Issue 85, 10 December 2009

Featuring


Using the NAG Toolbox for MATLAB to solve a PDE over a 2D region


The latest release of The NAG Toolbox for MATLAB® is available for both 32 bit and 64 bit versions of Linux (Microsoft Windows due in 2010). Containing more than 1,400 routines, the NAG Toolbox for MATLAB makes the full functionality of the NAG Library available via the MATLAB environment, thereby enhancing the basic functionality that MATLAB has to offer in many applications areas.

Using the NAG Toolbox for MATLAB, Part 4 discusses a single larger-scale application which makes use of several Toolbox functions to solve a partial differential equation over a two-dimensional region. The application has been made available as part of the NAG Toolbox for MATLAB Mark 22.

Previous articles in this series (Using the NAG Toolbox for MATLAB, Using the NAG Toolbox for MATLAB Part 2 and Using the NAG Toolbox for MATLAB Part 3) discussed some of the contents of the NAG Toolbox, presenting separate examples of the use of specific routines in areas such as surface fitting, function minimization, root finding, interpolation, quadrature and random number generation.


NAG Customer Survey 2009 ' Complete and receive a pedometer


Vital to the continued improvement in our products and evolution of our products and services is to listen to the most important people to us ' our users.

The NAG Customer Survey 2009 gives you the perfect opportunity to tell us your thoughts on our software and service and ultimately help shape future direction and functionality.

Fundamental to NAG's ethos is collaboration; please help us improve by collaborating with us on this survey. To begin the survey visit https://www.surveymonkey.com/s/6X7VHWD

To thank you for your time and thoughts we'll send you a pedometer for completing the survey. To give you an idea of time, we estimate that the survey will take approximately 15 minutes to complete.

* Please note that this survey invitation is for users of NAG software and is intended for completion only once. If you receive an email invitation to complete the survey over the next week or so, and have already completed it, please ignore the email.


Supercomputing: Should programming supercomputers be hard?


Many in the supercomputing community want programming to be made easier, but the fundamental issue is far more complex than that, says Andrew Jones, VP HPC Business, NAG.

Most people agree that programming parallel computers is hard, especially if performance at scale is required ' that is, if it involves a large number of processors. So it is hardly surprising that the question of how to make programming supercomputers easier is a popular topic at high-performance computing (HPC) conferences.

It is unarguable that parallel programming skills must become more common for two reasons. First, parallel hardware ' whether multicore processors or graphical process units ' is here for the foreseeable future. Secondly, wider use of HPC is critically important to scientific and industrial advancement. I define HPC as computing at a performance level substantially beyond a typical individual workstation.

Read the full article here.


HECToR dCSE Team Enhances Speed and I/O Performance of Oceanography Code - NEMO


The UK's national supercomputing facility, HECToR, is used by scientists needing capability supercomputing resources. A substantial part of the Research Councils' six-year (2007-2013) funding for HECToR is devoted to the Computational Science and Engineering (CSE) support provided by NAG and especially to the distributed CSE (dCSE) programme which, through lightweight peer review, funds focused multi-month performance and scalability projects in response to proposals from users. The first dCSE projects to complete are providing excellent examples of what can be achieved through dedicated CSE effort, with dramatic improvements in code performance and scalability which could potentially save millions of pounds and allow significant new science to be undertaken.

The objectives of this dCSE project were to improve the performance of the NEMO (Nucleus for European Modelling of the Ocean) code for scalar MPP architectures and reduce the amount of storage resource required. NEMO is a modelling framework for oceanographic research, operational oceanography seasonal forecast and climate studies. The key tasks of the project were (a) to address the potential I/O bottleneck arising from the more complex models, higher spatial resolutions and larger numbers of processors needed by the researchers; and (b) to investigate the performance of nested models in NEMO (which enable different parts of the ocean to be modelled with different resolution within the same global model).

The investigations into gridding and the removal of land-only cells performed under this dCSE project resulted in significant reductions to the Allocation Unit (AU) usage (notional cost) for a given simulation, by as much as 25% at larger processor counts, and a reduced runtime. NEMO has been converted to use netCDF 4.0 for its main output files resulting in a reduction in output file size of up to 3.55 times relative to the original netCDF 3.x code. It is expected that a production length research simulation should benefit in reduced run time also, due to fewer I/O bottlenecks resulting from the reduced I/O data sizes. Dr Andrew Coward, who is the manager of the Global Ocean Modelling Consortium on HECToR, estimated that their group used around 6M AUs running NEMO last year. Reducing the wall clock time of NEMO by up to 25% could result in a saving in notional cost of AUs by as much as '95,000 per year (up to '400,000 for the remainder of the service), for only six months of person effort.

Other consortia using NEMO on HECToR have used around 40M AUs over the same period. If the code modifications are accepted into the main code base, then these other users could benefit too, leading to the possibility of multi-million pound savings overall. Read the full story here


NAG Quant Day 2009: Watch talk recordings


On 22nd October NAG staged the fifth highly successful Quant Day in collaboration with Wilmott magazine. The NAG Quant Days always attract high quality speakers from academia and finance. This years event featured a diverse set of presentations on the subjects of Nearest Correlation Matrix and Using GPUs for computational finance from Professor Nick Higham, University of Manchester and Professor Mike Giles, University of Oxford.

If you didn't get the chance to attend the event but would like to watch the presentations, please visit our website here and click on 'View a full recording of the talk' links.

If you would like to ensure that you receive the first wave of invitations to next year's NAG Quant Day, please email us at nagnews@nag.co.uk.


Out & About with NAG


For more information on any of the above events visit NAG's ‘Out & About’ webpage


New NAG product implementations


The NAG Fortran Library, Mark 22 is now available for the following platforms:

  • Sun Solaris 64-bit using the Sun f95 Version 8.2'

The NAG Fortran Compiler, Release 5.2 is now available for the following platform:

  • Sun SPARC Solaris using the Sun C v 5.5 compiler

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.co.uk/doc/inun.asp


Managing your subscription


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.co.uk/NAGNews/index.asp.


NAGNews - Past Issues


We provide an online archive of past issues of NAGNews. For editions prior to these, please contact us.