From xemacs-m  Tue Mar 18 19:02:18 1997
Received: from mailbox2.ucsd.edu (mailbox2.ucsd.edu [132.239.1.54])
	by xemacs.org (8.8.5/8.8.5) with ESMTP id TAA17674
	for <xemacs-beta@xemacs.org>; Tue, 18 Mar 1997 19:02:17 -0600 (CST)
Received: from sdnp5.ucsd.edu (sdnp5.ucsd.edu [132.239.79.10]) by mailbox2.ucsd.edu (8.8.5/8.6.9) with SMTP id RAA04103 for <xemacs-beta@xemacs.org>; Tue, 18 Mar 1997 17:02:18 -0800 (PST)
Received: by sdnp5.ucsd.edu (SMI-8.6/SMI-SVR4)
	id RAA19636; Tue, 18 Mar 1997 17:04:16 -0800
Sender: dmoore@sdnp5.ucsd.edu
To: XEmacs Beta Mailing List <xemacs-beta@xemacs.org>
Subject: Re: relocating allocator?
References: <199703190029.QAA14028@roy>
X-Face: "oX;zS#-JU$-,WKSzG.1gGE]x^cIg!hW.dq>.f6pzS^A+(k!T|M:}5{_%>Io<>L&{hO7W4cicOQ|>/lZ1G(m%7iaCf,6Qgk0%%Bz7b2-W3jd0m_UG\Y;?]}4s0O-U)uox>P3JN)9cm]O\@,vy2e{`3pb!"pqmRy3peB90*2L
Mail-Copies-To: never
Mime-Version: 1.0 (generated by tm-edit 7.105)
Content-Type: text/plain; charset=US-ASCII
From: David Moore <dmoore@ucsd.edu>
Date: 18 Mar 1997 17:04:15 -0800
In-Reply-To: Damon Lipparelli's message of Tue, 18 Mar 1997 16:29:42 -0800
Message-ID: <rvbu8g1yls.fsf@sdnp5.ucsd.edu>
Lines: 32
X-Mailer: Gnus v5.4.28/XEmacs 19.15(beta100)

Damon Lipparelli <lipp@primus.com> writes:

> Does anyone remember the consensus on whether or not rel-alloc is a good
> idea?  I seem to recall it being suspected for some crashes and performance 
> problems, but I can't find a specific message.

The _mmap_ rel-alloc can cause severe performance problems in certain
situations.  Such as if you are growing a buffer often (and of course,
it hurts a lot on the initial growth of a large buffer).

So for example, if you like to leave a *shell* buffer open for eternity,
you pay a pretty good hit every time you exceed the 16k extra space it
preallocates.  If you load in a large file, you pay lots of pain.  If
you run the big file generation bench mark, it'll be slow.

On the other hand if you tend to mostly deal with smaller sized files,
it's not quite so bad. :)

I may also be overstating the above case, since you might pay costs with
large buffer growth w/o the rel-alloc turned on.  For example, if you
alternatively grow two different buffers.

There's a lot of potential for making the mmap rel-alloc quite a bit
better than the other approaches.  It's just sitting there begging to be
done.


-- 
David Moore <dmoore@ucsd.edu>       | Computer Systems Lab      __o
UCSD Dept. Computer Science - 0114  | Work: (619) 534-8604    _ \<,_
La Jolla, CA 92093-0114             | Fax:  (619) 534-1445   (_)/ (_)
<URL:http://oj.egbt.org/dmoore/>    | In a cloud bones of steel.

