Held in conjunction with HPDC (The International ACM Symposium on High-Performance Parallel and Distributed Computing) 2019, Phoenix, AZ, June 22-29
The focus of this workshop is to discover and discuss open challenges of the computing infrastructure convergence problem, and to draw preliminary solutions for tackling such problems. The workshop aims to act as a liaison, a common forum for experts from computing domains, such as HPC, big data, AI to discuss the infrastructure-related problems that could be tackled through a concerted approach. The workshop combines three major domains of high interest to HPDC under one umbrella that emphasizes convergence: high performance computing, big data analytics, large scale machine learning.
The case for deep convergence needs pragmatic arguments. Addressing the lack of performance studies for the deep convergence problem, in this workshop we will provide a dedicated forum for preliminary performance studies of processing representative big-data and HPC workloads on a common infrastructure. In our vision, the deep convergence entails defining common software-and-hardware ecosystems, catering to the needs of all computing domains. To achieve deep convergence, we advocate for a complete redesign of the software-hardware stacks through software-hardware concertation. This is achieved by identifying common software building blocks for all computing domains, mapping them to their most suited hardware, and offering efficient mechanisms for combining these in efficient, large-scale workflows. Such workflows would have the capability to efficiently describe combinations of HPC, AI, and BD workloads, that efficiently share resources on a deeply converged infrastructure.
Topics of Interest
We invite the submission of original work that is related to the topics below. The submitted articles can either be short (4 pages) position papers, or long (8 pages) research papers. We especially encourage submissions that look into the convergence of the three focus domains - HPC, BD, and AI. Topics of interest include, but are not limited to:
- Resource-management and scheduling for converged infrastructure
- Interoperability, interference mitigation, performance variability
- Novel system architectures and programming models for converged infrastructure
- Performance studies of converged infrastructure, performance portability and compatibility
- Converged storage systems, malleable and reconfigurable runtimes and middleware
- Hardware-software co-design and concertation
- Identifying common building blocks, libraries, and functionality for HPC, BD, and AI/ML/DL
- Convergence with the cloud, redefining elasticity and scalability for converged infrastructure
- New classes of applications, applications at the boundary of HPC, Big Data, AI, Machine Learning, or Deep Learning, or combinations of, that fit best on converged infrastructure
Proceedings of this workshop will be published in CD format and will be available at the conference (together with the HPDC conference proceedings).
- Paper submission: April 22, 2019
- Paper Acceptance Notification: April 29, 2019
- Camera-ready deadline: May 6, 2019
- Workshop: June 24, 2019