public inbox for gcc-bugs@sourceware.org help / color / mirror / Atom feed
From: "rguenther at suse dot de" <gcc-bugzilla@gcc.gnu.org> To: gcc-bugs@gcc.gnu.org Subject: [Bug tree-optimization/113441] [14 Regression] Fail to fold the last element with multiple loop since g:2efe3a7de0107618397264017fb045f237764cc7 Date: Tue, 05 Mar 2024 08:21:26 +0000 [thread overview] Message-ID: <bug-113441-4-tCXQnJMdqg@http.gcc.gnu.org/bugzilla/> (raw) In-Reply-To: <bug-113441-4@http.gcc.gnu.org/bugzilla/> https://gcc.gnu.org/bugzilla/show_bug.cgi?id=113441 --- Comment #43 from rguenther at suse dot de <rguenther at suse dot de> --- On Mon, 4 Mar 2024, rsandifo at gcc dot gnu.org wrote: > https://gcc.gnu.org/bugzilla/show_bug.cgi?id=113441 > > --- Comment #41 from Richard Sandiford <rsandifo at gcc dot gnu.org> --- > (In reply to Richard Biener from comment #40) > > So I wonder if we can use "local costing" to decide a gather is always OK > > compared to the alternative with peeling for gaps. On x86 gather tends > > to be slow compared to open-coding it. > Yeah, on SVE gathers are generally ?enabling? instructions rather than > something to use for their own sake. > > I suppose one problem is that we currently only try to use gathers for > single-element groups. If we make a local decision to use gathers while > keeping that restriction, we could end up using gathers ?unnecessarily? while > still needing to peel for gaps for (say) a two-element group. > > That is, it's only better to use gathers than contiguous loads if by doing that > we avoid all need to peel for gaps (and if the cost of peeling for gaps was > high enough to justify the cost of using gathers over consecutive loads). Yep. I do want to experiment with a way to have vectorizable_* register multiple variants of vectorization and have ways to stitch together and cost the overall vectorization as a cheaper (and more flexible) way to "iteration". It will to some extent blow up combinations to try but there might be a way to use greedy relaxation techniques to converge to a lowest cost variant. > One of the things on the list to do (once everything is SLP!) is to support > loads with gaps directly via predication, so that we never load elements that > aren't needed. E.g. on SVE, a 64-bit predicate (PTRUE .D) can be used with a > 32-bit load (LD1W .S) to load only even-indexed elements. So a single-element > group with a group size of 2 could be done cheaply with just consecutive loads, > without peeling for gaps. Yep. Gap handling leaves to be desired (also when no predication is available), I also plan to address some shortcomings in that area early stage1. Note that generally the idea is that gap peeling is very cheap - unless that is the only reason to have an epilogue at all. The exeption might be small round-trip loops but those are best handled with predication where there's no good reason to do peeling for gaps at all.
next prev parent reply other threads:[~2024-03-05 8:21 UTC|newest] Thread overview: 48+ messages / expand[flat|nested] mbox.gz Atom feed top 2024-01-17 12:38 [Bug c/113441] New: [14 Regression] Fail to fold the last element with multiple loop juzhe.zhong at rivai dot ai 2024-01-17 12:45 ` [Bug tree-optimization/113441] " juzhe.zhong at rivai dot ai 2024-01-17 13:22 ` [Bug tree-optimization/113441] [13/14 " rguenth at gcc dot gnu.org 2024-01-17 14:07 ` juzhe.zhong at rivai dot ai 2024-01-17 14:35 ` rguenth at gcc dot gnu.org 2024-01-22 12:38 ` juzhe.zhong at rivai dot ai 2024-01-22 12:41 ` tnfchris at gcc dot gnu.org 2024-01-22 12:42 ` juzhe.zhong at rivai dot ai 2024-01-22 13:19 ` juzhe.zhong at rivai dot ai 2024-01-22 13:52 ` [Bug tree-optimization/113441] [14 " rguenth at gcc dot gnu.org 2024-01-22 16:16 ` tnfchris at gcc dot gnu.org 2024-01-22 22:16 ` juzhe.zhong at rivai dot ai 2024-01-23 6:42 ` rguenth at gcc dot gnu.org 2024-01-23 8:15 ` juzhe.zhong at rivai dot ai 2024-01-23 8:17 ` rguenther at suse dot de 2024-01-23 8:25 ` juzhe.zhong at rivai dot ai 2024-01-23 10:29 ` rguenther at suse dot de 2024-01-23 10:30 ` tnfchris at gcc dot gnu.org 2024-01-23 12:32 ` tnfchris at gcc dot gnu.org 2024-01-23 12:50 ` rguenth at gcc dot gnu.org 2024-01-23 12:52 ` rguenth at gcc dot gnu.org 2024-01-23 12:56 ` rguenth at gcc dot gnu.org 2024-01-23 13:02 ` rguenth at gcc dot gnu.org 2024-01-23 13:05 ` tnfchris at gcc dot gnu.org 2024-01-23 13:12 ` tnfchris at gcc dot gnu.org 2024-01-23 13:21 ` juzhe.zhong at rivai dot ai 2024-01-23 13:28 ` tnfchris at gcc dot gnu.org 2024-02-22 16:18 ` [Bug tree-optimization/113441] [14 Regression] Fail to fold the last element with multiple loop since g:2efe3a7de0107618397264017fb045f237764cc7 tnfchris at gcc dot gnu.org 2024-02-26 8:10 ` tnfchris at gcc dot gnu.org 2024-02-26 8:17 ` rguenther at suse dot de 2024-02-27 8:01 ` tnfchris at gcc dot gnu.org 2024-02-27 8:08 ` rguenth at gcc dot gnu.org 2024-02-29 22:18 ` rsandifo at gcc dot gnu.org 2024-03-01 9:44 ` rguenth at gcc dot gnu.org 2024-03-01 9:53 ` rsandifo at gcc dot gnu.org 2024-03-01 10:44 ` rguenther at suse dot de 2024-03-04 12:07 ` rsandifo at gcc dot gnu.org 2024-03-04 13:26 ` rsandifo at gcc dot gnu.org 2024-03-04 14:28 ` rguenth at gcc dot gnu.org 2024-03-04 14:48 ` rsandifo at gcc dot gnu.org 2024-03-04 15:01 ` rsandifo at gcc dot gnu.org 2024-03-04 15:10 ` rguenth at gcc dot gnu.org 2024-03-04 16:16 ` rsandifo at gcc dot gnu.org 2024-03-04 22:52 ` rsandifo at gcc dot gnu.org 2024-03-05 8:21 ` rguenther at suse dot de [this message] 2024-03-05 10:44 ` rguenth at gcc dot gnu.org 2024-03-07 20:50 ` law at gcc dot gnu.org 2024-05-07 7:43 ` [Bug tree-optimization/113441] [14/15 " rguenth 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-113441-4-tCXQnJMdqg@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).