From owner-chemistry@ccl.net Wed Oct 25 07:01:01 2006 From: "Imran Predhanekar zmatrix#%#rediffmail.com" To: CCL Subject: CCL: include NBO, how? Message-Id: <-32865-061025065712-29137-fKHVTPYXK0WcnWtsMditwA-*-server.ccl.net> X-Original-From: "Imran Predhanekar" Date: Wed, 25 Oct 2006 06:57:11 -0400 Sent to CCL by: "Imran Predhanekar" [zmatrix(!)rediffmail.com] greetings I want to do NPA with gamess. Since it is not free, I downloaded a folder > from sources of CCL. Can you tell me how to include this with gamess. the manual is not quite clear. Or is there any other way to do NPA? Thanks in advance Imran From owner-chemistry@ccl.net Wed Oct 25 09:48:00 2006 From: "Nuno A. G. Bandeira nuno.bandeira-x-ist.utl.pt" To: CCL Subject: CCL: include NBO, how? Message-Id: <-32866-061025094229-13628-++6jXE/YOA/nYVClny3sqA|,|server.ccl.net> X-Original-From: "Nuno A. G. Bandeira" Content-Transfer-Encoding: 7bit Content-Type: text/plain; charset=ISO-8859-1; format=flowed Date: Wed, 25 Oct 2006 14:05:07 +0100 MIME-Version: 1.0 Sent to CCL by: "Nuno A. G. Bandeira" [nuno.bandeira .. ist.utl.pt] Imran Predhanekar zmatrix#%#rediffmail.com wrote: > Sent to CCL by: "Imran Predhanekar" [zmatrix(!)rediffmail.com] > greetings > I want to do NPA with gamess. Since it is not free, I downloaded a folder > >> from sources of CCL. Can you tell me how to include this with gamess. the > > manual is not quite clear. > Or is there any other way to do NPA? The free NBO cannot be embedded in GAMESS. It's too primitive. You can however compile gennbo 3.0 (you can download a precompiled binary for Windows from my website) but you must have a file .47 archive which I'm not sure gamess will give you. -- Nuno A. G. Bandeira, AMRSC Graduate researcher and molecular sculptor Inorganic and Theoretical Chemistry Group, Faculty of Science University of Lisbon - C8 building, Campo Grande, 1749-016 Lisbon,Portugal http://cqb.fc.ul.pt/intheochem/nuno.html Doctoral student /./ IST,Lisbon -- From owner-chemistry@ccl.net Wed Oct 25 11:46:00 2006 From: "Sambit Gaan sgaan a nicholaspiramal.co.in" To: CCL Subject: CCL: Require statics about GROMACS & AMBER MD simulation on cluster Message-Id: <-32867-061025070847-1968-rkNcuERPBErIcNgvPyaFjg]|[server.ccl.net> X-Original-From: "Sambit Gaan" Date: Wed, 25 Oct 2006 07:08:46 -0400 Sent to CCL by: "Sambit Gaan" [sgaan*nicholaspiramal.co.in] Dear All, I am using GROMACS for sometime to do Molecular Dynamics. I am a beginner in the HPC field. I have created a linux cluster having 5 nodes and each node is having 2 GB of RAM and Xeon processor. But sometime back I met one person who has got a lot of experience on HPC systems and MD. So he told me that it is better to use AMBER software for running large MD jobs than GROMACS. He told me that AMBER performs well on large clusters, where as GROMACS performance degrades as you go on increasing the number of processors. So can anybody give me some benchmark statics comparing AMBER with GROMACS. Tahnks. Sambit Gaan From owner-chemistry@ccl.net Wed Oct 25 22:53:01 2006 From: "Lubos Vrbka lists|a|vrbka.net" To: CCL Subject: CCL: Require statics about GROMACS & AMBER MD simulation on cluster Message-Id: <-32868-061025225106-22483-dSTmc4k7q7/tZkHuJp78Ew a server.ccl.net> X-Original-From: Lubos Vrbka Content-Transfer-Encoding: 7bit Content-Type: text/plain; charset=UTF-8; format=flowed Date: Thu, 26 Oct 2006 11:50:45 +0900 MIME-Version: 1.0 Sent to CCL by: Lubos Vrbka [lists_._vrbka.net] hi, > I am using GROMACS for sometime to do Molecular Dynamics. I am a beginner in the HPC field. I have created a linux cluster having 5 nodes and each node is having 2 GB of RAM and Xeon processor. > But sometime back I met one person who has got a lot of experience on HPC systems and MD. So he told me that it is better to use AMBER software for running large MD jobs than GROMACS. He told me that AMBER performs well on large clusters, where as GROMACS performance degrades as you go on increasing the number of processors. > So can anybody give me some benchmark statics comparing AMBER with GROMACS. with single precision, gromacs is much faster than amber, that is in double precision by default. when you use double precision in gromacs, the speed is in my experience more or less comparable. the degradation of speed for gromacs *might* be seen for >32 processors, but i've never run so big simulation, so i can't really tell. consult gromacs webpage, they have some benchmark data. (by the way, i wouldn't call running calculation on 5 processors 'high performance computing'). the degradation of speed will also depend on the size of your calculation (number of atoms) and on the network interconnection between nodes - these two factors will influence both codes more or less in the same way. also it's best (maybe required, see manual - i'm not sure now and cannot check at the moment) to run gromacs jobs on number of processors that are power of 2, so in your case 4. amber had such limitation in the past, but i don't know whether it still applies. for amber, if you don't need any special features, i recommend using pmemd for parallel runs. for your cluster size, i don't think you will see any dramatic difference between the two codes. the obvious advantage of gromacs is that it is for free... :) i used to run amber jobs and still do for bio simulations (let's say up to 16 processors - my systems are too small to scale well on larger number of processors). for my other stuff i use gromacs, also because it supports some things i need and amber doesn't have. regards, -- Lubos _,,_" http://www.lubos.vrbka.net