From: Geoff Keating <geoffk@geoffk.org>
To: Devang Patel <dpatel@apple.com>
Cc: gcc@gcc.gnu.org, Matt Austern <austern@apple.com>
Subject: Re: GCC 3.3, GCC 3.4
Date: Mon, 03 Feb 2003 22:32:00 -0000 [thread overview]
Message-ID: <jmd6m96lpk.fsf@desire.geoffk.org> (raw)
In-Reply-To: <14035A2A-37C2-11D7-8C5C-0003935AAA26@apple.com>
Devang Patel <dpatel@apple.com> writes:
> On Monday, February 3, 2003, at 12:04 AM, Tim Josling wrote:
>
> >
> >> On Thursday, January 30, 2003, at 03:38 PM, Neil Booth wrote:
> >>> I'm interested in improving the quality of GCC. Your statment above
> >>> indicates you're interested in pointing to a number and saying "look,
> >>> it's lower", regardless of what that means long-term. Hell, let's
> >>> not free anything at all and turn GC off altogether; GCC would be
> >>> faster
> >>> for 90% of files I imagine.
> >>
> >> Guess what? For smaller programs, it already does that! So, you are
> >> just describing status quo.
> >
> > It is one thing to stop collecting; it is another to turn off GC.
> >
> > Tests I did last year showed that even without doing collection, gcc
> > is slowed
> > down by the requirements of GC:
> >
> > 1. Allocating storage of similar sizes together, sacrificing locality.
> > 2. Wasting space by rounding up allocations to the next power of 2,
> > wasting
> > space and increasing working set sizes.
> > 3. Having bits of GC code all over the place impacting code locality
> > and
> > working set sizes.
> >
> > Turning off GC slows down the part of the build of GCC attributable
> > to the
> > xgcc/cc1/ld by 5%.
> >
> > However removing it and allocating sequentially (each allocation is
> > next to
> > the previous one) reduces build xgcc/cc1/ld compile time by about 5%.
> >
> > Add these together and the requirements of GC slow down the compiler
> > by about
> > 10% (though collecting gets half of it back)! And most compiles
> > don't need or
> > benefit from it!
>
> Matt Austern at Apple did similar experiment by using ggc placeholder.
> (same interface as ggc, no collection, allocate memory using mmap in
> chunks and
> satisfy memory allocation requests from these chunks etc..).
>
> He got around ~12% speedup for sample c++ source.
To be precise, Matt found that the speedup of using the ggc
placeholder was about equivalent to the speedup of using ggc-page but
disabling garbage collection. This contradicts the results that Tim
Josling got above.
--
- Geoffrey Keating <geoffk@geoffk.org>
next prev parent reply other threads:[~2003-02-03 22:32 UTC|newest]
Thread overview: 118+ messages / expand[flat|nested] mbox.gz Atom feed top
[not found] <1043976898.27601.ezmlm@gcc.gnu.org>
2003-02-03 19:50 ` Tim Josling
2003-02-03 21:54 ` Devang Patel
2003-02-03 22:32 ` Geoff Keating [this message]
2003-02-03 22:40 ` GC overhead (Re: GCC 3.3, GCC 3.4) Matt Austern
2003-02-03 22:53 ` Matt Austern
2003-02-04 1:14 ` Ziemowit Laski
2003-02-04 20:17 ` GCC 3.3, GCC 3.4 Tim Josling
2003-02-04 20:47 ` Matt Austern
2003-02-06 19:26 ` Tim Josling
2003-02-04 0:05 ` Tim Hollebeek
2003-02-04 20:07 ` Tim Josling
2003-02-04 2:03 ` Richard Henderson
2003-02-04 2:26 ` Zack Weinberg
2003-02-04 4:14 ` Daniel Berlin
[not found] <1044318563.708.247.camel@steven>
2003-02-04 1:13 ` Scott Robert Ladd
2003-02-04 0:31 Steven Bosscher
[not found] <200301312121.QAA22864@makai.watson.ibm.com>
2003-02-03 20:44 ` Tom Tromey
2003-02-03 21:02 ` Benjamin Kosnik
2003-02-03 21:10 ` Mark Mitchell
2003-02-03 21:31 ` David Edelsohn
2003-02-03 21:53 ` Mark Mitchell
2003-02-03 22:16 ` Scott Robert Ladd
2003-02-03 22:30 ` Jack Lloyd
2003-02-04 1:21 ` Scott Robert Ladd
2003-02-03 22:03 ` Scott Robert Ladd
2003-02-03 21:45 ` Benjamin Kosnik
2003-02-03 21:56 ` Mark Mitchell
2003-02-03 22:34 ` Tom Tromey
2003-02-03 23:03 ` David Edelsohn
2003-02-03 21:17 ` Gabriel Dos Reis
2003-02-03 21:29 ` Mark Mitchell
2003-02-03 21:41 ` David Edelsohn
2003-02-03 21:42 ` Gabriel Dos Reis
-- strict thread matches above, loose matches on Subject: below --
2003-02-02 6:17 Brad Lucier
[not found] <20030131165429.32d4d907.bkoz@redhat.com>
2003-02-01 1:32 ` Mark Mitchell
2003-02-01 1:43 ` Jan Hubicka
[not found] <200301312258.OAA14911@emf.net>
2003-02-01 0:00 ` Mike Stump
2003-01-31 16:29 Richard Kenner
2003-01-31 5:15 Wolfgang Bangerth
2003-01-31 10:39 ` Mike Stump
2003-01-31 17:54 ` Wolfgang Bangerth
2003-01-31 0:37 Benjamin Kosnik
2003-01-31 0:52 ` Zack Weinberg
2003-01-31 1:16 ` Gabriel Dos Reis
2003-01-31 1:34 ` Zack Weinberg
2003-01-31 17:53 ` Mark Mitchell
2003-01-31 1:14 ` Gabriel Dos Reis
2003-01-31 1:25 ` Benjamin Kosnik
2003-01-31 1:27 ` Gerald Pfeifer
2003-01-31 1:52 ` Benjamin Kosnik
2003-01-31 20:59 ` Joe Buck
2003-01-31 22:44 ` Tom Tromey
2003-01-31 6:09 ` Joe Buck
2003-01-31 7:39 ` Zack Weinberg
2003-01-31 20:43 ` Geoff Keating
[not found] ` <200301312127.QAA22861@caip.rutgers.edu>
[not found] ` <jmy9516lsd.fsf@desire.geoffk.org>
2003-02-01 4:32 ` Kaveh R. Ghazi
2003-01-31 11:52 ` Tom Lord
2003-01-31 20:52 ` Joe Buck
[not found] ` <200301312311.PAA15835@emf.net>
2003-02-01 1:28 ` Joe Buck
2003-02-01 1:49 ` Michel LESPINASSE
2003-01-31 21:07 ` Mike Stump
2003-01-31 13:50 ` Joseph S. Myers
2003-01-31 18:02 ` Mark Mitchell
2003-01-31 19:05 ` Devang Patel
2003-01-31 20:05 ` Mark Mitchell
2003-01-31 20:37 ` Graham Stott
2003-01-31 20:57 ` Devang Patel
2003-01-31 21:35 ` Tom Tromey
2003-01-31 21:02 ` Joseph S. Myers
2003-01-31 21:13 ` Joel Sherrill
2003-01-31 21:21 ` Devang Patel
2003-01-31 22:59 ` Joel Sherrill
2003-01-31 23:42 ` Devang Patel
2003-01-31 19:44 ` Matt Austern
2003-01-31 20:23 ` Mark Mitchell
2003-02-01 0:21 ` Michael Matz
2003-02-01 3:18 ` David Edelsohn
2003-01-31 20:31 ` Richard Henderson
2003-01-31 22:59 ` Andreas Jaeger
2003-01-31 23:12 ` Phil Edwards
2003-01-30 23:53 Karel Gardas
2003-01-31 0:13 ` Mike Stump
2003-01-30 20:00 Benjamin Kosnik
2003-01-30 21:29 ` Mark Mitchell
2003-01-30 21:42 ` Mike Stump
2003-01-31 13:31 ` Nathan Sidwell
2003-01-30 22:15 ` Joseph S. Myers
2003-01-30 22:56 ` Neil Booth
2003-01-30 22:58 ` Michael S. Zick
2003-01-30 23:22 ` Michael Matz
2003-01-30 23:25 ` Michael S. Zick
2003-01-31 11:16 ` Bernd Jendrissek
2003-01-30 23:55 ` Mike Stump
2003-01-31 0:11 ` Paolo Carlini
2003-01-31 0:15 ` Neil Booth
2003-01-31 0:29 ` Phil Edwards
2003-01-31 0:30 ` Neil Booth
2003-01-31 2:08 ` Mike Stump
2003-01-31 16:01 ` Michael S. Zick
2003-01-31 0:41 ` Mike Stump
2003-01-31 0:17 ` Benjamin Kosnik
2003-01-31 0:21 ` Neil Booth
2003-01-31 0:25 ` Matt Austern
2003-01-31 1:10 ` Benjamin Kosnik
2003-01-31 1:22 ` Gabriel Dos Reis
2003-01-31 1:55 ` Mike Stump
2003-01-31 1:54 ` Mike Stump
2003-01-31 2:40 ` Geoff Keating
2003-01-31 12:54 ` Richard Earnshaw
2003-01-31 20:22 ` Geoff Keating
2003-01-30 11:22 Mark Mitchell
2003-01-30 19:07 ` Mike Stump
2003-01-30 19:58 ` Mark Mitchell
2003-01-30 20:15 ` Mike Stump
2003-01-30 20:59 ` Mark Mitchell
2003-01-30 19:20 ` Matt Austern
2003-01-30 19:52 ` Mark Mitchell
2003-01-30 20:11 ` David Edelsohn
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=jmd6m96lpk.fsf@desire.geoffk.org \
--to=geoffk@geoffk.org \
--cc=austern@apple.com \
--cc=dpatel@apple.com \
--cc=gcc@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: link
Be 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).