From owner-chemistry@ccl.net Fri Feb 17 11:53:00 2012 From: "Philippe Bourassa philippe.bourassa|a|uqtr.ca" To: CCL Subject: CCL:G: Gaussian on a PC computer Message-Id: <-46324-120217093620-13954-/YdTg6+LcNFFwvqKQPRG2g---server.ccl.net> X-Original-From: Philippe Bourassa Content-Type: multipart/alternative; boundary=f46d042c647bdd628b04b929dd31 Date: Fri, 17 Feb 2012 09:36:13 -0500 MIME-Version: 1.0 Sent to CCL by: Philippe Bourassa [philippe.bourassa{=}uqtr.ca] --f46d042c647bdd628b04b929dd31 Content-Type: text/plain; charset=ISO-8859-1 The 64 bit version of Gaussian for Windows is already available, although only as a site license. 2012/2/17 Gkourmpis, Thomas Thomas.Gkourmpis : borealisgroup.com < owner-chemistry+*+ccl.net> > > Sent to CCL by: "Gkourmpis, Thomas" [Thomas.Gkourmpis{:}borealisgroup.com] > Hello > > According to my chats with the guys in Gaussian the Windows version (what > they call multicore) is a 32bit version so is able to use up to 2GB of RAM. > The issue with using all the cores in a computer is in essence a matter of > how much RAM can the program access, since Gaussian is using a system where > the program itself decides how many processors will use. In principle it > can use up to 4 but since you're limited by your RAM allocation most times > it will use anything between 1 and 3. Also don't forget that the 2GB are > total memory, so they're shared between the cores. Therefore in > calculations that need a lot of memory like SCFs or frequency you're better > off using one or two cores instead of four (the program will do that for > you). > > In all if you want to get top performance from Gaussian either get the > Linux version or pressure them to finish up and release the 64bit version > for Windows :-) > > I hope this helps > Thomas > > -----Original Message----- > > From: owner-chemistry+thomas.gkourmpis==borealisgroup.com__ccl.net[mailto: > owner-chemistry+thomas.gkourmpis==borealisgroup.com__ccl.net] On Behalf > Of willsd{:}appstate.edu > Sent: Friday, February 17, 2012 2:10 AM > To: Gkourmpis, Thomas > Subject: CCL:G: Gaussian on a PC computer > > > Sent to CCL by: [willsd(a)appstate.edu] > While I am not sure about the windows version, the 64-bit (intel) linux > version of G09 I use at my university will > certainly use all 8 cores I have on a shared memory node. I think you > must use the linda parallel version to run > jobs on multiple compute nodes, but I think the normal version requires > nothing special to run on shared cores > using the same memory space. > Steve Williams > > ----- Original Message ----- > > From: "Ronald Cook cookrl(-)tda.com" > Date: Thursday, February 16, 2012 7:16 pm > Subject: CCL:G: Gaussian on a PC computer > To: "Williams, Steve " > > > Hi Rapha > > > > I have been running G09W calculations on a Windows XP system using > > an Intel > > i7-2700K processor. The chip is normally clocked at 3.5GHz, but we > > havebeen running calculations for several months now with the system > > overclocked at 5GHz with no overheating problems. We see about 30% > > decrease in computational time. Furthermore, G09W does not use > > hyperthreading. So If you go into the task manager and select > > processesand right click on G0W9 and then uncheck processors 4-7 > > (which are virtual > > anyway) and just use processors 0-3 you can get an additional 30% > > decreasein computational time. > > > > Ron > > > > On Thu, Feb 16, 2012 at 2:53 PM, Raphael Martinez > > raphaelmartinez1983++gmail.com wrote: > > > > > > > > Sent to CCL by: "Raphael Martinez" [raphaelmartinez1983:- > > :gmail.com]> Hi everyone: > > > If I have an Intel quad core, 16GB ram computer with either linux or > > > Windows, does it Gaussian uses all 4 cores if I request to do so > > in the > > > input file? > > > If not, what would be the best processor to have in a personal > > computer to > > > run Gaussian? > > > Thanks > > > RaphaI> > > > > > > > > > > > > -- > > Ronald L Cook > > Principal Scientist > > TDA Research, Inc. > > cookrl++tda.com > > 303-940-2302http://www.ccl.net/chemistry/sub_unsub.shtmlhttp://www.ccl.net/spammers.txtThisEmail and any files transmitted with it are confidential and > intended solely for the use of the individual or the entity to whom > it is addressed. If you have received this Email by error, please > notify the sender and delete the material from any storage device. > Borealis extends no warranties and makes no representations as to > the accuracy or completeness of the information provided. It is the > customer's responsibility to inspect and test our products and > technical advice in order to satisfy itself as to the suitability > of the products and technical advice for the customer's particular > purpose.> > > --f46d042c647bdd628b04b929dd31 Content-Type: text/html; charset=ISO-8859-1 Content-Transfer-Encoding: quoted-printable The 64 bit version of Gaussian for Windows is already available, although o= nly as a site license.

2012/2/17 Gkourmp= is, Thomas Thomas.Gkourmpis : borealis= group.com <owner-chemistry+*+ccl.net>

Sent to CCL by: "Gkourmpis, Thomas" [Thomas.Gkourmpis{:}borealisgroup.com]
Hello

According to my chats with the guys in Gaussian the Windows version (what t= hey call multicore) is a 32bit version so is able to use up to 2GB of RAM. = The issue with using all the cores in a computer is in essence a matter of = how much RAM can the program access, since Gaussian is using a system where= the program itself decides how many processors will use. In principle it c= an use up to 4 but since you're limited by your RAM allocation most tim= es it will use anything between 1 and 3. Also don't forget that the 2GB= are total memory, so they're shared between the cores. Therefore in ca= lculations that need a lot of memory like SCFs or frequency you're bett= er off using one or two cores instead of four (the program will do that for= you).

In all if you want to get top performance from Gaussian either get the Linu= x version or pressure them to finish up and release the 64bit version for W= indows :-)

I hope this helps
Thomas

-----Original Message-----
> From: owner-chemistry+thomas.gkourmpis=3D=3Dborealisgroup.com__ccl.net [mailt= o:owner-chemistry+tho= mas.gkourmpis=3D=3Dborealisgroup.com__ccl.net] On Behalf Of willsd{:}appstate.edu
Sent: Friday, February 17, 2012 2:10 AM
To: Gkourmpis, Thomas
Subject: CCL:G: Gaussian on a PC computer


Sent to CCL by: [willsd(a)appstate.edu]
While I am not sure about the windows version, the 64-bit (intel) linux ver= sion of G09 I use at my university will
certainly use all 8 cores I have on a shared memory node. =A0I think you mu= st use the linda parallel version to run
jobs on multiple compute nodes, but I think the normal version requires not= hing special to run on shared cores
using the same memory space.
Steve Williams

----- Original Message -----
> From: "Ronald Cook cookrl(-)tda.com" <owner-chemistry() ccl.net>
Date: Thursday, February 16, 2012 7:16 pm
Subject: CCL:G: Gaussian on a PC computer
To: "Williams, Steve " <willsd() conrad.appstate.edu>

> Hi Rapha
>
> I have been running G09W calculations on a Windows XP system using
> an Intel
> i7-2700K processor. =A0The chip is normally clocked at 3.5GHz, but we<= br> > havebeen running calculations for several months now with the system > overclocked at 5GHz with no overheating problems. =A0We see about 30%<= br> > decrease in computational time. =A0Furthermore, G09W does not use
> hyperthreading. =A0So If you go into the task manager and select
> processesand right click on G0W9 and then uncheck processors 4-7
> (which are virtual
> anyway) and just use processors 0-3 you can get an additional 30%
> decreasein computational time.
>
> Ron
>
> On Thu, Feb 16, 2012 at 2:53 PM, Raphael Martinez
> raphaelmartinez1983++gm= ail.com <owner-chemistry++ccl.net> wrote:
>
> >
> > Sent to CCL by: "Raphael =A0Martinez" [raphaelmartinez1= 983:-
> :gmail.com]> Hi = everyone:
> > If I have an Intel quad core, 16GB ram computer with either linux= or
> > Windows, does it Gaussian uses all 4 cores if I request to do so<= br> > in the
> > input file?
> > If not, what would be the best processor to have in a personal > computer to
> > run Gaussian?
> > Thanks
> > RaphaI>
> >
> >
>
>
> --
> Ronald L Cook
> Principal Scientist
> TDA Research, Inc.
> cookrl++tda.com
> 303-94= 0-2302http://www.ccl.net/cgi-bin/ccl/send_ccl_messagehttp://www.= ccl.net/chemistry/sub_unsub.shtmlhttp://www.ccl.net/spammers.txtThis Em= ail and any files transmitted with it are confidential and
intended solely for the use of the individual or the entity to whom
it is addressed. If you have received this Email by error, please
notify the sender and delete the material from any storage device.
Borealis extends no warranties and makes no representations as to
the accuracy or completeness of the information provided. It is the
customer's responsibility to inspect and test our products and
technical advice in order to satisfy itself as to the suitability
of the products and technical advice for the customer's particular
purpose.



-=3D This is automatically added to each message by the mailing script =3D-=
E-mail to subscribers: CHEMISTRY+*+ccl.n= et or use:
=A0 =A0 =A0http://www.ccl.net/cgi-bin/ccl/send_ccl_message

E-mail to administrators: CHEM= ISTRY-REQUEST+*+ccl.net or use
=A0 =A0 =A0http://www.ccl.net/cgi-bin/ccl/send_ccl_message

Subscribe/Unsubscribe:
=A0 =A0 =A0http://www.ccl.net/chemistry/sub_unsub.shtml

Before posting, check wait time at: http://www.ccl.net

Job: http://www.ccl.n= et/jobs
Conferences: http://server.ccl.net/chemistry/announcements/co= nferences/

Search Messages: http://www.ccl.net/chemistry/searchccl/index.shtml
=A0 =A0 =A0h= ttp://www.ccl.net/spammers.txt

RTFI: http://www.ccl.net/chemistry/aboutccl/instructions/



--f46d042c647bdd628b04b929dd31--