High fusion computers: The IoTs, edges, data centers, and humans-in-the-loop as a computer

Wanling Gao, Lei Wang, Mingyu Chen, Jin Xiong, Chunjie Luo, Wenli Zhang, Yunyou Huang, Weiping Li, Guoxin Kang, Chen Zheng, Biwei Xie, Shaopeng Dai, Qian He, Hainan Ye, Yungang Bao, Jianfeng ZhanarXiv 2022 · OpenReview

TL;DR

This paper introduces "High Fusion Computers" (HFCs), a new paradigm for systems that integrate IoTs, edge devices, data centers, and human input. HFCs are designed to handle safety-critical, mission-critical tasks with intertwined dependencies, prioritizing accuracy over raw performance. The authors propose an open-source initiative to build these HFC systems, addressing significant challenges in their design and implementation.

What Problem Does This Paper Address?

The paper addresses the growing need for a new computing paradigm to handle emerging and future applications that are fundamentally different from traditional warehouse-scale computers. These new applications, such as autonomous driving, medical emergency management, and smart defense systems, involve direct interaction with the physical world, often with safety-critical or mission-critical implications. They feature complex, intertwined dependencies between IoT devices, edge computing nodes, data centers, and human operators. Existing distributed systems struggle to meet the unique requirements of these applications, which demand high accuracy, real-time responsiveness, and the ability to manage complex interactions and irreversible effects. The authors argue that these systems should be viewed and built as a unified "computer" rather than a collection of distributed components.

Methods

The core of the proposed solution is the funclet methodology and architecture. This approach breaks down complex systems into reusable, well-defined, and independently deployable building blocks called "funclets." These funclets operate across four layers: chiplet (hardware), HWlet (hardware components like memory and network processors), envlet (environment management, including the OS), and servlet (services). This layered architecture, combined with a focus on co-design across vertical (algorithm to hardware) and horizontal (IoT to human) dimensions, aims to manage the complexity of HFC systems. The authors also emphasize the importance of benchmarks and standards for evaluation and development, and propose a scenario simulator to test and verify HFC designs.

Datasets

Not discussed in the paper. The paper focuses on the architectural and methodological aspects of HFCs, rather than specific empirical evaluations on existing datasets.

Key Insights and Ideas

  • Shifting from "Distributed System" to "Computer": The most significant insight is the re-framing of interconnected IoT, edge, and data center systems, including humans, as a single, unified "computer." This paradigm shift is crucial for addressing the deeply intertwined nature of these systems.
  • Prioritizing Accuracy: For safety-critical and mission-critical applications, accuracy in prediction, interpretation, and action is paramount, even before achieving peak performance. This is a departure from traditional performance-centric computing.
  • "High Fusion" as a Unifying Concept: The term "High Fusion Computers" aptly captures the essence of integrating diverse components and bridging the physical and digital worlds.
  • Funclet Abstraction: The funclet approach offers a structured and modular way to build and manage the complexity of HFC systems, promoting reusability and interoperability.
  • The "Ecosystem Wall" Challenge: The authors highlight the difficulty of integrating disparate existing ecosystems (IoT, edge, data center) and the need for a unified approach to overcome this inertia.

Key Results

The paper is primarily a proposal and a call to action, outlining a vision and a plan for High Fusion Computers. It does not present empirical results from a fully realized HFC system. Instead, it:

  • Identifies and characterizes the unique requirements of seven emerging and future applications (e.g., medical emergency management, autonomous driving).
  • Quantifies the resource demands of these applications, suggesting they require orders of magnitude more resources than current systems.
  • Outlines significant challenges in organizability, collaboration, handling irreversible effects, ecosystem integration, and evaluation.
  • Proposes a concrete plan for developing HFCs, including a funclet architecture, benchmarks, tools, and an open-source implementation.

Limitations and Future Work

The paper acknowledges several limitations and points towards future work:

  • Complexity of HFC Systems: The inherent complexity of integrating diverse components and managing intertwined dependencies is a major challenge.
  • Ecosystem Walls: Overcoming the fragmentation of existing IoT, edge, and data center ecosystems is a significant hurdle.
  • Effective Evaluation: Developing appropriate benchmarks and simulators for HFC systems is crucial but challenging due to the complexity and real-world nature of the applications.
  • Open-Source Implementation: The authors plan to provide the first open-source implementation of the funclet architecture, which will be a key step in validating their approach.
  • Iterative Development: The development of HFC systems will be an iterative process, starting with specific application scenarios and essential funclets, gradually expanding scope and refining technologies.
  • Co-design: Future work will involve extensive co-design efforts across algorithms, software, and hardware, as well as horizontal collaboration among all HFC components.

Summary updated

Abstract

Emerging and future applications rely heavily upon systems consisting of Internet of Things (IoT), edges, data centers, and humans-in-the-loop. Significantly different from warehouse-scale computers that serve independent concurrent user requests, this new class of computer systems directly interacts with the physical world, considering humans an essential part and performing safety-critical and mission-critical operations; their computations have intertwined dependencies between not only adjacent execution loops but also actions or decisions triggered by IoTs, edge, datacenters, or humans-in-the-loop; the systems must first satisfy the accuracy metric in predicting, interpreting, or taking action before meeting the performance goal under different cases. This article argues we need a paradigm shift to reconstruct the IoTs, edges, data centers, and humans-in-the-loop as a computer rather than a distributed system. We coin a new term, high fusion computers (HFCs), to describe this class of systems. The fusion in the term has two implications: fusing IoTs, edges, data centers, and humans-in-the-loop as a computer, fusing the physical and digital worlds through HFC systems. HFC is a pivotal case of the open-source computer systems initiative. We laid out the challenges, plan, and call for uniting our community's wisdom and actions to address the HFC challenges. Everything, including the source code, will be publicly available from the project homepage: https://www.computercouncil.org/HFC/.
BibTeX
@misc{gao2022high,
  author = {Wanling Gao and Lei Wang and Mingyu Chen and Jin Xiong and Chunjie Luo and Wenli Zhang and Yunyou Huang and Weiping Li and Guoxin Kang and Chen Zheng and Biwei Xie and Shaopeng Dai and Qian He and Hainan Ye and Yungang Bao and Jianfeng Zhan},
  title = {{High fusion computers: The IoTs, edges, data centers, and humans-in-the-loop as a computer}},
  year = {2022},
  eprint = {2212.00721},
  archivePrefix = {arXiv},
  url = {https://arxiv.org/abs/2212.00721},
}
Figure