利用提示调优融合多种信息的低资源事件抽取方法

针对现有流水线式事件抽取方法依靠大量训练数据、在低资源情况下难以快速迁移运用等问题,利用提示调优技术,提出适用于低资源场景下的流水线式事件抽取方法（low-resource event extraction method using the multi-information fusion with prompt tuning,IFPT）。该方法通过构造语义映射和提示模板充分利用事件类型描述、实体类型等多种信息,能够高效使用有限训练数据,流水线式地完成事件检测和论元抽取。实验结果表明,在低资源情况下,IFPT方法论元抽取性能超过了所有基准模型,采取流水线方式能够达到与SOTA模型相媲美的性能。     

数字电视CAS中DES加密模块的FPGA实现

一种基于 FPGA 的数据加密标准算法的实现。就资源优先和性能优先分别使用循环法和流水线法对 DES 加密算法进行了设计,并对其进行了比较。通过采用子密钥简单产生和 ROM 优化 S盒的方法,对流水线法进行改进,达到了资源占用率低、加密速度快的效果。     

资源约束的FPGA流水线调度

循环是程序中十分耗时的部分,流水线能够加速循环执行但需要大量运算资源。由于FPGA资源有限,将循环代码在FPGA上加速时手动设计流水线不具有实际可行性。该文使用软件流水将循环自动映射到FPGA上,并实现资源约束下的流水线调度。通过探索整个或者局部资源组合空间,可以选择一个性能和面积比较平衡的设计。     

以访存为中心的阵列众核处理器核心流水线设计

传统的流水线设计是以转移指令为中心的,大量逻辑资源被用于提高处理器转移预测的能力,以保证向流水线发射和执行部件提供充足的指令流。在阵列众核处理器中提出了一种以访存为中心的核心流水线设计。通过提高访存装载指令在流水线中的执行优先级,以及访存装载指令的预测执行机制,可以有效减少顺序流水线因访存延迟所带来的停顿,提高流水线性能和能效比。测试结果表明,以4KB容量的装载指令访存地址表为例,访存为中心的流水线设计可以带来8.6%的流水线性能提升和7%的流水线能效比提高。     

GCC的流水冲突识别器和并行调度器

由于超长指令字处理器通常都有多级流水线和复杂的资源使用限制,如何准确地描述处理器的流水线模型,快速地判断是否存在资源冲突并不是个简单地任务。文章介绍GCC新引入的正则表达式语法的流水线描述机制。在将GCC移植到笔者所开发的SuperV芯片的过程中,利用该机制对SuperV芯片的流水线结构和资源使用限制进行详尽地描述,启动了GCC的指令级并行调度。通过并行调度,测试程序的性能提高了大约6%—35%。     

一种基于ILP的流水线调度算法

提出一种基于整数规划 (IL P)的流水线调度算法 ,可以在给定的流水时延槽之下优化资源需求 ,保证流水线长度的最小化 ;同时 ,在各个工作步中的操作与调度的先后顺序无关 ,整体性能优良 ,支持链式 (Chaining)操作 ,较适合于流水线数据通路的综合。     

AVS解码器流水线控制机制的一种改进设计

根据AVS视频标准中的解码算法特点,提出一种改进的AVS解码器流水线控制机制.该流水线对解码模块采用两级控制策略,不同级别流水线中的解码模块数据处理粒度不同,节省了级间缓存.同时,合理安排数据处理顺序,减少了数据等待时间.仿真结果表明,该设计在不影响系统解码性能的基础上节省了大量的存储器资源.     

多线程处理器资源分配策略

处理器资源如何在多个线程之间进行分配和共享是直接影响多线程处理器性能的关键问题。该文总结4种分配模型,提出其实现机制,讨论资源分配平衡问题,指出可根据目标应用和流水线不同阶段的特点,在各流水线阶段综合采用不同分配模型和实现机制,实现处理器资源的合理分配。     

基于异步映射的异步流水线设计

论文提出了"异步映射"的概念,即在传统同步流水线基础上,试图将控制部分和数据计算部分分离,将控制部分用优化的异步电路替换,得到兼具异步和同步电路优点的流水线结构。论文首先证明了基于锁存器的流水线在同步时钟和异步流水线控制下能够获得相近的性能和完全相同的结果,然后以"异步映射"的方法实现了一款基于锁存器的16位异步DSP的设计。通过对DSP的性能进行分析,"异步映射"的可行性得到了验证。"异步映射"也可以被应用于基于寄存器的流水线和复杂流水线结构,达到利用同步设计的丰富资源发挥异步设计长处的目的。     

基于SPSO的机械流水线负载任务分配方法

在弱通信条件下,传统的机械流水线任务分配方法采用任务的随机分配,无法根据流水线的实际负载能力,将应有的任务量分配到相应的流水线上。提出一种基于简化粒子群优化算法(Simplified particle swarm optimization algorithm,SPSO)的高效机械流水线任务分配方法,首先对每个流水线的实际负载能力进行动态评估。然后采用粒子群优化算法对所有流水线负载分配相应的计算任务。由于每个负载的任务量是根据实际的流水线性能来分配的,所以可以使得全局的效率最优化。最后通过实验对算法的性能进行验证。结果显示,改进方法在基于粒子群优化的机械流水线任务分配下,任务根据流水线性能,呈现很好的聚类,算法收敛性好,分配任务速度快,具有很好的应用价值。     

Blackfin DSP嵌入式图像数据采集机制设计

针对Blackfin Dsp媒体信号处理器硬件资源,着重分析了SDRAM上数据总线的冲突问题,并且提出了解决方案,在软件架构上提出了利用Blackfin DSP自带的已经封装好了的API功能函数,实现了对CMOS图像传感器以及DSP内部资源的配置,试验表明采集程序稳定性高,DSP功耗更低,更加节省SDRAM的传输带宽,能够满足实时图像采集和编码的要求。     

基于帕累托最优的云资源调度研究

针对在云环境中,服务资源在各用户间难以实现最优动态分配的问题,利用帕累托最优理论与粒子群优化算法相互结合应用于云计算模型中,对各种服务资源的效用进行最优化配置,最终使资源利用率达到一个最优的状态。通过CloudSim对云服务资源调度进行仿真实验,结果表明,采用帕累托最优算法优化后的云计算模型具有更好的系统性能,使得资源的调度和配置达到最优。     

Server Selection,Configuration and Reconfiguration Technology for IaaS Cloud with Multiple Server Types

We propose a server selection, configuration, reconfiguration and automatic performance verification technology to meet user functional and performance requirements on various types of cloud compute servers. Various servers mean there are not only virtual machines on normal CPU servers but also container or baremetal servers on strong graphic processing unit (GPU) servers or field programmable gate arrays (FPGAs) with a configuration that accelerates specified computation. Early cloud systems are composed of many PC-like servers, and virtual machines on these severs use distributed processing technology to achieve high computational performance. However, recent cloud systems change to make the best use of advances in hardware power. It is well known that baremetal and container performances are better than virtual machines performances. And dedicated processing servers, such as strong GPU servers for graphics processing, and FPGA servers for specified computation, have increased. Our objective for this study was to enable cloud providers to provision compute resources on appropriate hardware based on user requirements, so that users can benefit from high performance of their applications easily. Our proposed technology select appropriate servers for user compute resources from various types of hardware, such as GPUs and FPGAs, or set appropriate configurations or reconfigurations of FPGAs to use hardware power. Furthermore, our technology automatically verifies the performances of provisioned systems. We measured provisioning and automatic performance verification times to show the effectiveness of our technology.     

一种FPGA配置文件压缩算法

基于现场可编程门阵列(FPGA)的可重构系统具有高性能和高灵活性,但随着FPGA规模的不断扩大,配置文件规模相应增加,导致可重构计算时间过长。该文提出一种FPGA配置文件压缩算法VLZW,降低了对片外存储器的容量要求,通过减少每次重构传送的配置数据缩短了系统重构时间。     

基于双层规划的容器云资源动态配置算法

本文分析容器云资源动态配置决策问题,通过定义容器云资源的调度任务,求解得到容器云资源调度时间;利用容器云资源调度任务的最短时间矩阵,获取容器云资源调度所需的条件。在双层规划条件下,求解容器云资源调度的目标函数和约束函数;考虑到用户的任务情况和云数据中心的云资源状况,在虚拟机上构建一个到物理主机的矩阵,通过构建容器云资源动态配置结果在优化时的目标函数,结合约束条件,实现容器云资源的动态配置。实验结果表明,资源动态配置算法不仅可以提高容器云资源的利用率,还可以减少配置完成时间,具有更好的动态配置性能。     

基于多标签的内核配置图及其应用

Linux内核提供了灵活的内核配置项机制,便于针对不同的应用场景进行个性化定制.但内核配置项的数量巨大且增长快速,配置项的默认值在不同内核版本中经常改变,即使专业的内核团队设置配置项也面临很多挑战.针对上述问题,提出基于多标签的内核配置图,该图包含内核配置项间的依赖关系、功能标签、性能标签、安全标签和配置项使能率.此外...     

Supporting configurability and real time in RTD channels

Network communication services that can be configured to customize functionality provide significant advantages over monolithic versions, but can be difficult to construct if the services must also provide real‐time guarantees on message delivery. This paper describes how practical issues related to the combination of configurability and real time have been addressed in real‐time dependable (RTD) channels, a communication abstraction that has been prototyped using the Cactus system. The architecture and implementation of RTD channels are described, including facilities for translating an application's quality of service requirements into configuration‐specific resource requirements, and an admission control architecture that uses system‐wide information to determine whether sufficient resources exist to create new channels. In addition, the results of experiments that demonstrate that the desired channel properties—including real‐time guarantees—are provided for various configurations are presented. Finally, practical problems and lessons learned from implementing RTD channels on the general‐purpose MK 7.3 Mach operating system are discussed. Copyright © 2001 John Wiley & Sons, Ltd.     

Design space exploration of on-chip ring interconnection for a CPU–GPU heterogeneous architecture

Incorporating a GPU architecture into CMP, which is more efficient with certain types of applications, is a popular architecture trend in recent processors. This heterogeneous mix of architectures will use an on-chip interconnection to access shared resources such as last-level cache tiles and memory controllers. The configuration of this on-chip network will likely have a significant impact on resource distribution, fairness, and overall performance.     

Coordinating components in middleware systems

Configuration and coordination are central issues in the design and implementation of middleware systems and are one of the reasons why building such systems is more complex than constructing stand‐alone sequential programs. Through configuration, the structure of the system is established—which elements it contains, where they are located and how they are interconnected. Coordination is concerned with the interaction of the various components—when an interaction takes place, which parties are involved, what protocols are followed. Its purpose is to coordinate the behaviour of the various components to meet the overall system specification. The open and adaptive nature of middleware systems makes the task of configuration and coordination particularly challenging. We propose a model that can operate in such an environment and enables the dynamic integration and coordination of components by observing and coercing their behaviour through the interception of the messages exchanged between them. Copyright © 2003 John Wiley & Sons, Ltd.     

An adaptive resource management scheme in cloud computing

There are various significant issues in resource allocation, such as maximum computing performance and green computing, which have attracted researchers’ attention recently. Therefore, how to accomplish tasks with the lowest cost has become an important issue, especially considering the rate at which the resources on the Earth are being used. The goal of this research is to design a sub-optimal resource allocation system in a cloud computing environment. A prediction mechanism is realized by using support vector regressions (SVRs) to estimate the number of resource utilization according to the SLA of each process, and the resources are redistributed based on the current status of all virtual machines installed in physical machines. Notably, a resource dispatch mechanism using genetic algorithms (GAs) is proposed in this study to determine the reallocation of resources. The experimental results show that the proposed scheme achieves an effective configuration via reaching an agreement between the utilization of resources within physical machines monitored by a physical machine monitor and service level agreements (SLA) between virtual machines operators and a cloud services provider. In addition, our proposed mechanism can fully utilize hardware resources and maintain desirable performance in the cloud environment.