EVENTS COMING UP

• Tue Nov 8, 9:00 am: MAINTENANCE SHUTDOWN FOR FILE SYSTEM

  This is the first of two shutdowns intended to improve certain aspects of system performance and data centre stability. File systems are the focus of this first downtime. The most visible change for users is that their home and scratch directories will be moved. See “SYSTEM NEWS” below for details.

  Note that all logins and jobs will be killed at that 9:00 am. We expect to be back in the evening.

• Wed Nov 9, 12:00 noon: SNUG (SCINET USERS GROUP) MEETING

  The SciNet Users Group (SNUG) meetings are every month on the second Wednesday, and involve pizza, user discussion, feedback, and a one or two short talks on topics or technologies of interest to the SciNet community.

  This time, we will have

  • “How to deal with the new file system setup at SciNet” (TechTalk by SciNet)
  • User discussion
  • Pizza!

  Sign up at https://support.scinet.utoronto.ca/courses/?q=node/35

• Dec 14, 2011, Jan 11/Feb 8/Mar 14/Apr 11/May 9, 2012: FUTURE SNUG MEETINGS

  We are still looking for users (students, postdocs, staff, faculty, it does not matter) willing to giving a short talk (20-30 minutes) about interesting work that they did on SciNet clusters and how they did it! If you are up for it, email support@scinet.utoronto.ca.

  More info on future SNUGs and sign-up at https://support.scinet.utoronto.ca/courses/?q=SNUG_Meetings

• Wed Nov 23: VISUALIZING DATA WITH PARAVIEW

  ParaView is a powerful open-source tool for analyzing and visualizing large multi-dimensional datasets. It is based on the Visualization Toolkit (VTK) and can run on all major platforms (Linux, Mac, Windows) both as a standalone application and as a front-end client connecting to a remote parallel server. In this two-hour long introduction I will talk about the basic workflow of a ParaView session, importing datasets in various formats, using filters, and scripting for automating ParaView tasks. This tutorial will have many hands-on exercises, so attendees are encouraged to bring a laptop with the latest copy of ParaView installed (http://www.paraview.org/paraview/resources/software.html). All sample codes and data will be provided.

  This workshop will be given at SciNet by our colleague from UOIT, Alex Razoumov.

  Sign-up at https://support.scinet.utoronto.ca/courses/?q=node/46.

• Fri Nov 4,11,18,25, 2011: SCIENTIFIC COMPUTING 1: SOFTWARE DEVELOPMENT AND DESIGN

  Note: in contrast to an earlier announcement, the lectures will be given on Fridays, from 9:30 to 11:30.

  See below for more details of the course. Sign-up for this part of the course is now closed.

• Fall/Winter: SCIENTIFIC COMPUTING COURSE

  Many computational projects start off with knowledge of the science you want to do, and with a bit of programming experience. It can be an arduous journey to get to a (maintainable) piece of code which you trust to compute the right thing. This course is aimed at reducing your struggle, and make you a more efficient computational scientist. Topics include well-established best practices for developing software as it applies to scientific computations, common numerical techniques and packages (so you don’t reinvent the wheel), and aspects of high performance computing.

  The course consists of three parts: Part 1: Scientific Software Development & Design (Nov 4,11,18,25) Part 2: Numerical Tools for Physical Scientists (Jan 13,20,27 & Feb 3) Part 3: High Performance Scientific Computing (Feb 10,17 & Mar 2,9) + A wrap-up lecture on Mar 16. Each part consists of four lectures of two hours. You can sign up for separate parts, or for the course as a whole.

  More details, including the full syllabus, grading, and sign-up, can be found at

  http://wiki.scinethpc.ca/wiki/index.php/Scientific_Software_Development_Course

  Note that these parts can be taken as “mini-courses” (AST3100) by astrophysics graduate students and as “modular courses” (PHYS2109) by physics students.

  Sign up for the full course and for part 1 is closed, but you can still sign up separately for parts 2 and 3 (see below) that will be give in the Winter.

• Mon Dec 12, 2011: INTRODUCTION TO GPGPU WITH CUDA

  The goal of this course is for incoming students, new to GPGPU but familiar with scientific programming in C, to leave being able to start writing simple kernels for their own problems, and understand the tools, techniques and libraries that will be needed to improve and optimize the results.

  More info and sign up at https://support.scinet.utoronto.ca/courses/?q=node/42

• Fri Jan 13,20,27, Feb 3, 2012: SCIENTIFIC COMPUTING 2: NUMERICAL TOOLS FOR PHYSICAL SCIENTISTS

  Sign-up: https://support.scinet.utoronto.ca/courses/?q=node/44

• Fri Feb 10,17, Mar 2,9, 2012: SCIENTIFIC COMPUTING 3: HIGH PERFORMANCE SCIENTIFIC COMPUTING

  Sign-up: https://support.scinet.utoronto.ca/courses/?q=node/45

SYSTEM CHANGES

• On November 8, we will be adding disks to the scratch file system and making scratch span both of our DDN controllers. The performance of /scratch should improve as a result of more spindles and the use of a second controller while the available space will increase by at least 40%.
• Furthermore, on November 8, 2011, the home, scratch, project and hpss file systems will be restructured (note: not all users have access to the latter two). As a consequence, users’ files will reside in different locations than before. The home and scratch file system will be group-based, and groups will furthermore be clustered by the initial letter of the group name. For instance, the current home directory of user ‘resu’ in group ‘puorg’ would move from /home/resu to /home/p/puorg/resu. Likewise, /scratch/resu would be moved to /scratch/p/puorg/resu.

  Users are responsible for making sure all their scripts and applications only use relative paths, or use the predefined variables $HOME, $SCRATCH and $PROJECT.

  Note that the TechTalk of the Nov 9 SNUG will address the issue how to deal with this new setup.

• The High-Performance Storage System (HPSS) goes into full production with a concurrent change in /project policies. Users with storage allocations greater than 5 TB will find all their former /project files will now reside in HPSS and their /project quotas will be reduced to 5 TB.
• GPC: an OS update from CentOS 5.6 to CentOS 6 is being prepared, which will include updates to other programs (perl,gcc,python) as well. The ARC already uses the newer OS, and a few of the gpc nodes are using this as a test already, while we are in the process of porting all the modules to the new OS. We encourage users to try the new environment out using the instructions on the wiki.
• SciNet has a new public web site, www.scinethpc.ca, with a new look. The wiki and portal sites can now be accessed though the same domain, as wiki.scinethpc.ca and portal.scinethpc.ca, respectively. The courses site remains at https://support.scinet.utoronto.ca/courses for now.

  Come take a look at our new website!

ADDED TO THE WIKI IN OCTOBER

 All new wiki content below is listed and linked on the main page: 
 http://wiki.scinethpc.ca/wiki/index.php/SciNet_User_Support_Library#What.27s_New_On_The_Wiki)

• Tips on testing your codes for CentOS 6
• Syllabus of the Scientific Software Development Course
• Currently installed modules for the CentOS 6 nodes
• Slides of the SNUG talk by Manuel Saldaña on “MPI as a programming model for High-Performance Reconfigurable Computers”
• How to distinguish between CentOS 5 and 6 in your .bashrc

WHAT ELSE HAPPENED AT SCINET IN OCTOBER?

• Oct 6: Information session for NRAC applications
• Oct 12: SNUG meeting with a TechTalk by Manuel Saldaña (Arch Es Computing) on “MPI as a programming model for High-Performance Reconfigurable Computers”
• Oct 26: Introduction to scientific programming with modern FORTRAN