From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (qmail 9148 invoked by alias); 8 Jun 2016 12:30:01 -0000 Mailing-List: contact gcc-patches-help@gcc.gnu.org; run by ezmlm Precedence: bulk List-Id: List-Archive: List-Post: List-Help: Sender: gcc-patches-owner@gcc.gnu.org Received: (qmail 9116 invoked by uid 89); 8 Jun 2016 12:30:01 -0000 Authentication-Results: sourceware.org; auth=none X-Virus-Found: No X-Spam-SWARE-Status: No, score=-1.9 required=5.0 tests=AWL,BAYES_00,FREEMAIL_FROM,KAM_ASCII_DIVIDERS,RCVD_IN_DNSWL_LOW,SPF_PASS autolearn=no version=3.3.2 spammy=49911, 499,11, 4999, floats X-HELO: mail-wm0-f49.google.com Received: from mail-wm0-f49.google.com (HELO mail-wm0-f49.google.com) (74.125.82.49) by sourceware.org (qpsmtpd/0.93/v0.84-503-g423c35a) with (AES128-GCM-SHA256 encrypted) ESMTPS; Wed, 08 Jun 2016 12:29:49 +0000 Received: by mail-wm0-f49.google.com with SMTP id n184so179377811wmn.1 for ; Wed, 08 Jun 2016 05:29:48 -0700 (PDT) X-Google-DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=1e100.net; s=20130820; h=x-gm-message-state:mime-version:in-reply-to:references:from:date :message-id:subject:to:cc; bh=tI0P3mzKF2PvWj+0nKtzjJ4E+dPeycvciaZz//P3O4g=; b=LBYafxGvyRW51Oj8+AGOrBn7w8qrd4p8nuBPkmJXVOxxSDJhqqV6tItCKqhn5DRF+Q jqXqd79z2EVU5xBKZABHIEU/Gg1dsw7P7qYbVIKoqzqH+eJqzLMEpPPJ5wqmVCWFV+Yq PK4K5+s7B2r5RepYC0u5kxm/mDai3f9tyZ0plLUK7fECWHvGJvSXZKIZzgXCmsfT/FPH ydPnMzRYEzZdqNasfXd0pbzcz0muuhv4GfMtB7CiHhC5DQr+ea7u0dzdfSssSYXgp+RG qb6Kn5uIbl8BKFd81JDdkCbDvQ3jNj3SZlSX7svT3WP3gKgJdn7z7ohP/8G00JfNmZ8L TD3w== X-Gm-Message-State: ALyK8tJT9BWYO6bzzJc3stq42b8OaS05k7NYSi84MTKStwO6NcGCL9o7fHGVg1PIJcURRSMrtMuS0jysS44MHw== X-Received: by 10.194.2.225 with SMTP id 1mr4253224wjx.72.1465388985742; Wed, 08 Jun 2016 05:29:45 -0700 (PDT) MIME-Version: 1.0 Received: by 10.194.87.34 with HTTP; Wed, 8 Jun 2016 05:29:45 -0700 (PDT) In-Reply-To: <1339553936.18291.15.camel@gnopaine> References: <1339553936.18291.15.camel@gnopaine> From: Richard Biener Date: Wed, 08 Jun 2016 12:30:00 -0000 Message-ID: Subject: Re: [PATCH, RFC] First cut at using vec_construct for strided loads To: "William J. Schmidt" , Uros Bizjak Cc: GCC Patches , Richard Biener , Peter Bergner Content-Type: text/plain; charset=UTF-8 X-IsSubscribed: yes X-SW-Source: 2016-06/txt/msg00584.txt.bz2 On Wed, Jun 13, 2012 at 4:18 AM, William J. Schmidt wrote: > This patch is a follow-up to the discussion generated by > http://gcc.gnu.org/ml/gcc-patches/2012-06/msg00546.html. I've added > vec_construct to the cost model for use in vect_model_load_cost, and > implemented a cost calculation that makes sense to me for PowerPC. I'm > less certain about the default, i386, and spu implementations. I took a > guess at i386 from the discussions we had, and used the same calculation > for the default and for spu. I'm hoping you or others can fill in the > blanks if I guessed badly. > > The i386 cost for vec_construct is different from all the others, which > are parameterized for each processor description. This should probably > be parameterized in some way as well, but thought you'd know better than > I how that should be. Perhaps instead of > > elements / 2 + 1 > > it should be > > (elements / 2) * X + Y > > where X and Y are taken from the processor description, and represent > the cost of a merge and a permute, respectively. Let me know what you > think. Just trying to understand how you arrived at the above formulas in investigating strangely low cost for v16qi construction of 9. If we pairwise reduce elements with a cost of 1 then we arrive at a cost of elements - 1, that's what you'd get with not accounting an initial move of element zero into a vector and then inserting each other element into that with elements - 1 inserts. This also matches up with code-generation on x86_64 for vT foo (T a, T b, ...) { return (vT) {a, b, ... }; } for any vector / element type combination I tried. Thus the patch below. I'll bootstrap / test that on x86_64-linux and I'm leaving other targets to target maintainers. Ok for the i386 parts? Thanks, Richard. 2016-06-08 Richard Biener * targhooks.c (default_builtin_vectorization_cost): Adjust vec_construct cost. * config/i386/i386.c (ix86_builtin_vectorization_cost): Likewise. Index: gcc/targhooks.c =================================================================== --- gcc/targhooks.c (revision 237196) +++ gcc/targhooks.c (working copy) @@ -589,8 +589,7 @@ default_builtin_vectorization_cost (enum return 3; case vec_construct: - elements = TYPE_VECTOR_SUBPARTS (vectype); - return elements / 2 + 1; + return TYPE_VECTOR_SUBPARTS (vectype) - 1; default: gcc_unreachable (); Index: gcc/config/i386/i386.c =================================================================== --- gcc/config/i386/i386.c (revision 237196) +++ gcc/config/i386/i386.c (working copy) @@ -49503,8 +49520,6 @@ static int ix86_builtin_vectorization_cost (enum vect_cost_for_stmt type_of_cost, tree vectype, int) { - unsigned elements; - switch (type_of_cost) { case scalar_stmt: @@ -49546,8 +49561,7 @@ ix86_builtin_vectorization_cost (enum ve return ix86_cost->vec_stmt_cost; case vec_construct: - elements = TYPE_VECTOR_SUBPARTS (vectype); - return ix86_cost->vec_stmt_cost * (elements / 2 + 1); + return ix86_cost->vec_stmt_cost * (TYPE_VECTOR_SUBPARTS (vectype) - 1); default: gcc_unreachable (); > Thanks, > Bill > > > 2012-06-12 Bill Schmidt > > * targhooks.c (default_builtin_vectorized_conversion): Handle > vec_construct, using vectype to base cost on subparts. > * target.h (enum vect_cost_for_stmt): Add vec_construct. > * tree-vect-stmts.c (vect_model_load_cost): Use vec_construct > instead of scalar_to-vec. > * config/spu/spu.c (spu_builtin_vectorization_cost): Handle > vec_construct in same way as default for now. > * config/i386/i386.c (ix86_builtin_vectorization_cost): Likewise. > * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost): > Handle vec_construct, including special case for 32-bit loads. > > > Index: gcc/targhooks.c > =================================================================== > --- gcc/targhooks.c (revision 188482) > +++ gcc/targhooks.c (working copy) > @@ -499,9 +499,11 @@ default_builtin_vectorized_conversion (unsigned in > > int > default_builtin_vectorization_cost (enum vect_cost_for_stmt type_of_cost, > - tree vectype ATTRIBUTE_UNUSED, > + tree vectype, > int misalign ATTRIBUTE_UNUSED) > { > + unsigned elements; > + > switch (type_of_cost) > { > case scalar_stmt: > @@ -524,6 +526,11 @@ default_builtin_vectorization_cost (enum vect_cost > case cond_branch_taken: > return 3; > > + case vec_construct: > + elements = TYPE_VECTOR_SUBPARTS (vectype); > + gcc_assert (elements > 1); > + return elements / 2 + 1; > + > default: > gcc_unreachable (); > } > Index: gcc/target.h > =================================================================== > --- gcc/target.h (revision 188482) > +++ gcc/target.h (working copy) > @@ -146,7 +146,8 @@ enum vect_cost_for_stmt > cond_branch_not_taken, > cond_branch_taken, > vec_perm, > - vec_promote_demote > + vec_promote_demote, > + vec_construct > }; > > /* The target structure. This holds all the backend hooks. */ > Index: gcc/tree-vect-stmts.c > =================================================================== > --- gcc/tree-vect-stmts.c (revision 188482) > +++ gcc/tree-vect-stmts.c (working copy) > @@ -1031,11 +1031,13 @@ vect_model_load_cost (stmt_vec_info stmt_info, int > /* The loads themselves. */ > if (STMT_VINFO_STRIDE_LOAD_P (stmt_info)) > { > - /* N scalar loads plus gathering them into a vector. > - ??? scalar_to_vec isn't the cost for that. */ > + /* N scalar loads plus gathering them into a vector. */ > + tree vectype = STMT_VINFO_VECTYPE (stmt_info); > inside_cost += (vect_get_stmt_cost (scalar_load) * ncopies > - * TYPE_VECTOR_SUBPARTS (STMT_VINFO_VECTYPE (stmt_info))); > - inside_cost += ncopies * vect_get_stmt_cost (scalar_to_vec); > + * TYPE_VECTOR_SUBPARTS (vectype)); > + inside_cost += ncopies > + * targetm.vectorize.builtin_vectorization_cost (vec_construct, > + vectype, 0); > } > else > vect_get_load_cost (first_dr, ncopies, > Index: gcc/config/spu/spu.c > =================================================================== > --- gcc/config/spu/spu.c (revision 188482) > +++ gcc/config/spu/spu.c (working copy) > @@ -6908,9 +6908,11 @@ spu_builtin_mask_for_load (void) > /* Implement targetm.vectorize.builtin_vectorization_cost. */ > static int > spu_builtin_vectorization_cost (enum vect_cost_for_stmt type_of_cost, > - tree vectype ATTRIBUTE_UNUSED, > + tree vectype, > int misalign ATTRIBUTE_UNUSED) > { > + unsigned elements; > + > switch (type_of_cost) > { > case scalar_stmt: > @@ -6937,6 +6939,11 @@ spu_builtin_vectorization_cost (enum vect_cost_for > case cond_branch_taken: > return 6; > > + case vec_construct: > + elements = TYPE_VECTOR_SUBPARTS (vectype); > + gcc_assert (elements > 1); > + return elements / 2 + 1; > + > default: > gcc_unreachable (); > } > Index: gcc/config/i386/i386.c > =================================================================== > --- gcc/config/i386/i386.c (revision 188482) > +++ gcc/config/i386/i386.c (working copy) > @@ -36072,9 +36072,11 @@ static const struct attribute_spec ix86_attribute_ > /* Implement targetm.vectorize.builtin_vectorization_cost. */ > static int > ix86_builtin_vectorization_cost (enum vect_cost_for_stmt type_of_cost, > - tree vectype ATTRIBUTE_UNUSED, > + tree vectype, > int misalign ATTRIBUTE_UNUSED) > { > + unsigned elements; > + > switch (type_of_cost) > { > case scalar_stmt: > @@ -36115,6 +36117,11 @@ ix86_builtin_vectorization_cost (enum vect_cost_fo > case vec_promote_demote: > return ix86_cost->vec_stmt_cost; > > + case vec_construct: > + elements = TYPE_VECTOR_SUBPARTS (vectype); > + gcc_assert (elements > 1); > + return elements / 2 + 1; > + > default: > gcc_unreachable (); > } > Index: gcc/config/rs6000/rs6000.c > =================================================================== > --- gcc/config/rs6000/rs6000.c (revision 188482) > +++ gcc/config/rs6000/rs6000.c (working copy) > @@ -3405,6 +3405,7 @@ rs6000_builtin_vectorization_cost (enum vect_cost_ > tree vectype, int misalign) > { > unsigned elements; > + tree elem_type; > > switch (type_of_cost) > { > @@ -3504,6 +3505,19 @@ rs6000_builtin_vectorization_cost (enum vect_cost_ > > return 2; > > + case vec_construct: > + elements = TYPE_VECTOR_SUBPARTS (vectype); > + elem_type = TREE_TYPE (vectype); > + gcc_assert (elements > 1); > + /* 32-bit vectors loaded into registers are stored as double > + precision, so we need n/2 converts in addition to the usual > + n/2 merges to construct a vector of short floats from them. */ > + if (SCALAR_FLOAT_TYPE_P (elem_type) > + && TYPE_PRECISION (elem_type) == 32) > + return elements + 1; > + else > + return elements / 2 + 1; > + > default: > gcc_unreachable (); > } > >