Groundbreaking Performance and Power Efficiency for High Performance Computing (HPC)
Developing scientific applications for solving society's complex global issues — such as vaccine discovery, climate modelling, improving food production, and advancing green energy sources — places increasing demand on computing resources that can only be met effectively by advancing supercomputing technology. As nations and regional alliances race toward exascale computing, constraints with other architectures, including lack of choice, advancement, and control, lead to limited innovation.
Arm HPC solutions, including Arm Neoverse, address the needs of the HPC community today and for the future. Arm's open IP licensing model and consistent architectural advancements provide CPU and system designers the freedom and flexibility to innovate supercomputer design —as evidenced perfectly by the collaboration between Arm, Fujitsu, and RIKEN in delivering Fugaku. No other CPU architecture today can deliver the level of performance and power efficiency required by supercomputers as Arm Neoverse.
As HPC enters a new era of technological capabilities, RIKEN and Fujitsu based the Fugaku Supercomputer in Japan on Arm, for performance, efficiency, and scalability.
Meeting the Demands of HPC Applications
Arm collaborates with the HPC community, system designers, and its ecosystem of software partners to address the complex needs of developing HPC solutions. Advantages of Arm solutions include:
- CPU micro-architecture advances such as Scalable Vector Extensions, which allow application developers to extract more parallelism and performance from applications in a developer and deployment-friendly way.
- Partnerships with all leading EDA tool and foundry providers to ensure Arm technology can be implemented in the latest, most advanced, and performant manufacturing processes.
- A legacy of delivering the most power-efficient CPU architectures, enabling more cores and higher performance in a power-efficient platform, whether deployed on-prem and at scale like Fugaku, or delivered via public cloud instances based on AWS Graviton2.
In addition, Arm has enabled a robust and performant HPC software ecosystem including full-stack support from leading vendors – from the OS, to libraries and compilers, middleware and developer debug and profiling tools.
The use cases described below are some of the more common examples for Arm-based HPC solutions.
Performance, scalability and efficiency support an ecosystem of Arm-based HPC partner solutions targeting genome sequencing, protein folding, drug discovery, and more.
Computational Fluid Dynamics
Arm-based HPC delivers accelerated CFD simulations with improved performance for more detailed experimentation and increased accuracy.
Electronic Design Automation
From design to simulation and verification, Arm-based HPC provides the performance and efficiency needed in the EDA process.
Oil & Gas Exploration
Simulation models can be finetuned and iterated upon quickly thanks to reliable software stacks and ecosystems, including compilers, toolchains and math libraries optimized for Arm Neoverse.
Government Labs and University Research
Pushing the boundaries of scientific exploration, Arm-based HPC clusters lets scientists, researchers, and educators collaborate effectively and build long-term strategic partnerships.
Efficiency Breakthroughs in HPC to Drive “Production Science,” Global Research
HPC needed economic disruption; computer scientists in the UK obliged.
How Supercomputer Performance, Power Efficiency Can Coexist
Building the world’s fastest supercomputer doesn’t mean breaking the power budget, as engineers at Fujitsu and RIKEN proved.
- NVIDIA, Ampere Computing Raise Arm 26x in Supercomputing
- Trends to Watch in HPC
- October 18 – 10^18 – Exascale Day, a reason for optimism
- Why Arm’s SVE will bring Vector Computing from HPC to the Edge
- Is the Cloud Ready for HPC Workloads?
- Arm Ecosystem Paving the Way for AI-enabled Supercomputing
- From Cloud to Edge, Arm Artisan Physical IP Tuned for a Trillion Intelligent Devices
- Read more HPC blogs