public inbox for gcc-bugs@sourceware.org help / color / mirror / Atom feed
From: "rguenth at gcc dot gnu.org" <gcc-bugzilla@gcc.gnu.org> To: gcc-bugs@gcc.gnu.org Subject: [Bug tree-optimization/98856] [11 Regression] botan AES-128/XTS is slower by ~17% since r11-6649-g285fa338b06b804e72997c4d876ecf08a9c083af Date: Fri, 05 Mar 2021 07:44:24 +0000 [thread overview] Message-ID: <bug-98856-4-qoLZGifejo@http.gcc.gnu.org/bugzilla/> (raw) In-Reply-To: <bug-98856-4@http.gcc.gnu.org/bugzilla/> https://gcc.gnu.org/bugzilla/show_bug.cgi?id=98856 --- Comment #22 from Richard Biener <rguenth at gcc dot gnu.org> --- (In reply to Uroš Bizjak from comment #21) > (In reply to Uroš Bizjak from comment #20) > > (In reply to Richard Biener from comment #18) > > > Even on Skylake it's 2 (movq) + 3 (vpinsr), so there it's 6 vs. 3. Not > > > sure if we should somehow do this late somehow (peephole or splitter) since > > > it requires one more %xmm register. > > What happens if you disparage [v]pinsrd alternatives in vec_concatv2di? > > Please try this: > > diff --git a/gcc/config/i386/sse.md b/gcc/config/i386/sse.md > index db5be59f5b7..edf7b1a3074 100644 > --- a/gcc/config/i386/sse.md > +++ b/gcc/config/i386/sse.md > @@ -16043,7 +16043,12 @@ > (const_string "maybe_evex") > ] > (const_string "orig"))) > - (set_attr "mode" "TI,TI,TI,TI,TI,TI,V4SF,V2SF,V2SF")]) > + (set_attr "mode" "TI,TI,TI,TI,TI,TI,V4SF,V2SF,V2SF") > + (set (attr "preferred_for_speed") > + (cond [(eq_attr "alternative" "0,1,2,3") > + (symbol_ref "false") > + ] > + (symbol_ref "true")))]) > > (define_insn "*vec_concatv2di_0" That works to avoid the vpinsrq. I guess the case of a mem operand behaves similar to a gpr (plus the load uop), at least I don't have any contrary evidence (but I didn't do any microbenchmarks either). I'm not sure IRA/LRA will optimally handle the situation with register pressure causing spilling in case it needs to reload both gpr operands. At least for typedef long v2di __attribute__((vector_size(16))); v2di foo (long a, long b) { return (v2di){a, b}; } with -msse4.1 -O3 -ffixed-xmm1 -ffixed-xmm2 -ffixed-xmm3 -ffixed-xmm4 -ffixed-xmm5 -ffixed-xmm6 -ffixed-xmm7 -ffixed-xmm8 -ffixed-xmm9 -ffixed-xmm10 -ffixed-xmm11 -ffixed-xmm12 -ffixed-xmm13 -ffixed-xmm14 -ffixed-xmm15 I get with the patch foo: .LFB0: .cfi_startproc movq %rsi, -16(%rsp) movq %rdi, %xmm0 pinsrq $1, -16(%rsp), %xmm0 ret while without it's movq %rdi, %xmm0 pinsrq $1, %rsi, %xmm0 as far as I understand LRA dumps the new attribute is a hard one, even applying when other alternatives are worse. In this case we choose alt 7. Covering also alts 7 and 8 with the optimize-for-speed attribute causes reload fails - which is expected if there's no way for LRA to choose alt 1. The following seems to work for the small testcase above but not for the important case in the benchmark (meh). diff --git a/gcc/config/i386/sse.md b/gcc/config/i386/sse.md index db5be59f5b7..e393a0d823b 100644 --- a/gcc/config/i386/sse.md +++ b/gcc/config/i386/sse.md @@ -15992,7 +15992,7 @@ (match_operand:DI 1 "register_operand" " 0, 0,x ,Yv,0,Yv,0,0,v") (match_operand:DI 2 "nonimmediate_operand" - " rm,rm,rm,rm,x,Yv,x,m,m")))] + " !rm,!rm,!rm,!rm,x,Yv,x,!m,!m")))] "TARGET_SSE" "@ pinsrq\t{$1, %2, %0|%0, %2, 1} I guess the idea of this insn setup was exactly to get IRA/LRA choose the optimal instruction sequence - otherwise exposing the reload so late is probably suboptimal.
next prev parent reply other threads:[~2021-03-05 7:44 UTC|newest] Thread overview: 48+ messages / expand[flat|nested] mbox.gz Atom feed top 2021-01-27 14:28 [Bug tree-optimization/98856] New: " marxin at gcc dot gnu.org 2021-01-27 14:29 ` [Bug tree-optimization/98856] " marxin at gcc dot gnu.org 2021-01-27 14:44 ` rguenth at gcc dot gnu.org 2021-01-28 7:47 ` rguenth at gcc dot gnu.org 2021-01-28 8:44 ` marxin at gcc dot gnu.org 2021-01-28 9:40 ` rguenth at gcc dot gnu.org 2021-01-28 11:03 ` rguenth at gcc dot gnu.org 2021-01-28 11:19 ` rguenth at gcc dot gnu.org 2021-01-28 11:57 ` rguenth at gcc dot gnu.org 2021-02-05 10:18 ` rguenth at gcc dot gnu.org 2021-02-05 11:52 ` jakub at gcc dot gnu.org 2021-02-05 12:52 ` rguenth at gcc dot gnu.org 2021-02-05 13:43 ` jakub at gcc dot gnu.org 2021-02-05 14:36 ` jakub at gcc dot gnu.org 2021-02-05 16:29 ` jakub at gcc dot gnu.org 2021-02-05 17:55 ` jakub at gcc dot gnu.org 2021-02-05 19:48 ` jakub at gcc dot gnu.org 2021-02-08 15:14 ` jakub at gcc dot gnu.org 2021-03-04 12:14 ` rguenth at gcc dot gnu.org 2021-03-04 15:36 ` rguenth at gcc dot gnu.org 2021-03-04 16:12 ` rguenth at gcc dot gnu.org 2021-03-04 17:56 ` ubizjak at gmail dot com 2021-03-04 18:12 ` ubizjak at gmail dot com 2021-03-05 7:44 ` rguenth at gcc dot gnu.org [this message] 2021-03-05 7:46 ` rguenth at gcc dot gnu.org 2021-03-05 8:29 ` ubizjak at gmail dot com 2021-03-05 10:04 ` rguenther at suse dot de 2021-03-05 10:43 ` rguenth at gcc dot gnu.org 2021-03-05 11:56 ` ubizjak at gmail dot com 2021-03-05 12:25 ` ubizjak at gmail dot com 2021-03-05 12:27 ` rguenth at gcc dot gnu.org 2021-03-05 12:49 ` jakub at gcc dot gnu.org 2021-03-05 12:52 ` ubizjak at gmail dot com 2021-03-05 12:55 ` rguenther at suse dot de 2021-03-05 13:06 ` rguenth at gcc dot gnu.org 2021-03-05 13:08 ` ubizjak at gmail dot com 2021-03-05 14:35 ` rguenth at gcc dot gnu.org 2021-03-08 10:41 ` rguenth at gcc dot gnu.org 2021-03-08 13:20 ` rguenth at gcc dot gnu.org 2021-03-08 15:46 ` amonakov at gcc dot gnu.org 2021-04-27 11:40 ` [Bug tree-optimization/98856] [11/12 " jakub at gcc dot gnu.org 2021-05-13 10:17 ` cvs-commit at gcc dot gnu.org 2021-07-28 7:05 ` rguenth at gcc dot gnu.org 2022-01-21 13:20 ` rguenth at gcc dot gnu.org 2022-04-21 7:48 ` rguenth at gcc dot gnu.org 2023-04-17 21:43 ` [Bug tree-optimization/98856] [11/12/13/14 " lukebenes at hotmail dot com 2023-04-18 9:07 ` rguenth at gcc dot gnu.org 2023-05-29 10:04 ` jakub at gcc dot gnu.org
Reply instructions: You may reply publicly to this message via plain-text email using any one of the following methods: * Save the following mbox file, import it into your mail client, and reply-to-all from there: mbox Avoid top-posting and favor interleaved quoting: https://en.wikipedia.org/wiki/Posting_style#Interleaved_style * Reply using the --to, --cc, and --in-reply-to switches of git-send-email(1): git send-email \ --in-reply-to=bug-98856-4-qoLZGifejo@http.gcc.gnu.org/bugzilla/ \ --to=gcc-bugzilla@gcc.gnu.org \ --cc=gcc-bugs@gcc.gnu.org \ /path/to/YOUR_REPLY https://kernel.org/pub/software/scm/git/docs/git-send-email.html * If your mail client supports setting the In-Reply-To header via mailto: links, try the mailto: linkBe sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions for how to clone and mirror all data and code used for this inbox; as well as URLs for read-only IMAP folder(s) and NNTP newsgroup(s).