From jgevans at wi.mit.edu Mon Feb 16 09:53:52 2004 From: jgevans at wi.mit.edu (James Evans) Date: Mon, 16 Feb 2004 09:53:52 -0500 Subject: [CSBi-HPC] new software for p655 cluster Message-ID: <000c01c3f49c$b162bee0$8d169d12@ahola> Hi We are in the process of adding new software licenses to the p655 cluster. To start with, we plan on adding 32 base licenses for MatLab and the Optimization toolbox. Could you reply to me with your needs in terms of the various MatLab toolboxes (i.e. type and approximate number used per session)? We are also exploring use of NAMD, MatLab*P and Huygens2. Also, let me know if there are other software packages that you would like to try out. Does anyone need interactive sessions while using their software? Cheers James ---------------------------------------------------------------------- James G. Evans Ph.D. CSBi Research Scientist Whitehead MIT BioImaging Center 500 Technology Square (NE47-311) Cambridge MA 02139 U.S.A. phone 617 324 0300 fax 617 258 7226 email jgevans at wi.mit.edu ---------------------------------------------------------------------- From cfdewey at MIT.EDU Mon Feb 16 11:11:50 2004 From: cfdewey at MIT.EDU (C. Forbes Dewey, Jr.) Date: Mon, 16 Feb 2004 11:11:50 -0500 Subject: [CSBi-HPC] new software for p655 cluster In-Reply-To: <000c01c3f49c$b162bee0$8d169d12@ahola> References: <000c01c3f49c$b162bee0$8d169d12@ahola> Message-ID: <4030EBC6.9030309@mit.edu> An HTML attachment was scrubbed... URL: http://mailman.mit.edu/pipermail/csbi-hpc-users/attachments/20040216/11f14a44/attachment.htm From jgevans at wi.mit.edu Tue Feb 24 06:11:41 2004 From: jgevans at wi.mit.edu (James Evans) Date: Tue, 24 Feb 2004 06:11:41 -0500 Subject: [CSBi-HPC] Reminder: CSBi HPC UserGroup Wednesday 5pm MPI part 2 Message-ID: <000601c3fac6$fb29e000$0a00a8c0@ahola> Hi Don't forget, Kirk Jordan from IBM will be giving a more in depth look at using MPI tomorrow evening. Please contact me if you have any questions. Thanks James Meeting: CSBi High Performance Computing UserGroup Title: "MPI: An Overview of Message Passing - Part Two" Speaker: Dr. Kirk E. Jordan, Emerging Solutions Executive, IBM Life Sciences Location: 500 Technology Square (NE47) room 189 When: 5pm Wednesday February 25th Dinner will be served. ---------------------------------------------------------------------- James G. Evans Ph.D. CSBi Research Scientist Whitehead MIT BioImaging Center 500 Technology Square (NE47-311) Cambridge MA 02139 U.S.A. phone 617 324 0300 fax 617 258 7226 email jgevans at wi.mit.edu ---------------------------------------------------------------------- From aldavis at wi.mit.edu Tue Feb 24 10:12:01 2004 From: aldavis at wi.mit.edu (Al Davis) Date: Tue, 24 Feb 2004 10:12:01 -0500 Subject: [CSBi-HPC] CSBi IBM cluster - available fileystem space References: <000601c3fac6$fb29e000$0a00a8c0@ahola> Message-ID: <000101c3fae8$8dbbaef0$29289d12@wi.mit.edu> We temporarily ran out of space in one of the home directories on the WI/Bioimaging part of the cluster so I want to inform everyone about other available disk space on this part of the cluster. The cluster currently has 5 nodes and each one has local disk space that is availabe for use - it's always been there, but probably not visible nor useable by everyone. So I've created scratch directories in each one and opened up the permissions so anyone can read/write to files there. These directories are cross mounted on each node and are availabe through the following paths: path via host # path via hostname FS size (GB) /cluster/wi/1/scratch /scratch/mantra801 230 /cluster/wi/2/scratch /scratch/mantra802 230 /cluster/wi/3/scratch /scratch/mantra401 230 /cluster/wi/4/scratch /scratch/mantra402 130 /cluster/wi/5/scratch /scratch/mantra803 130 If you need permanent space for large files, create a directory with your username in one of the above paths and keep large files there so you don't fill up your home directory. If you just need temporary space while your application runs, use the local node scratch disk which will have the fastest access and is available as /scratch/`hostname` and please delete them at the end of your run. A couple of caveats about this space: 1. We don't have lots of disk space on the cluster (~1 TB total) so please be judicious with the space. 2. We currently don't have any archival backup capability on the cluster so beware. 3. We currently don't have a good network connection between the two cluster halves, so it's not easy to share large datasets between them. We're working on plans to remedy all the above, but it will be a little while till their fully implemented and in place. 4. We also have a common directory for programs and libraries which are/will be installed on the cluster (/cluster/programs) so if you have code which you think is useful for everyone, let me know and I can install it there rather than in your home directory. So far the only working code is matlab, but I'm installing namd and some other code and when it's up and running I'll let everyone know. If you encounter any problems, let me know. al -- Al Davis aldavis at wi.mit.edu Systems Manager 617.324.0519 WI/MIT BioImaging Center 500 Tech Sq, Rm 309