Computational Support for HPC Systems and Clusters
NAG numerical software engineers deliver expert software design services and programming support to researchers and industry users in many sectors and disciplines.
NAG specialists are provided to improve the runtime performance of your application using the effective distributed Computational Science and Engineering (dCSE) methodology. This quickly results in programmers and application users being more effective when using the capabilities of their own computing clusters or HPC facilities. Activities already undertaken have included mentoring in practical HPC skills, converting research code into robust and reliable production code and porting software to new compute platforms.
Examples of projects include:
- Scaling a finite element engineering package to large clusters
- Extending expressive and scalable finite element simulation beyond 1000 cores
- Tuning a molecular dynamics simulation of multi-scale flows for GPUs
- Updating HPC software for massively parallel simulation of 3D electromagnetic nano-structures
- Parallelising uncertainty quantification with differential geometric population Markov chain Monte-Carlo method
- Improving tensor manipulation and storage on HPC
- Reworking tight binding molecular dynamics for a CPU/GPU cluster
- PC implementation of agent-based models for cell cultures
Read these reports for more details.
Tailored services are available for universities; for commercial clients and for regional consortia. Please contact us for more information.