We propose a novel hardware and software co-exploration framework for efficient neural architecture search (NAS). Different from existing hardware-aware NAS which assumes a fixed hardware design and explores the neural architecture search space only, our framework simultaneously explores both the architecture search space and the hardware design space to identify the best neural architecture and hardware pairs that maximize both test accuracy and hardware efficiency. Such a practice greatly opens up the design freedom and pushes forward the Pareto frontier between hardware efficiency and test accuracy for better design tradeoffs. The framework iteratively performs a two-level (fast and slow) exploration. Without lengthy training, the fast exploration can effectively fine-tune hyperparameters and prune inferior architectures in terms of hardware specifications, which significantly accelerates the NAS process. Then, the slow exploration trains candidates on a validation set and updates a controller using the reinforcement learning to maximize the expected accuracy together with the hardware efficiency. In this paper, we demonstrate that the co-exploration framework can effectively expand the search space to incorporate models with high accuracy, and we theoretically show that the proposed two-level optimization can efficiently prune inferior solutions to better explore the search space. Experimental results on ImageNet show that the co-exploration NAS can find solutions with the same accuracy, 35.24% higher throughput, 54.05% higher energy efficiency, compared with the hardware-aware NAS.

Solid state drives (SSDs) have been widely deployed in personal computers, data centers, and cloud storages. In order to improve performance, SSDs are usually constructed with a number of channels with each channel connecting to a number of nand flash chips, each flash chip consisting of multiple dies and each die containing multiple planes. Based on this parallel architecture, I/O requests are potentially able to access parallel units simultaneously. Despite the rich parallelism offered by the parallel architecture, recent studies show that the utilization of flash parallel units is seriously low. This paper shows that the low parallel unit utilization is highly caused by the access conflict among I/O requests. In this paper, we propose parallel issue queueing (PIQ), a novel I/O scheduler at the host systems. PIQ groups I/O requests without conflicts into the same batch and I/O requests with conflicts into different batches. Hence, the multiple I/O requests in one batch can be fulfilled simultaneously by exploiting the rich parallelism of SSDs. Extensive experimental results show that PIQ delivers significant performance improvement especially for the applications which have heavy access conflicts.

Field programmable gate array (FPGA) has been one of the most popular platforms to implement convolutional neural networks (CNNs) due to its high performance and cost efficiency; however, limited by the on-chip resources, the existing single-FPGA architectures cannot fully exploit the parallelism in CNNs. In this paper, we explore heterogeneous FPGA-based designs to effectively leverage both task and data parallelism, such that the resultant system can achieve the minimum cost while satisfying timing constraints. In order to maximize the task parallelism, we investigate two critical problems: 1) buffer placement, where to place buffers to partition CNNs into pipeline stages and 2) task assignment, what type of FPGA to implement different CNN layers. We first formulate the system-level optimization problem with a mixed integer linear programming model. Then, we propose an efficient dynamic programming algorithm to obtain the optimal solutions. On top of that, we devise an efficient algorithm that exploits data parallelism within CNN layers to further improve cost efficiency. Evaluations on well-known CNNs demonstrate that the proposed techniques can obtain an average of 30.82% reduction in system cost under the same timing constraint, and an average of 1.5 times speedup in performance under the same cost budget, compared with the state-of-the-art techniques.