Welcome to NAGNews

Issue 67, 22 November 2007

In this issue:

• Top Story - NAG and AMD co-develop latest version of AMD Core Math Library
• People News - NAG honours retiring board member and colleagues
• Product News - NAG products for Apple Intel Mac
• General Information - NAG MSc Prize News
• Product Spotlight - Maple-NAG Connector and Webinar Invitation
• Forthcoming Events - UK
• Tips & Hints - Getting the most from your NAG Library
• Product News - New product implementations

NAG has teamed up once again with AMD to co-develop the next version of the AMD Core Math Library (ACML). ACML is a highly-tuned mathematical and statistical components library developed for high performance computing, particularly for users of 64-bit AMD systems. The library provides a suite of Basic Linear Algebra Subprograms (BLAS) and LAPACK routines. It also features a comprehensive set of additional components including Fast Fourier Transforms and Random Number Generators.

New features in version 4.0 of ACML include:

* Support for AMD Family 10h (Barcelona) processors – double and single precision BLAS and LAPACK numerical routines have been tuned to deliver enhanced performance on quad core Barcelona systems

* Upgraded to include the latest LAPACK routines (version 3.1.1) and all LAPACK 3.1.1 bug fixes and enhancements applied

To learn more about NAG's Consulting Service visit our website http://www.nag.com/bdu/consultancy.asp

During NAG's recent AGM colleagues and company and board members paid tribute to Steve Hague, NAG's first employee, Emmanuel Vergison, who has served on NAG's Board of Directors for over 15 years and Sven Hammarling, longstanding NAG employee. Steve, Emmanuel and Sven have contributed in many ways to NAG's history and those attending the meeting enjoyed heartfelt recollections about them from Rob Meyer, CEO of NAG, Brian Ford, Founder Director of NAG and Richard Field and Anne Trefethen, former and current Chairman of NAG respectively.

Steve formally retired from the position of Chief Operating Officer of NAG Ltd in June 2007, however he continues to work on a consulting basis with NAG on several strategic projects. Over his 36 years of service to NAG, Steve has had a number of technical, administrative and managerial roles, as the company developed into a world leader in its field. The company owes a debt of gratitude to Steve; the contributions he has made to the organisation are immense.

Emmanuel ably served NAG as a board member for many years, giving generously of his own time whilst pursuing a successful career at Solvay S.A in Belgium. Among his other many achievements he has lectured at various Universities in France and Belgium.

In the morning of the AGM, Sven Hammarling, former Principal Technical Consultant at NAG gave an interesting insight into the world of numerical software development. His career spanned his time at NAG and before that at the UK's National Physical Laboratory. Anne Trefethen noted that Sven is the most highly cited researcher in his field, his achievements and dedication to NAG are immense. Sven formally retired from NAG in December 2006, but remains a very active member of the NAG product development team.

We are pleased to be able to honour Steve, Sven and Emmanuel and thank them for their outstanding service to NAG.

NAG is delighted to announce the first NAG Library for users of the Apple Intel Mac platform. This landmark implementation adds to the large list of existing implementations of the NAG Library. Initially the library for Apple Intel MAC is available for Intel Fortran under Mac OS X (32-bit), however over the next few months we plan to roll out other NAG products for Apple Intel Mac. The planned products will include:

* The NAG Fortran Library, Max OS X 64, for Intel and l.c.fortran compilers
* The NAG C Library, Mac OS X 64, gcc compiler
* The NAG C Library, Mac OS X 64, gcc compiler
* The NAG f95 Compiler, Mac OS X 64

The NAG Library joins the existing suite of products for Apple Intel Mac which includes the NAG f95 Compiler that was released in September. Keep up to date with news about NAG products for Apple Intel MAC in future issues of NAGNews.

To learn more about how the NAG Library can benefit you, visit our website http://www.nag.com/numeric/numerical_libraries.asp

NAG has a longstanding history of collaboration with the University of Manchester. This includes members of the School of Mathematics making algorithmic contributions to the globally renowned NAG Library and also serving on NAG's technical advisory team. This contribution is reciprocated with NAG experts supporting PhD students in their research. So when NAG wanted to establish a recognition award it made perfect sense to collaborate with the University of Manchester's School of Mathematics.

The recipients of the first NAG Prize awards, presented in March 2007, have since gone on to further their studies. Ruediger Borsdorf, who was awarded the NAG Prize for Applied Numerical Computing, completed a thesis on the nearest correlation matrix, under his mentor Professor Nick Higham. He has now returned to Germany to continue studying for a degree in Technical Mathematics at The Chemnitz University of Technology, from which he interrupted to study for the MSc in Manchester.

Talking about the NAG Prize Ruediger commented “I think the NAG Prize is a great way of supporting and promoting students and their work. It also encourages healthy competition between students which should result in excellent work”.

The next NAG Prize will be awarded to students that achieve the best performances in the MSc in Mathematical Finance and the MSc in Mathematics and Computational Science which both started in September this year. Award winners will be announced in a future issue of NAGNews.

The Maple-NAG Connector was developed by NAG and Maplesoft to extend mathematical functionality to users of Maple, and to enable the use of routines from the NAG C Library from within Maple.

It is an integrated environment to make code development easier. With this connector, users can eliminate the complexities of conventional programming and save time with simplified calling sequences and searchable documentation. The entirety of the NAG C Library at Mark 7 is available in seamless combination with Maple's mathematical analysis and visualization tools.

NAG and Maplesoft are hosting a webinar that will discuss High Performance Application Development Using Maple, the NAG Library, and the new Maple-NAG Connector on 18 December 2007. If you would like to attend the webinar email us at mailto:nagnews@nag.co.uk and we'll send you an invitation.

To learn more about the Maple-NAG Connector or to download a trial version visit our website

• 18th Daresbusy Machine Evaluation Workshop - Runcorn, UK
  27-28 November 2007
  
  

  NAG will be attending the 18th Machine Evaluation Workshop on 27-28 November hosted by EPSRC as part of the Distributed Computing Support Programme. This Workshop is now established as a leading national event dedicated to distributed high performance scientific computing. The principal objective is to encourage close contact between the research communities from the Mathematics, Chemistry, Physics and Materials Programmes of EPSRC and the major vendors of workstations, software and peripherals.

  Edward Smyth, Senior Technical Consultant at NAG will be presenting an “Overview of HECToR” at the event. For more details visit the event website at http://www.cse.scitech.ac.uk/disco/mew18/index.html
  
  
• NAG Northwest Seminar - University of Manchester, UK
  28 November 2007
  
  

  The Numerical Algorithms Group (NAG) is holding a seminar at the University of Manchester, on 28 November 2007.

  The event promises to be immensely informative for anyone using mathematical and statistical techniques. Technical sessions will appeal to those wanting to acquire a more in-depth understanding of numerical algorithms and wishing to learn about advances in technical computing.

  The seminar will be held from 11.00 am to 4.00 pm and will comprise an extended technical session on NAG's numerical libraries and toolboxes with specific sessions on numerical algorithms for the biosciences and finance.

  We expect this to be a very popular seminar and whilst the event is free to attend, seating is limited, so to ensure a place please register at http://www.nag.com/market/northwestseminar2007.asp
  
  
• NAG is delighted to support the following finance based events being held at Brunel University, London.
  
  Credit & Interest Rate Workshop - 27 November 2007
  Asset Liability Management/Liability Driven Investment - 28 November 2007
  Extreme Value Theory and Copulas - 30 November 2007
  

  For more details please visit the organiser's website http://www.carisma.brunel.ac.uk/forthcomingconf.html - to register your attendance visit http://www.unicom.co.uk/registration.asp

  
  
  

With so many NAG Library implementations available, some users are asking about the alternative libraries provided.

Very often NAG provides at least two alternative libraries; one might feature ‘nag’ in the name and the other ‘acml’ or ‘mkl’. The appropriate User's Note goes into greater and more specific detail, but the general concept is simple.

These alternative libraries contain specific and optimised implementations of the BLAS and LAPACK. For example, “acml” is optimized for use with processors from AMD, and “mkl” with those of Intel. Because they are associated with a specific processor or system vendor, they are often referred to as “performance” or “vendor” libraries. If such a library is provided by NAG, and you have the appropriate computing environment for it (as described in the User's Note), in general we would recommend this as the appropriate library to choose for normal working. With such a library, if a user calls the BLAS DGETRF for example, then the ACML, MKL or other vendor BLAS (if present) will be called directly. If the user calls instead F07ADF, then this routine, internally, makes calls to BLAS routines such as DGEMM which reside in the performance libraries.

If, however, a performance library alternate is not available or the user elects to use a ‘pure’ nag library, then no optimised BLAS or LAPACK routines are called. Instead high level NAG code is used. The NAG versions are present to prove correctness of operation and are not designed, written or specially optimised (above the standard options described in the Installer's Note) for performance. This second library is therefore ‘steady and reliable’, rather than ‘fast and efficient’, but as we have tested the latter just as thoroughly as the former, the optimised “performance” library is the appropriate first choice if available, with the NAG-provided code as a slower but reliable alternative in case it is not.

Previous Tips & Hints can be found on our website http://www.nag.com/techtips/index.asp.

NAG is committed to offering new implementations of its broad range of numerical and statistical software components and compilers and tools. The following implementations for NAG products have become available since the last issue of our newsletter:

The NAG C Library (http://www.nag.com/numeric/CL/CLdescription.asp, Mark 8 is now also available for the following platform:

• AMD-64 Windows64 using the Microsoft Visual C++ Compiler, Double Precision
• X86-32 Linux using the gcc 3.2.2 Compiler, Double Precision

The NAG Fortran Library http://www.nag.com/numeric/FL/FLdescription.asp, Mark 21 is now also available for the following platform:

• Apple Intel Mac OS X 32-bit using the Intel Fortran v9.1 Compiler, Double Precision

For full details of these and all other available implementations, visit the NAG site. Comprehensive technical details of each implementation are given in the relevant Installation and User Notes at http://www.nag.com/doc/inun.asp

Please feel free to forward this newsletter to colleagues or to post it to your company's Intranet sites.

To subscribe, unsubscribe or learn more about managing your subscription, please see: http://www.nag.com/NAGNews/index.asp.