Heterogeneous Computing Systems for Complex Scientific Discovery Workflows
Christoph Hagleitner, Dionysios Diamantopoulos, Burkhard Ringlein et al. · Design, Automation, and Test in Europe · 2021
With Moore's law progressively running out of steam, heterogeneous computing architectures have been powering the top supercomputers in the world for many years and are now finding broader adoption across the industry. The trend towards sustainable computing also requires domain-specific heterogeneous hardware architectures, which promise further gains in energy efficiency. At the same time, today's high performance computing applications have evolved from monolithic simulations in a single domain to multidisciplinary complex workflows. In this paper, we explore how these trends affect system design decisions and what this means for future computing system architectures.
Cited by 8 works. doi:10.23919/date51398.2021.9474061
Authors
- Christoph Hagleitner
- Dionysios Diamantopoulos
- Burkhard Ringlein
- Constantinos Evangelinos
- Charles Johns
- Rong Chang
- Bruce D’Amora
- James A. Kahle
- James Sexton
- Michael A. Johnston
- Edward O. Pyzer‐Knapp
References
- Improved protein structure prediction using potentials from deep learning
- HAQ: Hardware-Aware Automated Quantization With Mixed Precision
- CAP twelve years later: How the "rules" have changed
- Anton 2: Raising the Bar for Performance and Programmability in a Special-Purpose Molecular Dynamics Supercomputer
- “Found in Translation”: predicting outcomes of complex organic chemistry reactions using neural sequence-to-sequence models
- Learning from the Harvard Clean Energy Project: The Use of Neural Networks to Accelerate Materials Discovery
- Universal Memcomputing Machines
- Challenges and Opportunities of Near-Term Quantum Computing Systems
- A 3.0 TFLOPS 0.62V Scalable Processor Core for High Compute Utilization AI Training and Inference
- Parallel and Distributed Thompson Sampling for Large-scale Accelerated Exploration of Chemical Space