Experiences with ML-Driven Design: A NoC Case Study

There has been a lot of recent interest in applying machine learning (ML) to the design of systems, which purports to aid human experts in extracting new insights leading to better systems. In this work, we share our experiences with applying ML to improve one aspect of networks-on-chips (NoC) to uncover new ideas and approaches, which eventually led us to a new arbitration scheme that is effective for NoCs under heavy contention. However, a significant amount of human effort and creativity was still needed to optimize just one aspect (arbitration) of what is only one component (the NoC) of the overall processor. This leads us to conclude that much work (and opportunity!) remains to be done in the area of ML-driven architecture design.

• Publication year

  2020

• Venue

  International Symposium on High-Performance Computer Architecture

• Publication date

  2020-02-01

• Fields of study

  Computer Science, Engineering

• Identifiers
  DOI 10.1109/HPCA47549.2020.00058
• External record

  Open on Semantic Scholar

• Source metadata

  Semantic Scholar

• No claims are published for this paper.

• No concepts are published for this paper.

• Optimizing Routerless Network-on-Chip Designs: An Innovative Learning-Based Framework

  2019cited by this paper
• Reinforcement learning

  2019cited by this paper
• Bit-level Perceptron Prediction for Indirect Branches

  2019cited by this paper
• Perceptron-Based Prefetch Filtering

  2019cited by this paper
• An Energy-Efficient Network-on-Chip Design using Reinforcement Learning

  2019cited by this paper
• An orchestrated NoC prioritization mechanism for heterogeneous CPU-GPU systems

  2019cited by this paper
• Dynamic Voltage and Frequency Scaling in NoCs with Supervised and Reinforcement Learning Techniques

  2019cited by this paper
• A Survey of Machine Learning Applied to Computer Architecture Design

  2019cited by this paper
• Toward More Efficient NoC Arbitration : A Deep Reinforcement Learning Approach

  2018cited by this paper
• Extending the Power-Efficiency and Performance of Photonic Interconnects for Heterogeneous Multicores with Machine Learning

  2018cited by this paper
• LEAD: Learning-enabled Energy-Aware Dynamic Voltage/frequency scaling in NoCs

  2018cited by this paper
• Lost in Abstraction: Pitfalls of Analyzing GPUs at the Intermediate Language Level

  2018cited by this paper
• Learning Memory Access Patterns

  2018cited by this paper
• Long short term memory based hardware prefetcher: a case study

  2017cited by this paper
• Machine learning enabled power-aware Network-on-Chip design

  2017cited by this paper
• Interpretable Convolutional Neural Networks

  2017cited by this paper
• There and back again: Optimizing the interconnect in networks of memory cubes

  2017cited by this paper
• Dynamic error mitigation in NoCs using intelligent prediction techniques

  2016cited by this paper
• Efficient synthetic traffic models for large, complex SoCs

  2016cited by this paper
• Perceptron learning for reuse prediction

  2016cited by this paper
• Semantic locality and context-based prefetching using reinforcement learning

  2015cited by this paper
• Human-level control through deep reinforcement learning

  2015influential reference
• On the characterization of OpenCL dwarfs on fixed and reconfigurable platforms

  2014cited by this paper
• SynFull: Synthetic traffic models capturing cache coherent behaviour

  2014cited by this paper
• A Fast Circuit Topology for Finding the Maximum of N k-bit Numbers

  2013cited by this paper
• HARAQ: Congestion-Aware Learning Model for Highly Adaptive Routing Algorithm in On-Chip Networks

  2012cited by this paper
• The gem5 simulator

  2011cited by this paper
• A 48-Core IA-32 Processor in 45 nm CMOS Using On-Die Message-Passing and DVFS for Performance and Power Scaling

  2011cited by this paper
• Packet chaining: Efficient single-cycle allocation for on-chip networks

  2011cited by this paper
• Probabilistic Distance-Based Arbitration: Providing Equality of Service for Many-Core CMPs

  2010cited by this paper
• Aérgia: exploiting packet latency slack in on-chip networks

  2010cited by this paper
• GARNET: A detailed on-chip network model inside a full-system simulator

  2009cited by this paper
• On-Chip Networks

  2009cited by this paper
• Rodinia: A benchmark suite for heterogeneous computing

  2009cited by this paper
• Self-Optimizing Memory Controllers: A Reinforcement Learning Approach

  2008cited by this paper
• Age-based packet arbitration in large-radix k-ary n-cubes

  2007cited by this paper
• A case for (partially) TAgged GEometric history length branch prediction

  2006cited by this paper
• Dynamic branch prediction with perceptrons

  2001cited by this paper
• Route packets, not wires: on-chip interconnection networks

  2001cited by this paper
• A fast arbitration scheme for terabit packet switches

  1999cited by this paper
• The iSLIP scheduling algorithm for input-queued switches

  1999influential reference
• Q-learning

  1992cited by this paper
• Self-improving reactive agents based on reinforcement learning, planning and teaching

  1992cited by this paper
• Virtual clock: a new traffic control algorithm for packet switching networks

  1990cited by this paper
• Analysis and simulation of a fair queueing algorithm

  1989cited by this paper

• Athena: Synergizing Data Prefetching and Off-Chip Prediction via Online Reinforcement Learning

  2026cites this paper
• DRCD: a regional-contention-driven arbitration policy for CPU–GPU heterogeneous systems

  2025influential citation
• Traffic Patterns in Networks-on-Chip: A Survey

  2025cites this paper
• Intelligent learning-based routing algorithm for optical network-on-chips

  2025cites this paper
• DRLAR: A deep reinforcement learning-based adaptive routing framework for network-on-chips

  2024cites this paper
• A comprehensive study and holistic review of empowering network-on-chip application mapping through machine learning techniques

  2024cites this paper
• Communication Synchronization-Aware Arbitration Policy in NoC-Based DNN Accelerators

  2024cites this paper
• An Efficient Deep Reinforcement Learning-Based Automatic Cache Replacement Policy in Cloud Block Storage Systems

  2024cites this paper
• A Top-Down Modeling Approach for Networks-on-Chip Components Design: A Switch as Case Study

  2023cites this paper
• LARE: A Linear Approximate Reinforcement Learning Based Adaptive Routing for Network-on-Chips

  2023cites this paper
• DPBC-VCP: A Network-On-Chip Prioritization Mechanism Combined with VCP for CPU-GPU Heterogeneous Systems

  2023cites this paper
• SecureNoC: A Learning-Enabled, High-Performance, Energy-Efficient, and Secure On-Chip Communication Framework Design

  2022cites this paper
• An End-to-End Automatic Cache Replacement Policy Using Deep Reinforcement Learning

  2022cites this paper
• Chapter Seven - Power-efficient network-on-chip design by partial topology reconfiguration

  2022cites this paper
• Pythia: A Customizable Hardware Prefetching Framework Using Online Reinforcement Learning

  2021cites this paper
• Distilling Arbitration Logic from Traces using Machine Learning: A Case Study on NoC

  2021influential citation
• Continual Learning Approach for Improving the Data and Computation Mapping in Near-Memory Processing System

  2021cites this paper
• Designing a Cost-Effective Cache Replacement Policy using Machine Learning

  2021cites this paper
• Q-adaptive: A Multi-Agent Reinforcement Learning Based Routing on Dragonfly Network

  2021cites this paper
• Adapt-NoC: A Flexible Network-on-Chip Design for Heterogeneous Manycore Architectures

  2021cites this paper
• A Survey of Machine Learning for Computer Architecture and Systems

  2021cites this paper
• The Tribes of Machine Learning and the Realm of Computer Architecture

  2020cites this paper
• This paper is included in the Proceedings of the 22nd USENIX Conference on File and Storage Technologies.

  year unknowncites this paper