Search

Taking advantage of green energy is becoming increasingly popular. However, when it comes to the storage I/O, where storage devices usually schedule nodes based on the intensity of the storage workload, the irregular fluctuations of green energy lead to the mismatch of peaks and valleys between green energy and storage workloads, which increases the difficulty of scheduling storage resources. To address this issue, a tiering-cooperative energy-efficient scheme for hybrid NVM-SSD storage system, called MixSave, is proposed. MixSave customizes the energy-saving strategies for each layer and tunes them cooperatively between layers, reducing overall energy usage. It efficiently tiers NVM and SSD and replicates data among them based on their performance and energy characteristics, achieving both high performance and large capacity. The workload-driven strategy is used by the NVM layer, which is used for caching, to ensure the best performance, and the number of active NVM devices depends on the storage workload. Meanwhile, the SSD layer employs a green energy-oriented strategy to conserve traditional energy, and the number of active SSD devices will be proportionally to the amount of green energy supplied. Evaluation results show that, when compared with energy-save-unaware approaches, MixSave degrades performance by less than 4% and saves traditional energy by 73% to 80% and 55% to 61% under light and heavy workloads, respectively.

Deep‐learning methods have been developed in recent years and have achieved dramatic improvements for image denoising. The existing deep‐learning methods can be conducted using two major models: Encoder–decoder and high‐resolution, where the high‐resolution model has superior resolution ability for detail description and restoration. In this study, a high‐resolution‐based network called multiscale residual fusion network (MRF‐Net) is proposed, which employed the spatial and contextual information of images. In detail, dilated convolution layers are used to enlarge the network's receptive field and learned sufficient features in a multiscale feature extracting module. The function of dilated convolution is reinterpreted here and it is viewed as a complex downsampling operation. Therefore, multiscale feature analysis could be performed in the proposed network by dilated convolution. Multilevel feature maps are sequentially obtained through a residual projection module, where considerable contextual and spatial information was collected from the multiscale features. In a residual fusion module, all maps were aggregated to generate a residual image effectively for noise removal. Experiments demonstrated that the MRF‐Net outperformed several state‐of‐the‐art model‐based and deep‐learning methods in both blind and non‐blind image denoising tests. Meanwhile, ablation studies were executed to verify the denoising performance of each module. Moreover, this method exhibited high computational efficiency, thus demonstrating its practicability. [ABSTRACT FROM AUTHOR]