From: Stefan Monnier <monnier@iro.umontreal.ca>
To: Dmitry Antipov <dmantipov@yandex.ru>
Cc: emacs-devel@gnu.org
Subject: Re: Proposal: block-based vector allocator
Date: Wed, 06 Jun 2012 15:18:32 -0400 [thread overview]
Message-ID: <jwvmx4gi7th.fsf-monnier+emacs@gnu.org> (raw)
In-Reply-To: <4FCF701C.6070106@yandex.ru> (Dmitry Antipov's message of "Wed, 06 Jun 2012 18:58:36 +0400")
>> I explained this earlier: using a vector-block for largish vectors is
>> not efficient (because the overhead of the vector-block is not shared
>> among enough vectors).
>> E.g. for a vector of size VECTOR_BLOCK_BYTES, using the vector-block
>> code is a complete waste
> ...of just one pointer, so 8/4088, or 0.2% in terms of space for this rare
> case; an overhead of having one more mem_node is 6x larger.
No, no, no: since this VECTOR_BLOCK_BYTES will fill the whole
vector-block it will incur the full cost of a mem_mode anyway when
allocating the vector-block.
>> For the case of a vector of size VECTOR_BLOCK_BYTES, allocating in
>> a vector block will always be a bad idea, no matter the scenario.
> Allocating a lot of VECTOR_BLOCK_BYTES / 2 + sizeof (Lisp_Object)
> vectors (and negligible amount of others) will waste ~50% of space in
> blocks;
I'm not too worried about this, actually, what I'm after is different:
The cost (CPU or memory, it doesn't matter much here, and we ignore
fragmentation) of allocating a vector is something like:
and CPU costs alike, ignores fragmentation):
- outside of a vector-block: malloc + mem_node
- inside a vector-block: vector-block + frac * (malloc + mem_node + a-bit-more)
where "frac" is a fraction that's more or less vector-size /
VECTOR_BLOCK_BYTES.
So for small vectors, "frac" is small and since we expect the
vector-block overhead to be significantly smaller than malloc+mem_node
we win. But past a certain value of "frac", we're better off allocating
outside of a vector-block. As I said, I don't know exactly where that
threshold is, but using VECTOR_BLOCK_BYTES / 2 should be good enough.
And it also happens to help with the problem you mention, reducing the
worst-case 50% (which really means doubling the memory use) to
a worst-case 25% (which only means an extra 33% of memory use, which
I find a lot more acceptable) lost in fragmentation.
Stefan
next prev parent reply other threads:[~2012-06-06 19:18 UTC|newest]
Thread overview: 62+ messages / expand[flat|nested] mbox.gz Atom feed top
2011-12-06 5:22 Proposal: block-based vector allocator Dmitry Antipov
2011-12-06 13:35 ` Stefan Monnier
2011-12-06 15:14 ` Dmitry Antipov
2011-12-06 19:39 ` Stefan Monnier
2011-12-07 5:05 ` Dmitry Antipov
2011-12-07 12:27 ` Carsten Mattner
2011-12-07 13:52 ` Stefan Monnier
2011-12-07 16:08 ` Dmitry Antipov
2011-12-07 16:30 ` Stefan Monnier
2011-12-08 8:50 ` Dmitry Antipov
2011-12-08 13:52 ` Stefan Monnier
2011-12-08 1:53 ` Stephen J. Turnbull
2011-12-08 4:41 ` Dmitry Antipov
2011-12-08 14:10 ` Stefan Monnier
2011-12-08 16:48 ` Dmitry Antipov
2011-12-08 19:58 ` Stefan Monnier
2011-12-09 7:32 ` Eli Zaretskii
2011-12-09 9:04 ` Dmitry Antipov
2011-12-09 14:05 ` Stefan Monnier
2011-12-09 16:15 ` Dmitry Antipov
2011-12-09 21:04 ` Stefan Monnier
2011-12-11 13:18 ` Dmitry Antipov
2011-12-12 3:07 ` Dmitry Antipov
2011-12-12 16:24 ` Stefan Monnier
2011-12-09 4:44 ` Stephen J. Turnbull
[not found] ` <jwvaa1yjs21.fsf-monnier+emacs@gnu.org>
2012-05-17 7:58 ` Dmitry Antipov
2012-05-18 17:40 ` Stefan Monnier
2012-05-21 12:19 ` Dmitry Antipov
2012-05-21 13:02 ` Andreas Schwab
2012-05-21 13:48 ` Dmitry Antipov
2012-05-21 15:07 ` Andreas Schwab
2012-05-22 5:23 ` Ken Raeburn
2012-05-21 20:12 ` Stefan Monnier
2012-05-22 8:24 ` Dmitry Antipov
2012-05-31 13:44 ` Dmitry Antipov
2012-05-31 15:43 ` Paul Eggert
2012-06-01 5:15 ` Dmitry Antipov
2012-06-01 5:44 ` Paul Eggert
2012-06-01 9:06 ` Dmitry Antipov
2012-06-01 17:36 ` Stefan Monnier
2012-06-02 0:32 ` Paul Eggert
2012-06-02 7:41 ` Eli Zaretskii
2012-06-03 6:49 ` Paul Eggert
2012-06-03 14:26 ` Eli Zaretskii
2012-05-31 21:16 ` Stefan Monnier
2012-06-01 7:34 ` Dmitry Antipov
2012-06-01 17:40 ` Stefan Monnier
2012-06-01 17:43 ` Stefan Monnier
2012-06-06 7:02 ` Dmitry Antipov
2012-06-06 13:13 ` Stefan Monnier
2012-06-06 14:58 ` Dmitry Antipov
2012-06-06 19:18 ` Stefan Monnier [this message]
2012-06-07 10:03 ` Dmitry Antipov
2012-06-07 14:07 ` Stefan Monnier
2012-06-08 5:50 ` Dmitry Antipov
2012-06-08 6:17 ` Stefan Monnier
2012-06-08 8:49 ` Dmitry Antipov
2012-06-08 8:53 ` Eli Zaretskii
2012-06-08 9:41 ` Eli Zaretskii
2012-06-08 10:00 ` Eli Zaretskii
2012-06-08 6:57 ` Eli Zaretskii
2012-06-08 6:38 ` Paul Eggert
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=jwvmx4gi7th.fsf-monnier+emacs@gnu.org \
--to=monnier@iro.umontreal.ca \
--cc=dmantipov@yandex.ru \
--cc=emacs-devel@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: 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).