Flash controllers play critical roles in determining the performance of flash storage. But current controller architectures based on asynchronous interface cannot meet the ever-increasing demands of performance for ne...
详细信息
ISBN:
(纸本)9781450333580
Flash controllers play critical roles in determining the performance of flash storage. But current controller architectures based on asynchronous interface cannot meet the ever-increasing demands of performance for new large-scale flash storage systems, due to their limitations of I/O bandwidth and scalability. In this paper, we propose an advanced NAND flash controller architecture, called Dysource, to improve performance and scalability of flash storage. Based on the high-speed source synchronous interface supported by ONFI standard, we propose a dynamic scheduling strategy to extend exploration of parallelism to target level. Moreover, we design a novel instruction "MOVE" and incorporate it into existing instruction set to speed up garbage collection for higher performance. The experimental results suggest that the performance of the Dysource controller is 4.6 to 9.4 times better than the traditional asynchronous controller as for organization of 8 channels x 8 targets. Moreover, by employing architecture of multiple Dysources, the storage system can scale to arbitrary number of channels and 32 targets at most per channel with high growth rate of performance. Copyright 2015 ACM.
Knowledge representation learning (KRL) is one of the important research topics in artificial intelligence and Natural language processing. It can efficiently calculate the semantics of entities and relations in a low...
详细信息
With the rapid development and application of cloud computing, there exist plenty of clouds that are distributed on the open Internet, decentralized in the management, evolving with various services providing diverse ...
详细信息
Currently, the performance problems of software systems gets more and more attentions. Among various diagnosis methods based on system traces, principal component analysis (PCA) based methods are widely used due to th...
详细信息
Functional programming languages have a long history and receive more and more attention today. The paper focuses on the development of functional languages and aims to introduce the concepts, such as higher-order fun...
详细信息
According to Moore's law the complexity of VLSI circuits has doubled approximately every two years, resulting in simulation becoming the major bottleneck in the circuit design process. parallel and distributed sim...
详细信息
Interconnection network plays an essential role in the architecture of large-scale high performance computing (HPC) systems. In the paper, we construct a novel family of lowdiameter topologies, Galaxyfly, using techni...
详细信息
Decision trees are famous machine learning classifiers which have been widely used in many areas, such as healthcare, text classification and remote diagnostics, etc. The service providers usually host a decision tree...
详细信息
Cooperation of CPU and hardware accelerator on SoC FPGA to accomplish computational intensive tasks, provides significant advantages in performance and energy efficiency. However, current operating systems provide lit...
详细信息
The key escrow problem and high computational cost are the two major problems that hinder the wider adoption of hierarchical identity-based signature (HIBS) scheme. HIBS schemes with either escrow-free (EF) or online/...
详细信息
暂无评论