These are the early results of NAG’s distributed Computational Science and Engineering (dCSE) support program for HECToR (UK’s national supercomputing facility), now consisting of over 30 dedicated application optimisation projects complementing the traditional HPC user support provided by NAG.
 
In the first project to complete, a key materials science code, CASTEP used by academic researchers and industry was enhanced with band-parallelism to allow the code to scale to more than 1,000 cores. The speed of CASTEP on a fixed number of cores was also improved by up to 4x on the original, representing a potential saving of around $3m of computing resources over the remainder of the HECToR service.

The CASTEP project showed the collaborative nature of the dCSE program, with the University of York undertaking the core development (8 person months) in conjunction with NAG HPC staff and the Science and Technology Facilities Council.
 
In another project, an ocean modelling application known as NEMO  (Nucleus for European Modelling of the Ocean with blue, white and green oceans ad adaptative mesh) underwent  optimisation including I/O techniques and variable resolution approaches to run 25% faster on relevant use cases.

This represents a $600,000 saving in computing resources for that project with potentially multi-million dollar savings across all NEMO users. The 6 person-month project was performed by a collaboration of the National Oceanography Centre and the University of Edinburgh working with NAG HPC staff.
 
A third project optimized a quantum Monte Carlo code (CASINO ) for better performance on multicore nodes by introducing shared memory techniques and hierarchical parallelism.

This resulted in performance gains of up to 4x on quad-core nodes and further performance gains from I/O optimisations for simulations using more than 10,000 cores.

Following NAG’s work, the scientists were able to run on 40,000 cores of the Jaguar Petaflops supercomputer at Oak Ridge National Laboratory. This 12 person-month dCSE project was undertaken by NAG HPC staff working with users at University College London, and is estimated to have saved the researchers around $1m in computing resources on HECToR.
 
“These three examples of HPC software projects show the real performance advantages – and cost savings – to researchers from enhancing applications to run optimally on the latest HPC machines”, said Andrew Jones, (right) VP of HPC Consulting at NAG. “Investment in application performance and algorithms appropriate to the computer architecture has now become critical for efficient use of HPC resources and users’ time.”
 
NAG’s dCSE projects enhance the performance of dozens of community HPC applications used on HECToR and all performance improvements are fed back into the community so that non-HECToR users can benefit too. Further examples can be found on HECToR .

NAG has also apened a new office in Taipei to serve the Greater China region and a simultaneous expansion of supporting staff in both Germany and France. The goal of this multi-pronged expansion is to give more users of technical computing software a better service as they access the 1600+ NAG Library mathematics and statistical routines and make use of NAG HPC consultants.

“When we set out our plan we knew having skilled NAG employees based in the key geographic regions was essential to providing service excellence,” said Ian Reid, chief commercial officer at NAG. “This is particularly true for our consultancy work and the relationships that we have with regional partners and academics.”