神威E级原型机互连网络和消息机制

本文描述了神威E级原型机的互连网络和消息机制.神威E级原型机是继神威蓝光、神威·太湖之光之后神威家族的第三代计算机.该计算机作为一台E级计算机的原型机,峰值性能3.13 PFlops,其最大的特色之一就是采用28 Gbps传输技术,设计开发了新一代的神威高阶路由器和神威高性能网络接口两款芯片,在传统胖树的基础上,设计了双轨泛树拓扑结构,定义实现了新颖的神威消息原语和消息库,实现了一种基于包级粒度动态切换的双轨乱序消息机制,通信性能比神威·太湖之光互连网络提升了4倍,为神威E级计算机互连网络的研制奠定了基础.     

基于资源调度的集群节能系统的设计与实现

集群计算机系统的运算性能跨入百万亿次、千万亿次时代,节能降耗已成为集群计算机系统必须面对的重要问题之一,本文从系统级节能的角度出发,结合神威高性能集群计算机系统的系统监测、作业管理、IPMI带外电源管理和TuxOnIce系统休眠技术,设计并实现了基于资源调度的集群节能系统,通过对空闲结点的关机或休眠,能够有效地降低集群系统空闲时的能耗,使神威高性能集群计算机系统成为真正的绿色计算机。     

闪迪ExtremeII至尊极速240GB固态硬盘 闪迪的新旗舰

闪迪Extreme是其消费级SSD中的旗舰产品,不过在今年CES上闪迪推出UItraPlusSSD后,Extreme这个旗舰级就基本没有优势了,所以不久前闪迪又推出新的旗舰产品——ExtremeII至尊极速SSD。新的旗舰ExtremeII至尊极速SSDSnUItraPlus在主控的采用上有明显的区别。UltraPlus采用的是Maryell9175“梵高节能型”主控,主要用于提升电脑的性能,主打高性能低功耗,而ExtremeII至尊极速SSD采用的是Marvell9187“莫纳2．0”高性能主控,它针对的是技术发烧友、     

基于磁盘和固态硬盘的混合存储系统研究综述

大数据和云计算环境下海量增长的数据对存储系统的超高容量和体系结构带来了极大的挑战。目前存储系统的发展趋向于大容量、低成本和高性能,然而任何单一的存储器件如传统的机械磁盘（HDD）、固态硬盘（SSD）、非易失型性随机存储器等由于其固有的物理特性的限制,并不能满足以上的需求。将不同的存储介质混合组合成高效的存储系统是一个好的解决方法,固态硬盘作为一种高可靠性、低能耗、高性能的存储器被越来越广泛地运用到混合存储系统。通过将固态硬盘与传统磁盘进行组合,利用固态硬盘的高性能和传统磁盘低成本大容量的特点,能够为用户提供大容量的存储空间,保证系统的高性能,同时还能降低成本。通过阐述SSD与HDD混合存储系统的研究现状,对不同的SSD与HDD混合存储系统进行分类总结;然后针对缓存架构和设备同层架构这两种目前最流行的存储架构中涉及到的关键技术和不足进行讨论;最后对基于SSD和HDD的混合存储技术进行概括总结,并对今后该领域的研究重点和方向进行展望。     

厂商简讯

东芝宣布推出Blade X-gale系列薄型、刀片式高性能SSD（固态硬盘）产品。该新型SSD产品共有64GB、128GB和256GB三种容量规格,适用于空间有限的产品。     

基于神威蓝光处理器的向量数学软件包

首先介绍了SIMD扩展技术,并分析了使用SIMD扩展的3种方式,认为通过调用特定目标平台优化的第三方库是应用领域软件开发者快速开发高效并行程序的较好的方式;其次,介绍了国产神威处理器SW-1600平台,并利用SIMD扩展和循环展开等技术开发了SW-VML(SW Vector Math Library),开发过程中提出了访存对界、简化向量条件分支的优化方法,解决了非对界访存、向量与标量数组转换影响性能的问题,并根据SW编译器对OpenMP的支持,开发了多线程OpenMp版;最后,在SW-1600平台上采用不同向量规模对SW-VML进行了测试,测试结果显示,SIMD向量化相对于串行程序加速比为2.08,4线程相对单线程平均加速比为2.26.SW-VML是在国产神威系列处理器上开发高效程序的向量函数软件包,也是在神威蓝光高性能计算平台单计算节点开发高性能程序的基础软件工具包.     

“神威”的神威

12月12日，中国气象局副局长郑国光在神威IB网络产品发布会上透露，中国气象科学研究院气候预报研究的高性能计算系统已成功应用神威InfiniBand网络产品。中国国家气象局的大型计算机运行系统正在变得更加强大和完善，堪当现今复杂多变气候条件下的研究分析和预测的重任。     

面向Lustre文件系统的固态盘存储加速技术研究

磁盘的随机I/O延时制约了存储系统的性能提高,具备高性能随机I/O特性的固态盘(solid state disk,SSD)逐渐成为关注的热点.分析了磁盘、Flash型SSD以及DRAM型SSD三类设备不同的性能特点,讨论了SSD存储加速技术的研究现状,提出了一种面向Lustre文件系统的固态盘存储加速系统架构,介绍和分析了各模块的构成与原理,提出了对象迁移策略.     

希捷SSD助力Nor-Tech

希捷科技公司日前宣布,其Seagate Pulsar固态硬盘（SSD）应用于Nor-Tech屡获奖项的先进计算系统;Nor-Tech是高性能计算（HPC）领域的领先企业。受益于希捷Pulsar SSD的速度和可靠性,Nor-Tech得以制造出了高度移动化的便携式集群（Portable Clusters）,这些产品使高性能计算解决方案能够用于各种各样的移动应用和苛刻的户外环境。     

HDD与SSD华山论剑:各有千秋

装载着HDD的华为赛门铁克N8500和装载着SSD的EMCVNX5700,在NFS和CIFS两种基准构造的完全不同的测试场景中均有优异的性能表现,说明HDD与SSD均有足够的实力胜任不同NAS应用中的高性能需求.     

Correlation and relaxation labelling: An experimental investigation on fast algorithms

This paper compares experimental results between three popular matching functions: the cross-correlation coefficient (CCC); the sum of squared difference (SSD); and sum of the absolute value of difference (SAVD), within our newly developed correlation-relaxation (C-R) framework (Wu 1995). The C-R framework is a general method for determining optical flow and has been applied to determining cloud motion from satellite images. SSD and SAVD are simpler and faster functions to calculate, when compared with CCC, and their uses can lead to significant savings in computer time in the initial selection of displacement candidates. Given that the image distortion is Gaussian noise, and the motion is translational, the study shows that while computationally more expensive, the performance of the CCC function is better, or at least no worse, than using SSD and SAVD in the selection of initial displacement candidates. Similarly, the performance of SSD is better, or no worse, than using SAVD. Computationally, SSD is the fastest among the three functions. In the presence of high level distortion, however, the poor quality of initial candidates selected using SSD and SAVD usually means a large number of iterations of the subsequent relaxation labelling process. In contrast, the CCC function gives high quality initial candidates, and only a small number of iterations are needed. The CCC function also usually leads to better final quality in motion estimations than that produced using the SSD or the SAVD function in the C-R algorithm. In the presence of moderate and low level distortion, however, the performance of SSD can be adequate, and its use can lead to faster processing without much sacrifice to the overall motion estimation quality.     

Buffer flush and address mapping scheme for flash memory solid-state disk

The flash memory solid-state disk (SSD) is emerging as a killer application for NAND flash memory due to its high performance and low power consumption. To attain high write performance, recent SSDs use an internal SDRAM write buffer and parallel architecture that uses interleaving techniques. In such architecture, coarse-grained address mapping called superblock mapping is inevitably used to exploit the parallel architecture. However, superblock mapping shows poor performance for random write requests. In this paper, we propose a novel victim block selection policy for the write buffer considering the parallel architecture of SSD. We also propose a multi-level address mapping scheme that supports small-sized write requests while utilizing the parallel architecture. Experimental results show that the proposed scheme improves the I/O performance of SSD by up to 64% compared to the existing technique.     

ASA-FTL: An adaptive separation aware flash translation layer for solid state drives

The flash-memory based Solid State Drive (SSD) presents a promising storage solution for increasingly critical data-intensive applications due to its low latency (high throughput), high bandwidth, and low power consumption. Within an SSD, its Flash Translation Layer (FTL) is responsible for exposing the SSD’s flash memory storage to the computer system as a simple block device. The FTL design is one of the dominant factors determining an SSD’s lifespan and performance. To reduce the garbage collection overhead and deliver better performance, we propose a new, low-cost, adaptive separation-aware flash translation layer (ASA-FTL) that combines sampling, data clustering and selective caching of recency information to accurately identify and separate hot/cold data while incurring minimal overhead. We use sampling for light-weight identification of separation criteria, and our dedicated selective caching mechanism is designed to save the limited RAM resource in contemporary SSDs. Using simulations of ASA-FTL with both real-world and synthetic workloads, we have shown that our proposed approach reduces the garbage collection overhead by up to 28% and the overall response time by 15% compared to one of the most advanced existing FTLs. We find that the data clustering using a small sample size provides significant performance benefit while only incurring a very small computation and memory cost. In addition, our evaluation shows that ASA-FTL is able to adapt to the changes in the access pattern of workloads, which is a major advantage comparing to existing fixed data separation methods.     

A hybrid filesystem for hard disk drives in tandem with flash memory

The traditional hard disk drive (HDD) is often a bottleneck in the overall performance of modern computer systems. With the development of solid state drives (SSD) based on flash memory, new possibilities are available to improve secondary storage performance. In this work, we propose a new hybrid SSD–HDD storage system and a selection of algorithms designed to assign pages across an HDD and an SSD to optimise I/O performance. The hybrid system combines the advantages of the SSD’s fast random seek speed with the sequential access speed and large storage capacity of the HDD to produce significantly improved performance in a variety of situations. We further improve performance by allowing concurrent access across the two types of storage devices. We show the drive assignment problem is NP-complete and accordingly propose effective heuristic solutions. Extensive experiments using both synthetic and real data sets show our system with a small SSD can outperform a striped dual HDD and remain competitive with a dual SSD.     

一种闪存敏感的多级缓存管理方法

基于闪存的固态硬盘(solid state driver,简称SSD)已经广泛应用于各种移动设备、PC机和服务器.与磁盘相比,尽管SSD具有数据存取速度高、抗震、低功耗等优良特性,但SSD自身也存在读写不对称、价格昂贵等不利因素,这使得SSD 短期内不会完全取代磁盘.将SSD和磁盘组合构建混合系统,可以发挥不同的硬件特性,提升系统性能.基于 MLC 型 SSD 和 SLC 型 SSD 之间的特性差异,提出了一种闪存敏感的多级缓存管理策略——FAMC.FAMC将SSD用在内存和磁盘之间作扩展缓存,针对数据库系统、文件管理中数据访问的特点,有选择地将内存牺牲页缓存到不同类型的SSD.FAMC同时考虑写请求模式和负载类型对系统性能的影响,设计实现对SSD友好的数据管理策略.此外,FAMC基于不同的数据置换代价提出了适用于SSD的缓冲区管理算法.基于多级缓存存储系统对FAMC的性能进行了评测,实验结果表明,FAMC可以大幅度降低系统响应时间,减少磁盘I/O.     

改进SSD算法的多目标检测

目标检测作为计算机视觉的核心,在人脸识别、人脸跟踪、大规模场景识别等方面具有广泛应用,其中One-stage领域的SSD算法检测速度和检测性能较为突出,但在环境较为复杂的多目标检测情况下仍会出现误检和漏检。针对这一问题,提出一种改进SSD算法的多目标检测方法,通过优化SSD内部网络和提高样本适用性的方式改善检测性能;其中,采用修改网络输出和添加抗旋转层ARConv来统一网络结构,降低模型训练时间,减少漏检;并提出P-NMS算法和限制函数优化训练样本,减少误检;在测试阶段,提出单张图片批量测试方法,有效提高模型召回率。实验结果表明,改进后算法具有更强的鲁棒性,并且能有效降低误检、漏检率提升网络性能。     

大数据时代——从冯·诺依曼到计算存储融合

海量数据的出现和硬件计算能力的提升,催生了第三次人工智能的发展热潮,大数据时代来临。首先,分析了拥有冯·诺依曼体系结构的计算机在大数据时代遭遇的存储墙、带宽墙和功耗高问题,引出为适应和满足大数据处理需求的计算机体系结构的发展趋势;接着,分析计算机体系结构层面的计算存储融合技术、软硬件结构、offloading算法的设计思路与技术特点,以及在商业系统中的应用,为高性能计算、数据中心建设和智能SSD产品设计等提供启发意义;分析微观层面基于硅穿孔的3D堆叠封装技术和最新的产业动态;最后,阐述代表计算存储一体化发展目标的类脑计算和最新的研究进展。     

一种基于闪存固态硬盘的辅助缓冲池设计

基于磁盘数据库系统的瓶颈主要在磁盘I／O,通常采用缓冲池的设计,将读到的数据页先放入到内存缓冲池后再进行操作。因此,缓存池的大小直接决定了数据库的性能。通过研究基于闪存固态硬盘的特性,提出了一种基于闪存固态硬盘的辅助缓冲池设计。最后,通过修改开源数据库MySQL InnoDB存储引擎,并通过TPC-C实验对比分析了启用辅助缓冲池后数据库的性能可有100％-320％的提高。     

一种新型的高性能计算机存储系统的研究与实现

文中提出了一种改善计算机存储系统写请求特别是小写请求性能的新的存储结构。采用有着高存储速度、高可靠性的固态盘和廉介的硬盘空间共同和为磁光盘的写高速缓存,并结合顺序文件存取技术,从存储体系结构角度出发,研究并实现了一种以较低成本实现快速、可靠、大容量的存储系统。试验表明新型存储系统不必修改现有文件系统,即能大幅提高存储系统性能。中用于要求高可靠性的军用环境以及高可靠性和快速大容量要求的民用系统。     

Chameleon: A High Performance Flash/FRAM Hybrid Solid State Disk Architecture

Flash memory solid state disk (SSD) is gaining popularity and replacing hard disk drive (HDD) in mobile computing systems such as ultra mobile PCs (UMPCs) and notebook PCs because of lower power consumption, faster random access, and higher shock resistance. One of the key challenges in designing a high-performance flash memory SSD is an efficient handling of small random writes to non-volatile data whose performance suffers from the inherent limitation of flash memory that prohibits in-place update. In this paper, we propose a high performance Flash/FRAM hybrid SSD architecture called Chameleon. In Chameleon, metadata used by the flash translation layer (FTL), a software layer in the flash memory SSD, is maintained in a small FRAM since this metadata is a target of intensive small random writes, whereas the bulk data is kept in the flash memory. Performance evaluation based on an FPGA implementation of the Chameleon architecture shows that the use of FRAM in Chameleon improves the performance by 21.3%. The results also show that even for bulk data that cannot be maintained in FRAM because of the size limitation, the use of fine-grained write buffering is critically important because of the inability of flash memory to perform in-place update.