Mailing-List: contact cygwin-help AT cygwin DOT com; run by ezmlm
List-Subscribe: <mailto:cygwin-subscribe AT cygwin DOT com>
List-Archive: <http://sources.redhat.com/ml/cygwin/>
List-Post: <mailto:cygwin AT cygwin DOT com>
List-Help: <mailto:cygwin-help AT cygwin DOT com>, <http://sources.redhat.com/ml/#faqs>
Sender: cygwin-owner AT cygwin DOT com
Mail-Followup-To: cygwin AT cygwin DOT com
Delivered-To: mailing list cygwin AT cygwin DOT com
X-Authentication-Warning: just.smarts.com: elbergr owned process doing -bs
Date: Wed, 27 Aug 2003 20:55:47 -0400 (EDT)
From: Rich Elberger <elbergr AT smarts DOT com>
X-X-Sender: elbergr AT just
To: Rolf Campbell <Endlisnis AT mailc DOT net>, "" <cygwin AT cygwin DOT com>
Subject: Re: parallel make
In-Reply-To: <bijhiq$r5r$1@sea.gmane.org>
Message-ID: <Pine.SOL.4.50.0308272040450.17915-100000@just>
References: <Pine DOT SOL DOT 4 DOT 50 DOT 0308271643000 DOT 17159-100000 AT frame> <bijhiq$r5r$1 AT sea DOT gmane DOT org>
MIME-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII


This is starting to sound promising.  I hope others have similar
experiences.

Unfortunately I cannot just take make and the cyg dll.  The problem comes
in with how different utils interpret drive paths:

our old gnu:  //c/...
perl: c:/...
cyg:  /cygdrive/c/... (we mount NFS exports, hence the drive notation)...

So, we basically need to move things from the inside out rather than start
patching things on the outside of the system, eventually getting to the
core.  This makes the migration pretty expensive, so I would like to see
more cases before making the resource justification case.  If it's a set
of production machines, no problem, but the change would have to occur on
all development workstations.

I highly appreciate everyone's help because I think this is the only forum
I could ever get such kind of feedback.

-- rich

On Wed, 27 Aug 2003, Rolf Campbell wrote:

> make -j does work in cygwin, to an extent.  If you try to use too many
> processes, cygwin seems to flip out.  -j20 does seem to work fine though
> (it only starts acting strang around -j100).
>
> On my large build system, we have a slow disk, and I find that when the
> disk cache is empty, -j4 speeds it up about 40%.  When the build system
> is cached, -j slows things down by a few percent.  This is dealing with
> 1 processor.  Of course, if you have multiple processors, the speed
> should scale linearly (with 4 cpu's, -j4 is about 70% faster).
>
> Rich Elberger wrote:
>
> > Hi folks,
> > Currently our build environment uses parallel make (-j jobs option) on all
> > our unixes using gnu tools.  We use an older version of gnu tools on our
> > windows boxes.  The older make on the windows box does not do parallel
> > make (or at least correctly).  I want to upgrade to the latest cygwin to
> > see if parallel make works, but this will require significant changes to
> > our build engine, so I would like to confirm a few things if possible.
> >
> > 1. Does the -j jobs option work well on windows. (part b: does it work
> > with the MSVC (6/7) compiler (which probably doesn't make a difference
> > anyway)?
> > 2. Has anyone done this in a very large project, and if so, do you have
> > any performance gain stats (which, I acknowledge, is tied to
> > processor-intensive makes and how many processors the machine has).
> >
> > I realize that dos does not allow for threading so I don't know if this is
> > a cmd.exe-related issue or not (since cmd.exe is the parent shell, I
> > don't know if this affects the behavior).
> >
> > thanks in advance --
> >
> >
>
>
>
> --
> Unsubscribe info:      http://cygwin.com/ml/#unsubscribe-simple
> Problem reports:       http://cygwin.com/problems.html
> Documentation:         http://cygwin.com/docs.html
> FAQ:                   http://cygwin.com/faq/
>
>

--
Unsubscribe info:      http://cygwin.com/ml/#unsubscribe-simple
Problem reports:       http://cygwin.com/problems.html
Documentation:         http://cygwin.com/docs.html
FAQ:                   http://cygwin.com/faq/