQqbrZfE6ti4 YnQhxhAxBEli/PFc0qpCkUTg6GPowvOzwf9PKtKjvhk8pn+z3hKVCQ0DYX5PSjof/HBr YFERktI8C4GohIHr79Ubak6HU6lzwsYVyD+0drxYuKEABVUMqu0Vz4uha47Wqj7riy4c kNUA== X-Gm-Message-State: AOJu0YxUXNo0xmVrFznZ6V1eVMqFvSV2lHtuh3Ep0e5d4lJA5QXhzzZY 4bvvHhv4a1ujePXm85dRbAwOeYoUKfTKcvTVlGIpIscU X-Google-Smtp-Source: AGHT+IGYwbzoKpbMXdyJIMZCJcOn6RnGzHKoFq0SkHrvlJ04bGRwoWZltc6XFwNgwpofulR9JTik+R3QAS033xzgEaY= X-Received: by 2002:a05:651c:10ba:b0:2bc:dcbe:f566 with SMTP id k26-20020a05651c10ba00b002bcdcbef566mr646901ljn.11.1693296656295; Tue, 29 Aug 2023 01:10:56 -0700 (PDT) MIME-Version: 1.0 References: <4c3c9a1c-e182-30a9-342d-525adfb8cffd@gmail.com> In-Reply-To: From: Richard Biener Date: Tue, 29 Aug 2023 10:09:17 +0200 Message-ID: Subject: Re: [PATCH] [tree-optimization/110279] swap operands in reassoc to reduce cross backedge FMA To: Di Zhao OS Cc: Jeff Law , Martin Jambor , "gcc-patches@gcc.gnu.org" Content-Type: text/plain; charset="UTF-8" Content-Transfer-Encoding: quoted-printable X-Spam-Status: No, score=-1.3 required=5.0 tests=BAYES_00,DKIM_SIGNED,DKIM_VALID,DKIM_VALID_AU,DKIM_VALID_EF,FREEMAIL_FROM,KAM_SHORT,RCVD_IN_DNSWL_NONE,SPF_HELO_NONE,SPF_PASS,TXREP autolearn=ham autolearn_force=no version=3.4.6 X-Spam-Checker-Version: SpamAssassin 3.4.6 (2021-04-09) on server2.sourceware.org List-Id: On Tue, Aug 29, 2023 at 9:49=E2=80=AFAM Di Zhao OS wrote: > > Hi, > > > -----Original Message----- > > From: Richard Biener > > Sent: Tuesday, August 29, 2023 3:41 PM > > To: Jeff Law ; Martin Jambor > > Cc: Di Zhao OS ; gcc-patches@gcc.gnu.org > > Subject: Re: [PATCH] [tree-optimization/110279] swap operands in reasso= c to > > reduce cross backedge FMA > > > > On Tue, Aug 29, 2023 at 1:23=E2=80=AFAM Jeff Law via Gcc-patches > > wrote: > > > > > > > > > > > > On 8/28/23 02:17, Di Zhao OS via Gcc-patches wrote: > > > > This patch tries to fix the 2% regression in 510.parest_r on > > > > ampere1 in the tracker. (Previous discussion is here: > > > > https://gcc.gnu.org/pipermail/gcc-patches/2023-July/624893.html) > > > > > > > > 1. Add testcases for the problem. For an op list in the form of > > > > "acc =3D a * b + c * d + acc", currently reassociation doesn't > > > > Swap the operands so that more FMAs can be generated. > > > > After widening_mul the result looks like: > > > > > > > > _1 =3D .FMA(a, b, acc_0); > > > > acc_1 =3D .FMA(c, d, _1); > > > > > > > > While previously (before the "Handle FMA friendly..." patch), > > > > widening_mul's result was like: > > > > > > > > _1 =3D a * b; > > > > _2 =3D .FMA (c, d, _1); > > > > acc_1 =3D acc_0 + _2; > > > > How can we execute the multiply and the FMA in parallel? They > > depend on each other. Or is it the uarch can handle dependence > > on the add operand but only when it is with a multiplication and > > not a FMA in some better ways? (I'd doubt so much complexity) > > > > Can you explain in more detail how the uarch executes one vs. the > > other case? > > > > > > If the code fragment is in a loop, some architecture can execute > > > > the latter in parallel, so the performance can be much faster than > > > > the former. For the small testcase, the performance gap is over > > > > 10% on both ampere1 and neoverse-n1. So the point here is to avoid > > > > turning the last statement into FMA, and keep it a PLUS_EXPR as > > > > much as possible. (If we are rewriting the op list into parallel, > > > > no special treatment is needed, since the last statement after > > > > rewrite_expr_tree_parallel will be PLUS_EXPR anyway.) > > > > > > > > 2. Function result_feeds_back_from_phi_p is to check for cross > > > > backedge dependency. Added new enum fma_state to describe the > > > > state of FMA candidates. > > > > > > > > With this patch, there's a 3% improvement in 510.parest_r 1-copy > > > > run on ampere1. The compile options are: > > > > "-Ofast -mcpu=3Dampere1 -flto --param avoid-fma-max-bits=3D512". > > > > > > > > Best regards, > > > > Di Zhao > > > > > > > > ---- > > > > > > > > PR tree-optimization/110279 > > > > > > > > gcc/ChangeLog: > > > > > > > > * tree-ssa-reassoc.cc (enum fma_state): New enum to > > > > describe the state of FMA candidates for an op list. > > > > (rewrite_expr_tree_parallel): Changed boolean > > > > parameter to enum type. > > > > (result_feeds_back_from_phi_p): New function to check > > > > for cross backedge dependency. > > > > (rank_ops_for_fma): Return enum fma_state. Added new > > > > parameter. > > > > (reassociate_bb): If there's backedge dependency in an > > > > op list, swap the operands before rewrite_expr_tree. > > > > > > > > gcc/testsuite/ChangeLog: > > > > > > > > * gcc.dg/pr110279.c: New test. > > > Not a review, but more of a question -- isn't this transformation's > > > profitability uarch sensitive. ie, just because it's bad for a set o= f > > > aarch64 uarches, doesn't mean it's bad everywhere. > > > > > > And in general we shy away from trying to adjust gimple code based on > > > uarch preferences. > > > > > > It seems the right place to do this is gimple->rtl expansion. > > > > Another comment is that FMA forming has this deferring code which I > > think deals exactly with this kind of thing? CCing Martin who did this > > work based on AMD uarchs also not wanting cross-loop dependences > > on FMAs (or so). In particular I see > > > > if (fma_state.m_deferring_p > > && fma_state.m_initial_phi) > > { > > gcc_checking_assert (fma_state.m_last_result); > > if (!last_fma_candidate_feeds_initial_phi (&fma_state, > > &m_last_result_set)) > > cancel_fma_deferring (&fma_state); > > > > and I think code to avoid FMAs in other/related cases should be here > > as well, like avoid forming back-to-back FMAs. > > The changes in this patch is controlled by "param_avoid_fma_max_bits", so > I think it should only affect architectures with similar behavior. (The > parameter was added in a previous patch "Deferring FMA transformations > in tight loops", which seems to be dealing with the same issue.) That's what I said. Is the pipeline behavior properly modeled by the scheduler description? Your patch seems to not only affect loops but the FMA forming case already present does - is the case you are after not in a tight loop? In principle with a loop with enough scheduling freedom this should be a non-issue - unless the pipeline model isn't accurate and we put the FMAs too close together? Richard. > > Richard. > > > > > Jeff > > Thanks, > Di Zhao