4Network-on-Chips (NoCs) are the de facto choice for designing the interconnect fabric in multicore chips due to their regularity, efficiency, simplicity, and scalability. However, NoC suffers from excessive static power and dynamic energy due to transistor leakage current and data movement between the cores and caches. Power consumption issues are only exacerbated by ever decreasing technology sizes. Dynamic Voltage and Frequency Scaling (DVFS) is one technique that seeks to reduce dynamic energy; however this often occurs at the expense of performance. In this paper, we propose LEAD Learning-enabled Energy-Aware Dynamic voltage/frequency scaling for multicore architectures using both supervised learning and reinforcement learning approaches. LEAD groups the router and its outgoing links into the same V/F domain and implements proactive DVFS mode management strategies that rely on offline trained machine learning models in order to provide optimal V/F mode selection between different voltage/frequency pairs. We present three supervised learning versions of LEAD that are based on buffer utilization, change in buffer utilization and change in energy/throughput, which allow proactive mode selection based on accurate prediction of future network parameters. We then describe a reinforcement learning approach to LEAD that optimizes the DVFS mode selection directly, obviating the need for label and threshold engineering. Simulation results using PARSEC and Splash-2 benchmarks on a $4 \times 4$ concentrated mesh architecture show that by using supervised learning LEAD can achieve an average dynamic energy savings of 15.4% for a loss in throughput of 0.8% with no significant impact on latency. When reinforcement learning is used, LEAD increases average dynamic energy savings to 20.3% at the cost of a 1.5% decrease in throughput and a 1.7% increase in latency. Overall, the more flexible reinforcement learning approach enables learning an optimal behavior for a wider range of load environments under any desired energy vs. throughput tradeoff.
Designing and implementing wireless networks on chips (WiNoCs) presents numerous engineering challenges, in the areas of computer architecture, multiple access, wireless propagation, physical layer communications processing, and device design and fabrication. In this article we provide a survey on WiNoC propagation and the issues involved with WiNoC channel modeling. We address both attenuation and dispersion, and illustrate the dramatic differences between the miniature WiNoC case and more familiar terrestrial wireless channels. Remaining channel modeling research is outlined.
2 October 2017 in IEEE Transactions on Antennas and Propagation
We propose a novel antenna design enabled by 3-D printing technology for future wireless intrachip interconnects aiming at applications of multicore architectures and system-on-chips. In our proposed design we use vertical quarter-wavelength monopoles at 160 GHz on a ground plane to avoid low antenna radiation efficiency caused by the silicon substrate. The monopoles are surrounded by a specially designed dielectric property distribution. This additional degree of freedom in design enabled by 3-D printing technology is used to tailor the electromagnetic wave propagation. As a result, the desired wireless link gain is enhanced and the undesired spatial crosstalk is reduced. Simulation results show that the proposed dielectric loading approach improves the desired link gain by 8–15 dB and reduces the crosstalk by 9–23 dB from 155 to 165 GHz. As a proof-of-concept, a 60 GHz prototype is designed, fabricated, and characterized. Our measurement results match the simulation results and demonstrate 10–18 dB improvement of the desired link gain and 10–30 dB reduction in the crosstalk from 55 to 61 GHz. The demonstrated transmission loss of the desired link at a distance of 17 mm is only 15 dB, which is over 10 dB better than the previously reported work.
6 May 2022 in IEEE Transactions on Circuits and Systems I: Regular Papers
The complexity and size of recent deep neural network (DNN) models have increased significantly in pursuit of high inference accuracy. Chiplet-based accelerator is considered a viable scaling approach to provide substantial computation capability and on-chip memory for efficient process of such DNN models. However, communication using metallic interconnects in prior chiplet-based accelerators poses a major challenge to system performance, energy efficiency, and scalability. Photonic interconnects can adequately support communication across chiplets due to features such as distance-independent latency, high bandwidth density, and high energy efficiency. Furthermore, the salient ease of broadcast property makes photonic interconnects suitable for DNN inference which often incurs prevalent broadcast communication. In this paper, we propose a scalable chiplet-based DNN accelerator with photonic interconnects named ASCEND. ASCEND introduces (1) a novel photonic network that supports seamless intra- and inter- chiplet broadcast communication, and flexible mapping of diverse convolution layers, and (2) a tailored dataflow that exploits the ease of broadcast property and maximizes parallelism by simultaneously processing computations with shared input data. Simulation results using multiple DNN models show that ASCEND achieves 71% and 67% reduction in execution time and energy consumption, respectively, as compared to other state-of-the-art chiplet-based DNN accelerators with metallic or photonic interconnects.
28 December 2021 in IEEE Transactions on Parallel and Distributed Systems
Chiplet-based convolution neural network (CNN) accelerators have emerged as a promising solution to provide substantial processing power and on-chip memory capacity for CNN inference. The performance of these accelerators is often limited by inter-chiplet metallic interconnects. Emerging technologies such as photonic interconnects can overcome the limitations of metallic interconnects due to several superior properties including high bandwidth density and distance-independent latency. However, implementing photonic interconnects in chiplet-based CNN accelerators is challenging and requires combined effort of network architectural optimization and CNN dataflow customization. In this article, we propose SPRINT, a chiplet-based CNN accelerator that consists of a global buffer and several accelerator chiplets. SPRINT introduces two novel designs: (1) a photonic inter-chiplet network that can adapt to specific communication patterns in CNN inference through wavelength allocation and waveguide reconfiguration, and (2) a CNN dataflow that can leverage the broadcasting capability of photonic interconnects while minimizing the costly electrical-to-optical and optical-to-electrical signal conversions. Simulations using multiple CNN models show that SPRINT achieves up to 76% and 68% reduction in execution time and energy consumption, respectively, as compared to other state-of-the-art chiplet-based architectures with either metallic or photonic interconnects.
6 February 2017 in IEEE Transactions on Computer-Aided Design of Integrated Circuits and Systems
As aggressive scaling continues to push Multi- Processor System-on-Chips (MPSoCs) to new limits, complex hardware structures combined with stringent area and power constraints will continue to diminish reliability.Waning reliability in integrated circuits will increase the susceptibility of transient and permanent faults. There is an urgent demand for adaptive Error Correction Coding (ECC) schemes in Network-on-Chips (NoCs) to provide fault tolerance and improve overall resiliency of MPSoC architectures. The goal of adaptive ECC schemes should be to maximize power savings when faults are infrequent and increase application speedup by boosting fault coverage when faults are frequent. In this paper, we propose Runtime Adaptive Scrubbing (RAS), a novel multi-layered error correction and detection scheme with three modes of operation enabled by an area-efficient configurable encoder for encoding packets on the switch-to-switch (s2s) layer, thus preventing faults from accumulating up the network stack and onto the end-to-end (e2e) layer. As fault rates fluctuate we propose a dynamic methodology for improving fault localization and intelligently adapt fault coverage on demand to sustain graceful network degradation. RAS successfully improves network resiliency, fault localization, and fault coverage as compared to traditional static switch-toswitch (s2s) schemes. Simulation results demonstrate that static RAS improves network speedup by 10% for Splash-2/PARSEC benchmarks on a 8 8 mesh network while reducing area overhead by 14% and incurring on an average 6.6% power penalty by boosting fault tolerance when fault rates increase. Further, our dynamic RAS scheme maintains 97.88% of network performance for real applications while incurring 20% power penalty.