Search papers, labs, and topics across Lattice.
This paper addresses the challenges of optimizing network-on-chip (NoC) placement and packet routing in FPGA designs, which have become increasingly complex. By integrating a NoC link congestion cost into the versatile place & route (VPR) engine and employing turn model routing algorithms, the authors achieve a remarkable 90.7% reduction in NoC congestion while only slightly increasing bandwidth demand. Further improvements are realized through a boolean satisfiability (SAT) formulation for routing, culminating in a total congestion reduction of 95.1% compared to baseline methods, alongside an 8.8% reduction in wirelength via a NoC-aware reinforcement learning agent.
Reducing NoC congestion by over 95% in FPGA designs could revolutionize how we approach chip integration and performance optimization.
To help scale to ever-larger and more complex designs, recent FPGA architectures now integrate network-on-chips (NoCs). NoCs help transfer high-bandwidth data over long distances within the chip without using scarce low-delay long routing wire segments. While NoC-enhanced FPGAs aid system integration and design reuse, they also complicate FPGA computer-aided design (CAD) flows by introducing new constraints and metrics. Placement and routing needs to optimize NoC metrics like latency and bandwidth utilization and avoid link oversubscription (congestion), while simultaneously optimizing the programmable routing resource usage of the design modules attached to NoC routers. In this work we develop several new approaches to reduce NoC congestion while minimizing the impact on other design metrics. First, we incorporate a NoC link congestion cost into the placement engine of the open-source CAD flow, versatile place & route (VPR). Second, we integrate turn model NoC routing algorithms into the placement engine to leverage path diversity to further reduce congestion. On average over a suite of 29 benchmarks combining placement congestion modeling with turn model packet routing reduces NoC congestion by 90.7% at the cost of increasing aggregate bandwidth demand by 4%. In cases where the enhanced placement engine and NoC routing fail to fully resolve congestion, we formulate NoC routing as a boolean satisfiability (SAT) problem. This approach yields significant additional improvements; the combined algorithm reduces congestion by 95.1% compared to the baseline placement. Finally, we enhance the reinforcement learning (RL) agent in VPR鈥檚 placement engine by introducing a NoC-aware move type, resulting in an 8.8% reduction in wirelength on designs that make extensive use of the NoC.
