|
|
||||
|
Bulletin 140 - 2005 July 20
1. User feedback HPCCC users are reminded there is an anonymous feedback form available at http://www.hpccc.gov.au/hpccc/helpdesk/feedback.shtml Good or bad, we encourage you to make your opinions known. 2. HPCCC and HPSC web pages The HPCCC User News and Advice menu (http://www.hpccc.gov.au/hpccc/user_news_advice/) now includes notes about new additions to the web page. The list will highlight new items for 3 months after they are first added. Recent additions are:
The HPCCC WWW pages also contain instructions for accessing the intranet from outside the CSIRO or Bureau sub-nets. The HPSC WWW pages have a new search facility. The CSIRO Data Store Userguide was updated to reflect current management strategies. 3. Upgrade of SUPER-UX to Revision 15.1 The HPCCC is planning to upgrade the operating system on the SX-6 nodes from SUPER-UX Release 13.1 to Release 15.1. The plan is to have some nodes available under SUPER-UX 15.1 in August, and have the final transition complete by the end of September 2005. There will be a stable SUPER-UX 13.1 platform through to the end of September. Information on the implications of the change will be made available soon through the WWW pages. The HPCCC plans to make the change in phases, with subsets of the nodes being converted to SUPER-UX 15.1, and made available to users for testing. It will be important for users to test their applications under the new version, to allow debugging and resolution, so that Operational and critical jobs can migrate smoothly. The queues for the nodes on the new operating system will be named bm3, bmmn3, bmrt3, etc, through the completion of conversions. These arrangements will allow for testing on the new operating system while providing unchanged 13.1 services for research and operational needs. 4. Fortran90/SX Rev.313 compiler available On Tuesday 26th July - see SYSTEM CHANGE NOTICE 2005-A017 - we plan to install Release 313 of the NEC Fortran90 compiler for the SX system as the developmental compiler. R313, released with SUPER-UX R15.1, is the latest NEC compiler release. The default compiler will continue to be R302 until all users have had the opportunity to evaluate R313. The compiler will be made available on SX nodes, tx7, gale, cherax and farrer. In the tests carried out by NEC Australia staff on sample codes from HPCCC users, there are some cases where R313 compile time has gone up 25% and one down 30%, the execution times are about even with R302. A few PSRs were corrected. Release notes will be available shortly at the HPCCC WWW pages under
User Documentation 5. Upgrade of SX MPI libraries NEC has released a significant bug fix release for the MPI libraries, and which is part of the upcoming SUPER-UX R15.1 upgrade. The HPCCC thus plans to upgrade the MPI libraries for the SX system to enable users to prepare for this change to MPI. NEC Australia has advised there are some incompatibilities between our current version, and the new version. The new version will be required for use at SUPER-UX R15.1 and all users with MPI codes will have to relink. The corrections and changes are documented in Release notes that will be available shortly at the HPCCC WWW pages under
User Documentation The plan for deployment will be advised at a later date. We are investigating mechanisms for allowing access to multiple versions of the MPI components, so that users can test the new versions prior to their becoming the default. Please take note of this important change and requirement for relinking. Further advice will appear in HPCbulls and on the HPCCC web. 6. Change to bash shell on the SX-6 nodes There is no supplied bash shell on the SX-6 nodes. In the early days of the SX-6 installation, HPCCC staff provided a bash shell by simply providing a symbolic link to the Korn shell ksh. This has several draw-backs - error messages refer to the wrong shell, and startup processing cannot be provided in non-login shells, e.g. with rsh commands to the SX-6es. With the SUPER-UX upgrade, the HPCCC plans to change the bash setup, by providing an executable which executes /bin/ksh on SX-6 nodes, and provides startup processing capabilities. This will allow bash shell users to do commands like: rsh sx600:\$WORKDIR/runme A test version implementing this capability is available now as /bin/newbash on the SX-6 nodes and the TX7s. Users of bash are asked to try this facility and report their experiences: please contact the HPCCC to have your shell changed. (Bureau users can use ypchsh on a TX7 to make the change, but CSIRO users need to contact us to have the change made for them, because of the different authentication schemes used). 7. Superseded versions of MathKeisan and perl Superseded and no longer used versions of MathKeisan and perl will be removed from SX-6 nodes on Tuesday 26th July. 8. Cluster downtime notification (CSIRO) A reminder that the CSIRO cluster burnet is scheduled to be down for maintenance and upgrades on the second Saturday of each month. The next downtime will be on 13th August. 9. NAG and Gaussian moved (CSIRO) The NAG library and Gaussian package have been moved from /usr on sx604 to /cs/home. This change should be transparent to users. 10. Latest Intel Fortran and C compilers (CSIRO) The Intel 9.0 compilers are in cherax:/cs/datastore/u/csssg/share/Intel_compilers/ CSIRO HPSC has floating licences for these compilers, and they can be used on any platform (at least Linux ia32 or ia64) in CSIRO. You are welcome to try these, and give us feedback. Please contact HPSC staff for instructions on installing these compilers on systems outside the HPSC. You will need to know how to access a licence server. The compilers will shortly be installed on the HPSC systems. 11. cherax downtimes and upgrades - disk and Linux The expansion of the disc system underlying the main file systems on cherax has been delayed, pending the arrival of further disc to allow a higher-performance configuration to be built. The disc has now been installed, and there will be further downtime, probably on Saturday 13th August. As well, we plan to upgrade the Altix operating system from Redhat Linux to SUSE Linux Enterprise Server 9.2 with SGI Propack 4 at the same time. More details will be provided soon of the changes involved. 12. Denied access for CSIRO users From Sunday afternoon 17th July or before, to about midday on Monday 18th July, CSIRO users were not able to log in to HPCCC and HPSC systems. This was because of a fault in a NEXUS Domain controller, which authenticates all logins. The problem was fixed by a re-boot of the domain controller. We plan to have a better way to diagnose and fix such problems in the future. 13. New qsub and qsubnew versions At times, we will implement new versions of the development wrapper named qsubnew. A new version was recently installed there, which corrects some problems with complex PBS directive lines. This new version will be installed as the default local qsub wrapper on Tuesday 26th July. 14. Strategies for those moving applications to multi-processing A user recently asked for advice on the best strategy for using multiple processors on a code. We have provided advice to the user, and encapsulated it in an FAQ at http://www.hpccc.gov.au/hpccc/userguides/faq/parallel_choice.php and http://intra.hpsc.csiro.au/userguides/faq/parallel_choice.php 15. MATLAB (CSIRO) CSIRO HPSC has purchased two floating licences for MATLAB. The software can potentially be run anywhere in CSIRO (including Windows and ia32 Linux platforms). "MATLAB is a high-level language and interactive environment that enables you to perform computationally intensive tasks faster than with traditional programming languages such as C, C++, and Fortran." (http://www.mathworks.com/) This software will be available under the HPSC software licensing scheme (see http://intra.hpsc.csiro.au/forms/HPSC_LicensedSoftware.pdf) to CSIRO staff. For casual use for evaluation, etc, please contact HPSC staff. The software and licences will be available soon. 16. New Altix system for iVEC (CSIRO) A Silicon Graphics Inc. 160 processor Altix supercomputer will shortly be installed at the Australian Resources Research Centre node of iVEC, the hub of advanced computing in Western Australia. The purchase was made possible by the WA Government's $3.1 million commitment to iVEC and CSIRO's High Performance Scientific Computing. It represents a huge increase in computing power and will deliver a minimum 20-fold increase in performance over the current system. The Altix is a shared-memory system with a single pool of 320 GB of RAM. Each processor can access all 320 GB. This makes parallel programming much easier, and given that the computer has almost one teraFLOP of power, ensures that all iVEC and WA CSIRO users will have the tools to help solve some of the world's toughest problems. Dr Rhys Francis, Director of High Performance Scientific Computing for CSIRO said he was pleased to be able to help with the purchase of such a significant addition in computing power for researchers in the West.
|
|
Comments to: © Copyright 2010, CSIRO Australia Use of this web site and information available from it is subject to our Legal Notice and Disclaimer and Privacy Statement |