From xemacs-m  Tue Feb 25 04:50:29 1997
Received: from pat.idt.unit.no (0@pat.idt.unit.no [129.241.103.5])
	by xemacs.org (8.8.5/8.8.5) with ESMTP id EAA00583
	for <xemacs-beta@xemacs.org>; Tue, 25 Feb 1997 04:50:28 -0600 (CST)
Received: from ynde.idt.unit.no (1466@ynde.idt.unit.no [129.241.200.28])
	by pat.idt.unit.no (8.8.5/8.8.5) with ESMTP id LAA28664
	for <xemacs-beta@xemacs.org>; Tue, 25 Feb 1997 11:50:27 +0100 (MET)
Received: (from palat@localhost)
	by ynde.idt.unit.no (8.8.5/8.8.5) id LAA10562;
	Tue, 25 Feb 1997 11:50:23 +0100 (MET)
Sender: Sudeep.Palat@item.ntnu.no
To: XEmacs Beta <xemacs-beta@xemacs.org>
Subject: Re: lazy lock cpu usage
References: <9702250930.AA14387@mail.esrin.esa.it>
Mime-Version: 1.0 (generated by tm-edit 7.105)
Content-Type: text/plain; charset=US-ASCII
From: Sudeep Kumar Palat <Sudeep.Palat@item.ntnu.no>
Date: 25 Feb 1997 11:50:23 +0100
In-Reply-To: Simon Marshall's message of Tue, 25 Feb 97 09:30:53 GMT
Message-ID: <yv9afotgnww.fsf@ynde.idt.unit.no>
Lines: 25
X-Mailer: Gnus v5.4.15/XEmacs 20.1

>>>>> "Simon" == Simon Marshall <Simon.Marshall@esrin.esa.it> writes:

    Sudeep> Start xemacs -q and load a large file, for e.g., gnus.el.
    Sudeep> Turn on font lock.  Hold down the cursor key and look at
    Sudeep> cpu usage (I used top).  When the buffer has scrolled down
    Sudeep> 20%, the cpu usage by XEmacs is about 14% (it keeps
    Sudeep> increasing, presumably because, XEmacs is unable to
    Sudeep> process the key input at the keyboard repeat rate).

    Simon> So with Font Lock mode and without Lazy Lock mode has a
    Simon> dramatic slowdown?
Slowdown in the cpu usage.  Meaning lazy lock takes more CPU cycles.

    Sudeep> Turn on lazy lock, and repeat.  Scrolling 20% down, the
    Sudeep> cpu usage has shot up to about 50%.

    Simon> I don't have an XEmacs 20, only XEmacs 19.15b95.

    Simon> With lazy-lock-hide-invisible non-nil (the default for
    Simon> XEmacs) lazy-lock.el version 1 always does more work since
    Simon> it cannot rely on window-start and window-end.
Yes, setting lazy-lock-hide-invisible nil brought down the cpu usage
to a comparable 17%.  

sudeep

