Table of Contents
Fetching ...

Wireless Interconnection Network (WINE) for Post-Exascale High-Performance Computing

Hong Ki Kim, Yong Hun Jang, Hee Soo Kim, Won Young Kang, Young-Chai Ko, Sang Hyun Lee

TL;DR

Addressing interconnect bottlenecks in post-exascale HPC, the paper proposes WINE, a versatile wireless interconnect framework combining s-WINE and r-WINE with wired backbones. Using the WINE-cellar digital twin, indoor THz and FSO wireless links are analyzed, showing hundreds of Gbps per wireless path and coexistence with $100$ Gbps wired lanes; real-system scale is illustrated by extrapolations toward TOP500-class performance (e.g., exceeding $1.206$ EFLOPS). WINE demonstrates reduced inter-rack hop counts and tail latency on representative workloads, with observed benchmark uplifts of roughly 4–11% and projections toward ~1.122 PFlops on TOP500-scale systems. These results indicate that topology-adaptive, workload-aware wireless interconnects can materially advance post-exascale HPC, while outlining practical research directions for protocols, interference management, and digital-twin administration.

Abstract

Interconnection networks, or `interconnects,' play a crucial role in administering the communication among computing units of high-performance computing (HPC) systems. Efficient provisioning of interconnects minimizes the processing delay wherein computing units await information sharing between each other, thereby enhancing the overall computation efficiency. Ideally, interconnects are designed with topologies tailored to match specific workflows, requiring diverse structures for different applications. However, since modifying their structures mid-operation renders impractical, indirect communication incurs across distant units. In managing numerous long-routed data deliveries, heavy burdens on the network side may lead to the under-utilization of computing resources. In view of state-of-the-art HPC paradigms that solicit dense interconnections for diverse computation-hungry applications, this article presents a versatile wireless interconnecting framework, coined as Wireless Interconnection NEtwork (WINE). The framework exploits cutting-edge wireless technologies that promote workload adaptability and scalability of modern interconnects. Design and implementation of wirelessly reliable links are strategized under network-oriented scrutiny of HPC architectures. A virtual HPC platform is developed to assess WINE's feasibilities, verifying its practicality for integration into modern HPC infrastructures.

Wireless Interconnection Network (WINE) for Post-Exascale High-Performance Computing

TL;DR

Addressing interconnect bottlenecks in post-exascale HPC, the paper proposes WINE, a versatile wireless interconnect framework combining s-WINE and r-WINE with wired backbones. Using the WINE-cellar digital twin, indoor THz and FSO wireless links are analyzed, showing hundreds of Gbps per wireless path and coexistence with Gbps wired lanes; real-system scale is illustrated by extrapolations toward TOP500-class performance (e.g., exceeding EFLOPS). WINE demonstrates reduced inter-rack hop counts and tail latency on representative workloads, with observed benchmark uplifts of roughly 4–11% and projections toward ~1.122 PFlops on TOP500-scale systems. These results indicate that topology-adaptive, workload-aware wireless interconnects can materially advance post-exascale HPC, while outlining practical research directions for protocols, interference management, and digital-twin administration.

Abstract

Interconnection networks, or `interconnects,' play a crucial role in administering the communication among computing units of high-performance computing (HPC) systems. Efficient provisioning of interconnects minimizes the processing delay wherein computing units await information sharing between each other, thereby enhancing the overall computation efficiency. Ideally, interconnects are designed with topologies tailored to match specific workflows, requiring diverse structures for different applications. However, since modifying their structures mid-operation renders impractical, indirect communication incurs across distant units. In managing numerous long-routed data deliveries, heavy burdens on the network side may lead to the under-utilization of computing resources. In view of state-of-the-art HPC paradigms that solicit dense interconnections for diverse computation-hungry applications, this article presents a versatile wireless interconnecting framework, coined as Wireless Interconnection NEtwork (WINE). The framework exploits cutting-edge wireless technologies that promote workload adaptability and scalability of modern interconnects. Design and implementation of wirelessly reliable links are strategized under network-oriented scrutiny of HPC architectures. A virtual HPC platform is developed to assess WINE's feasibilities, verifying its practicality for integration into modern HPC infrastructures.
Paper Structure (10 sections, 5 figures, 1 table)

This paper contains 10 sections, 5 figures, 1 table.

Figures (5)

  • Figure 1: Modern HPC architecture, administration, and scaling.
  • Figure 2: HPC interconnection principles.
  • Figure 3: WINE-integrated HPC platform.
  • Figure 4: Digital twin HPC environment and WINE utilities.
  • Figure 5: WINE link coordination and performances.