Skip site navigation (1)Skip section navigation (2)
Date:      Thu, 10 Feb 2005 22:46:43 +0000
From:      Thomas Hurst <tom.hurst@clara.net>
To:        Antony T Curtis <antony.t.curtis@ntlworld.com>
Cc:        Scott Long <scottl@freebsd.org>
Subject:   Re: MySQL benchmarks
Message-ID:  <20050210224643.GA47912@voi.aagh.net>
In-Reply-To: <1108071290.59338.8.camel@pcgem.rdg.cyberkinetica.com>
References:  <20050209205943.34c39e15.flynn@energyhq.es.eu.org> <420A909C.8070701@freebsd.org> <1108071290.59338.8.camel@pcgem.rdg.cyberkinetica.com>

next in thread | previous in thread | raw e-mail | index | archive | help
* Antony T Curtis (antony.t.curtis@ntlworld.com) wrote:

> If I remember correctly, MyISAM with skip-locking should rarely use
> fsync() calls... so if possible, the test could be re-run using MyISAM
> tables to see if there is any performance difference.

Poor performance is seen on read-only tests too; no fsync() overhead
there.  However, this message caught my eye:

  http://lists.freebsd.org/pipermail/freebsd-threads/2005-February/002848.html

  "Linux uses ptmalloc2 as its memory allocator, an extremely efficient
  implementation whose performance under a heavily loaded multithreaded
  system is impressive. FreeBSD does not."

There are a few malloc implimentations in ports which are supposedly
very good under threaded and multi-CPU conditions, including an older
ptmalloc, but I can't seem to make MySQL work with any of them using
LD_PRELOAD (it hangs with ptmalloc and SEGV's after a few seconds of
wdrain with Hoard).  This on 5-STABLE as of Jan 14, though, so don't let
that put anyone here off trying.

-- 
Thomas 'Freaky' Hurst
    http://hur.st/



Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?20050210224643.GA47912>