Replied: Thu, 16 May 2002 11:27:17 -0500 Replied: "John B Rundle" Return-Path: rundle@cires.colorado.edu Delivery-Date: Thu May 16 10:49:56 2002 Return-Path: Received: from cires.colorado.edu (cires.Colorado.EDU [128.138.136.5]) by grids.ucs.indiana.edu (8.10.2+Sun/8.10.2) with ESMTP id g4GFntu12642 for ; Thu, 16 May 2002 10:49:55 -0500 (EST) Received: from poincare (doorstop.Colorado.EDU [128.138.136.107]) by cires.colorado.edu (8.9.3/8.9.3/ITS-5.0/standard) with SMTP id JAA18369 for ; Thu, 16 May 2002 09:50:26 -0600 (MDT) Message-ID: <000a01c1fcf1$3d121e40$b48810ac@cires.edu> Reply-To: "John B Rundle" From: "John B Rundle" To: "Geoffrey Charles Fox" Subject: I have a question about MPI Date: Thu, 16 May 2002 09:49:17 -0600 Organization: CIRES - University of Colorado MIME-Version: 1.0 Content-Type: multipart/alternative; boundary="----=_NextPart_000_0007_01C1FCBE.F25F4440" X-Priority: 3 X-MSMail-Priority: Normal X-Mailer: Microsoft Outlook Express 6.00.2600.0000 X-MimeOLE: Produced By Microsoft MimeOLE V6.00.2600.0000 Content-Length: 4195 This is a multi-part message in MIME format. ------=_NextPart_000_0007_01C1FCBE.F25F4440 Content-Type: text/plain; charset="iso-8859-1" Content-Transfer-Encoding: quoted-printable regarding the Broadcast operation ("BCAST")...my question is in general, = is it a blocking operation? ...i.e., there is a set of processes operating = on the nodes, and they all will get to the BCAST statement eventually, but presumably at different clock times. Do they all wait until the = process with the original copy of the data broadcasts it, and do they then have to confirm reception, before all the processes can proceed from that = point? I would assume they would, since then a process might not be operating with the latest, updated value of the variable....if not, I shall have = to=20 use blocking sends and receives, which makes the coding mildly more=20 tricky... _________________________________________ John Rundle ** Until September 1, 2002 ** Professor of Physics Director, Colorado Center for Chaos & Complexity 216 UCB University of Colorado Boulder, CO 80309 (303)-492-5642 (Tel) (303)-492-5070 (FAX) rundle@cires.colorado.edu ** After September 1, 2002 ** Interdisciplinary Professor of Physics, Geology and Engineering Director, Center for Computational Science and Engineering c/o Department of Physics University of California One Shields Avenue Davis, CA 95616-8677 (530) 752-1500 (Tel) (530) 752-4717 (FAX) _________________________________________ ------=_NextPart_000_0007_01C1FCBE.F25F4440 Content-Type: text/html; charset="iso-8859-1" Content-Transfer-Encoding: quoted-printable
regarding the Broadcast operation = ("BCAST")...my=20 question is in general, is
it a blocking operation?  ...i.e., = there is a=20 set of processes operating on
the nodes, and they all will get to the = BCAST=20 statement eventually,
but presumably at different clock = times.  Do=20 they all wait until  the process
with the original copy of the data = broadcasts it,=20 and do they then have
to confirm reception, before all the = processes can=20 proceed from that point?
I would assume they would, since then a = process=20 might not be operating
with the latest, updated value of the=20 variable....if not, I shall have to
use blocking sends and receives, which = makes the=20 coding mildly more
tricky...
_________________________________________
 
John Rundle
** Until September 1, = 2002=20 **
Professor of Physics
Director, Colorado Center for Chaos &=20 Complexity
216 UCB
University of Colorado
Boulder, CO =20 80309
 
(303)-492-5642  = (Tel)
(303)-492-5070 =20 (FAX)
rundle@cires.colorado.edu
 
** After September 1, 2002 = **
Interdisciplinary=20 Professor of Physics, Geology and Engineering
Director, Center for=20 Computational Science and Engineering
c/o Department of = Physics
University=20 of California
One Shields Avenue
Davis, CA  = 95616-8677
 
(530) 752-1500 (Tel)
(530) 752-4717=20 (FAX)
_________________________________________
------=_NextPart_000_0007_01C1FCBE.F25F4440--