Torrent: A Distributed DMA for Efficient and Flexible Point-to-Multipoint Data Movement

2025-12-19

AI Summary

The mismatch between computational capacity and on-chip communication bandwidth in SoCs severely limits performance for data-parallel workloads such as AI, while existing NoCs lack native multicast support—hardware enhancements or protocol modifications compromise compatibility and scalability. This paper proposes Torrent, a distributed DMA architecture featuring the novel Chainwrite logical chaining mechanism, enabling stateless, semantically faithful point-to-multipoint transmission without modifying NoC hardware or protocols. Coupled with a topology-aware dual-scheduling algorithm, Torrent jointly optimizes throughput and energy efficiency. RTL simulation and FPGA prototyping, along with 16nm ASIC synthesis and measurement, demonstrate up to 7.88× speedup over unicast baselines; per-destination chain overhead is only 82 cycles, area overhead is 207 μm² per destination, and ASIC measurements show that the Torrent infrastructure occupies merely 1.2% of total die area and 2.3% of total power.

Technology Category

• Search and Optimization: Distributed Search
• Data Mining & Knowledge Management: Scalability, Parallel & Distributed Systems
• Machine Learning: Hardware-aware ML

Application Category

• Systems and Infrastructure for Web, Mobile and WoT: Data management and stream processing for Web, mobile and wireless applications
• Economics, Online Markets and Human Computation: Incentives in network design for Web infrastructures and ecosystems
• Security and Privacy: Large-scale security measurements

Abstract

The growing disparity between computational power and on-chip communication bandwidth is a critical bottleneck in modern Systems-on-Chip (SoCs), especially for data-parallel workloads like AI. Efficient point-to-multipoint (P2MP) data movement, such as multicast, is essential for high performance. However, native multicast support is lacking in standard interconnect protocols. Existing P2MP solutions, such as multicast-capable Network-on-Chip (NoC), impose additional overhead to the network hardware and require modifications to the interconnect protocol, compromising scalability and compatibility. This paper introduces Torrent, a novel distributed DMA architecture that enables efficient P2MP data transfers without modifying NoC hardware and interconnect protocol. Torrent conducts P2MP data transfers by forming logical chains over the NoC, where the data traverses through targeted destinations resembling a linked list. This Chainwrite mechanism preserves the P2P nature of every data transfer while enabling flexible data transfers to an unlimited number of destinations. To optimize the performance and energy consumption of Chainwrite, two scheduling algorithms are developed to determine the optimal chain order based on NoC topology. Our RTL and FPGA prototype evaluations using both synthetic and real workloads demonstrate significant advantages in performance, flexibility, and scalability over network-layer multicast. Compared to the unicast baseline, Torrent achieves up to a 7.88x speedup. ASIC synthesis on 16nm technology confirms the architecture's minimal footprint in area (1.2%) and power (2.3%). Thanks to the Chainwrite, Torrent delivers scalable P2MP data transfers with a small cycle overhead of 82CC and area overhead of 207um2 per destination.

Problem

• Enables efficient point-to-multipoint data movement without hardware changes
• Optimizes performance and energy via topology-aware scheduling algorithms
• Provides scalable multicast-like transfers with minimal area and power overhead

Innovation

• Distributed DMA architecture for P2MP transfers
• Chainwrite mechanism using logical linked-list chains
• Scheduling algorithms optimize chain order for NoC topology