unofficial mirror of bug-gnu-emacs@gnu.org 
 help / color / mirror / code / Atom feed
From: Stefan Monnier via "Bug reports for GNU Emacs, the Swiss army knife of text editors" <bug-gnu-emacs@gnu.org>
To: Ihor Radchenko <yantar92@posteo.net>
Cc: 65491@debbugs.gnu.org
Subject: bug#65491: [PATCH] Improve performance allocating vectors
Date: Mon, 28 Aug 2023 12:32:01 -0400	[thread overview]
Message-ID: <jwvbker2lhb.fsf-monnier+emacs@gnu.org> (raw)
In-Reply-To: <87il8z323q.fsf@localhost> (Ihor Radchenko's message of "Mon, 28 Aug 2023 10:14:49 +0000")

> The reason might be that vector sizes are distributed non-uniformly and
> some segments of vector_free_lists are filled more than others.

Indeed, I'd expect that at any given time, many buckets of
`vector_free_lists` are empty.

Also, when ELisp code allocates many structs of the same size, the first
few allocations may be satisfied by free elements in the appropriate
bucket, but very quickly that bucket will become empty, so we'll look
for the next non-empty bucket, and the remaining slightly-smaller free
vector will then be put back into a lower bucket but the next allocation
will go through the same loop to find that same "slightly-smaller free
vector", to make it yet a bit smaller, ... until it's all consumed after
which we'll look for the next bigger free vector etc...

So your code makes a lot of sense from that point of view.

Many mallocs approach the problem by creating whole pages dedicated to
a given object size, so when the bucket is empty, N new free vectors of
that exact size are created (in a new vector-block) and put into the
appropriate bucket so the next N allocations of that same size will
quickly find a free vector.  Since all those vector-blocks are made of
vectors of the same size, the size info can be shared between them
instead of each one of them carrying its own size.
[ In our case, vectors need to carry their size for other purposes than
  memory management, so it's not clear we'd gain much there.
  But it could help make `live_small_vector_holding` faster.  ]


        Stefan






  reply	other threads:[~2023-08-28 16:32 UTC|newest]

Thread overview: 46+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2023-08-24  9:59 bug#65491: [PATCH] Improve performance allocating vectors Ihor Radchenko
2023-08-26  7:14 ` Eli Zaretskii
2023-08-26  7:27   ` Ihor Radchenko
2023-08-26  7:31     ` Eli Zaretskii
2023-08-26  7:51       ` Ihor Radchenko
2023-08-26  8:07         ` Ihor Radchenko
2023-08-26  9:01         ` Eli Zaretskii
2023-08-26  7:47     ` Ihor Radchenko
2023-08-26 12:01 ` Mattias Engdegård
2023-08-26 14:54   ` Ihor Radchenko
2023-08-26 14:55   ` Stefan Monnier via Bug reports for GNU Emacs, the Swiss army knife of text editors
2023-08-27  9:54     ` Mattias Engdegård
2023-09-16 14:58     ` Mattias Engdegård
2023-09-16 16:12       ` Eli Zaretskii
2023-09-16 16:17         ` Eli Zaretskii
2023-09-16 16:32           ` Mattias Engdegård
2023-09-16 16:54             ` Eli Zaretskii
2023-09-16 17:03               ` Mattias Engdegård
2023-09-16 17:11                 ` Eli Zaretskii
2023-09-17  3:02                 ` Po Lu via Bug reports for GNU Emacs, the Swiss army knife of text editors
2023-09-17 17:02                   ` Stefan Monnier via Bug reports for GNU Emacs, the Swiss army knife of text editors
2023-09-18  2:19                     ` Po Lu via Bug reports for GNU Emacs, the Swiss army knife of text editors
2023-09-18  2:27                       ` Stefan Monnier via Bug reports for GNU Emacs, the Swiss army knife of text editors
2023-09-18  3:08                         ` Po Lu via Bug reports for GNU Emacs, the Swiss army knife of text editors
2023-09-18  4:10                           ` Stefan Monnier via Bug reports for GNU Emacs, the Swiss army knife of text editors
2023-09-16 16:54             ` Mattias Engdegård
2023-09-16 17:09               ` Eli Zaretskii
2023-09-16 17:22                 ` Mattias Engdegård
2023-09-16 18:19                   ` Eli Zaretskii
2023-09-16 19:04                     ` Mattias Engdegård
2023-09-16 19:46                 ` Paul Eggert
2023-09-17  5:18                   ` Eli Zaretskii
2023-09-17 15:22                     ` Paul Eggert
2023-09-17 16:15                       ` Eli Zaretskii
2023-09-17 16:37                         ` Paul Eggert
2023-09-17 16:44                           ` Eli Zaretskii
2023-09-18 16:10                             ` Mattias Engdegård
2023-09-18 17:13                               ` Eli Zaretskii
2023-09-19 13:28                               ` Mattias Engdegård
2023-09-19 14:04                                 ` Eli Zaretskii
2023-09-19 14:05                                   ` Mattias Engdegård
2023-09-25 16:06       ` Mattias Engdegård
2023-08-27 16:21 ` Stefan Monnier via Bug reports for GNU Emacs, the Swiss army knife of text editors
2023-08-28 10:14   ` Ihor Radchenko
2023-08-28 16:32     ` Stefan Monnier via Bug reports for GNU Emacs, the Swiss army knife of text editors [this message]
2023-08-28 12:47   ` Stefan Monnier via Bug reports for GNU Emacs, the Swiss army knife of text editors

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

  List information: https://www.gnu.org/software/emacs/

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=jwvbker2lhb.fsf-monnier+emacs@gnu.org \
    --to=bug-gnu-emacs@gnu.org \
    --cc=65491@debbugs.gnu.org \
    --cc=monnier@iro.umontreal.ca \
    --cc=yantar92@posteo.net \
    /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: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
Code repositories for project(s) associated with this public inbox

	https://git.savannah.gnu.org/cgit/emacs.git

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).