Mailing-List: contact cygwin-help AT sourceware DOT cygnus DOT com; run by ezmlm
List-Subscribe: <mailto:cygwin-subscribe AT sources DOT redhat DOT com>
List-Archive: <http://sources.redhat.com/ml/cygwin/>
List-Post: <mailto:cygwin AT sources DOT redhat DOT com>
List-Help: <mailto:cygwin-help AT sources DOT redhat DOT com>, <http://sources.redhat.com/ml/#faqs>
Sender: cygwin-owner AT sources DOT redhat DOT com
Delivered-To: mailing list cygwin AT sources DOT redhat DOT com
From: "Gerrit P. Haase" <gp AT familiehaase DOT de>
Organization: Esse keine toten Tiere
To: ian DOT ray AT nokia DOT com, cygwin AT cygwin DOT com
Date: Tue, 9 Oct 2001 09:56:58 +0200
MIME-Version: 1.0
Content-type: text/plain; charset=US-ASCII
Content-transfer-encoding: 7BIT
Subject: RE: cygwin 1.3.[23] grindingly slow
Reply-to: cygwin AT cygwin DOT com
Message-ID: <3BC2C9EA.27805.2CB1E831@localhost>
In-reply-to: <C78815621F631C4AAFAE6FEC74ECDC9A120874@caebe004.NOE.Nokia.com>
X-mailer: Pegasus Mail for Win32 (v3.12cDE)
X-Hops: 1
X-Sender: 320081107336-0001 AT t-dialin DOT net

ian DOT ray AT nokia DOT com schrieb am 2001-10-09, 10:42:

>[...]
>> Is it faster or slower if HOMEDRIVE and HOMEPATH is set?
>> What is HOMEPATH and HOMEDRIVE? Aren't that Windows environment
>> variables? I never used them for Cygwin.
>
>I tested the patch on a (big) recursive make, it improved time from
>12minutes to 10minutes.

About 15%, that is not bad.

>I used strace to look for performance bottlenecks.
>
>The patch made the improvement if HOMEDRIVE and HOMEPATH are set,
>because (at least in my test) doing a query for user information (using
>NetUserGetInfo() API) seems to be slow.
>
>YMMV. This is precisely the kind of feedback that I was interested in.
>There is no point in submitting a patch if it does not make an
>improvement :)
>
>FYI: I originally made the patch to snapshot cygwin-snapshot-20011001-1.

It applies correct to CVS sources.

>Now, I wonder if all I have actually saved is some DLL loading time: and
>that you would only see a benefit during a big job, such as, for example
>a recursive make???

If that causes a speedup about 15% it is o.k. to apply it.

>[snip]
>> So at least ls -lRa is *faster* without the patch (which is
>> version cygwin-1.3.3-2), the tests with the patch is a recent
>> debugging cygwin (pre-1.3.4, maybe this is a little slower
>> because of debugging?).
>
>Could you try again with a non-debugging version?

I'm just building a fresh debug-version.  I will strip
it off then and try with both to see if makes a difference.

Gerrit


-- 
=^..^=

--
Unsubscribe info:      http://cygwin.com/ml/#unsubscribe-simple
Bug reporting:         http://cygwin.com/bugs.html
Documentation:         http://cygwin.com/docs.html
FAQ:                   http://cygwin.com/faq/