基于2D Cache结构的H.264运动补偿访存带宽优化方法

H.264/AVC的运动补偿处理环节需要消耗大量的内存访问带宽,这成为制约其性能的关键因素.分析表明,如此巨大的带宽消耗具体来自5个方面:像素数据的重复读取、地址对齐、突发访问、SDRAM页切换和内存竞争冲突.提出一种基于2D Cache结构的运动补偿带宽优化方法,充分利用像素的重用以减少数据的重复读取.同时通过结合数据在SDRAM中映射方式的优化,将众多短而随机的访问整合为地址对齐的突发访问,并减少了访问过程中页切换的次数.此外还提出了访存的组突发访问模式,以解决SDRAM竞争冲突所引入的开销.实验结果表明采用上述优化设计后,运动补偿的访存带宽降低了82.9～87.6%,同现存优化效率较高的方法相比,带宽进一步减少了64%～87%.在达到相同带宽减少幅度的前提下,所提出的新方法比传统Cache结构电路面积减少91%.该方法目前已在一款多媒体SoC芯片设计中实际应用.     

基于存储队列的Cache访问性能优化研究

高性能处理器普遍采用片上集成大容量复杂结构的一级Cache提高处理器性能,但随着Cache容量和复杂度的增加,访问Cache所产生的访存延迟和功耗明显增加;基于存储队列,提出了一种通过减少Cache访问次数来降低功耗和延迟的方法,利用存储队列来缓存Load/Store指令的数据,并且当存储队列不满时,通过空闲入口暂存已经完成的仿存数据,提高了连续访存数据的复用率,减少了Cache的访问次数;仿真结果显示,该方法在增加少量的控制逻辑基础上,显著减少了Cache的访问次数,降低了Cache的功耗,减少了访存延迟,加快了执行速度。     

基于可变步长的访存延迟测量模型的研究与实现

评测访存延迟对于优化应用访存模式和数据放置有重要的指导意义,然而数据Cache、多线程、数据预取等技术却严重干扰了访存延迟测量的精度。设计并实现了基于可变步长的访存延迟测量模型,在一块空间内根据用户指定的步长创建访问序列环,循环访问这个序列得出平均时间,即为访存延迟。最后对Intel的通用处理器和飞腾处理器在不同数据大小、步长、线程数等情况下的访存延迟进行了测量比较,该模型能够显示存储层次并精确显示测量延迟。     

基于可变步长的访存延迟测量模型的研究与实现

评测访存延迟对于优化应用访存模式和数据放置有重要的指导意义,然而数据Cache、多线程、数据预取等技术却严重干扰了访存延迟测量的精度。设计并实现了基于可变步长的访存延迟测量模型,在一块空间内根据用户指定的步长创建访问序列环,循环访问这个序列得出平均时间,即为访存延迟。最后对Intel的通用处理器和飞腾处理器在不同数据大小、步长、线程数等情况下的访存延迟进行了测量比较,该模型能够显示存储层次并精确显示测量延迟。     

影响机器人多处理机控制系统性能原因及其对策

多处理机系统的优越性,已越来越受人们的青睐.然而,由于其存在Cache一致问题,所以严重制约多处理机系统在机器人中的应用.如何提高多处理机系统的性能是本文研究的主要问题.我们通过仿真实验提出了:在编译时设置用基于访存时标、互斥锁机制、Java监控程序这些方法,使有腿的步行机器人,在其下位机的多处理机系统中Cache中数据保持一致,从而充分发挥了Cache的优点,缩短了访存时间,减少互连网络、存储器的带宽压力和冲突,其结果是有效地提高了机器人多处理机计算机系统的性能.     

多线程向量处理器中向量数据存储结构的设计与实现

多线程和向量技术相结合是当前微处理器设计的一个重要趋势.提出一种多线程向量处理器中向量数据存储结构,利用多线程切换来隐藏访存延迟,并让向量数据直接访问二级cache来提高带宽.模拟实验表明在所提出的存储结构下,访存带宽随线程数线性增长,向量数据访问带宽明显高于标量数据访问带宽.     

基于简单常见模式编码（S-FPC）的压缩Cache层次设计

本文基于简单常见模式压缩编码设计了一种新颖的片内压缩Cache层次结构。在该结构中,L1数据Cache和L2Cache都以压缩格式保存数据,但具有不同的布局。其中,L1数据Cache的布局能触发部分Cache行预取,同时又能避免普通预取技术可能导致的Cache污染增加以及带宽浪费的现象,而且没有预取缓冲开销。实验结果表明,与传统Cache结构相比,本文的设计方案可以显著增加L1数据Cache和L2Cache的有效容量,并且不会增加L1数据Cache的访存延迟,对L1数据Cache平均能增加33％的有效容量,减少L1数据Cachhe失效率达21％,程序执行速度提高了13％。     

移动图形处理器的纹理Cache设计

为了提高移动图形处理器中统一架构染色器的效率,减少其与片外存储器间的访问次数,提出了一种4端口纹理高速缓存结构。该结构采用基于Mipamp算法的纹理映射和基于细化层次(Level of Detail,LOD)选择不同单端口Cache的存储方式,提高了纹理Cache的命中率。此外为了提高数据吞吐率,采用4端口并行读取纹素。设计了FIFO缓冲区预取数据,降低访存延迟。利用SV搭建实验平台对纹理图像进行测试,结果表明纹理Cache的平均命中率为92.5%,数据吞吐率接近单端口Cache的4倍。     

结合访存失效队列状态的预取策略

随着存储系统的访问速度与处理器的运算速度的差距越来越显著,访存性能已成为提高计算机系统性能的瓶颈.通过对指令Cache和数据Cache失效行为的分析,提出一种预取策略--结合访存失效队列状态的预取策略.该预取策略保持了指令和数据访问的次序,有利于预取流的提取.并将指令流和数据流的预取相分离,避免相互替换.在预取发起时机的选择上,不但考虑当前总线是否空闲,而且结合访存失效队列的状态,减小对处理器正常访存请求的影响.通过流过滤机制提高预取准确性,降低预取对访存带宽的需求.结果表明,采用结合访存失效队列状态的预取策略,处理器的平均访存延时减少30%,SPEC CPU2000程序的IPC值平均提高8.3%.     

分块内存的数据分布优化

为了提高访存效率,提供可以与计算流水线并行执行的多个独立的访存流水线,魂芯DSP片上存储器设计时采用分块内存结构,并在核内提供多个独立的地址生成单元用于访存操作.针对分块内存的结构特点,编译器对程序中的存储访问构建关于变量的冲突图,对分块内存进行存储块分配,优化数据在分块内存的分布.以数据在分块内存的优化分布为基础,指导程序中访存操作在地址生成单元的优化分配,使得编译器生成的代码可以最大程度地挖掘程序中数据访问的并行性.实验表明,基于分块内存的数据分配分布优化为其它优化如地址寄存器的分簇、访存向量化、软件流水等经典优化提供了良好基础,保证了编译器生成的代码可以充分发挥魂芯DSP提供的指令级并行能力.     

Communication-aware task assignment algorithm for MPSoC using shared memory

In a Multi-Processor System-on-a-Chip (MPSoC) based on Network-on-Chip (NoC), which processes massive data in a distributed fashion, communication is concentrated on shared memory. This paper proposes an assignment algorithm that can minimize the total power consumption for data communication in executing application programs and a switch structure that can reduce communication congestion resulting from simultaneous accesses to the shared memory. The proposed assignment algorithm gives higher priority to the tasks transferring a larger amount of data to shared memory, so that these tasks can be assigned to the PEs close to shared memory. The proposed switch structure was designed to support multi-port memory, which is often used for shared memory. The ports of the proposed switch are dedicated to be connected with in/out ports of shared memory in order to increase communication bandwidth between PEs and shared memories. By adopting the proposed scheme, the congestion caused by the concentrated requests to the memory can be reduced. Experimental results show that power consumption for transferring data in High-Definition (HD) H.264 decoder, Motion-JPEG decoder, MP3 decoder and 2D Wavelet transform codes has been reduced by 23.9% on the average, when compared with the cases of applying the well-known FC, BN and SA algorithms. The area has been slightly increased by 1.7% compared to conventional NoC structures.     

FPGA architecture and implementation of sparse matrix-vector multiplication for the finite element method

The Finite Element Method (FEM) is a computationally intensive scientific and engineering analysis tool that has diverse applications ranging from structural engineering to electromagnetic simulation. The trends in floating-point performance are moving in favor of Field-Programmable Gate Arrays (FPGAs), hence increasing interest has grown in the scientific community to exploit this technology. We present an architecture and implementation of an FPGA-based sparse matrix-vector multiplier (SMVM) for use in the iterative solution of large, sparse systems of equations arising from FEM applications. FEM matrices display specific sparsity patterns that can be exploited to improve the efficiency of hardware designs. Our architecture exploits FEM matrix sparsity structure to achieve a balance between performance and hardware resource requirements by relying on external SDRAM for data storage while utilizing the FPGAs computational resources in a stream-through systolic approach. The architecture is based on a pipelined linear array of processing elements (PEs) coupled with a hardware-oriented matrix striping algorithm and a partitioning scheme which enables it to process arbitrarily big matrices without changing the number of PEs in the architecture. Therefore, this architecture is only limited by the amount of external RAM available to the FPGA. The implemented SMVM-pipeline prototype contains 8 PEs and is clocked at 110 MHz obtaining a peak performance of 1.76 GFLOPS. For 8 GB/s of memory bandwidth typical of recent FPGA systems, this architecture can achieve 1.5 GFLOPS sustained performance. Using multiple instances of the pipeline, linear scaling of the peak and sustained performance can be achieved. Our stream-through architecture provides the added advantage of enabling an iterative implementation of the SMVM computation required by iterative solution techniques such as the conjugate gradient method, avoiding initialization time due to data loading and setup inside the FPGA internal memory.     

基于FPGA的细粒度并行K-means算法加速器的设计与实现

本文在深入分析K-means算法计算特征的基础上,基于FPGA平台提出并实现了一种细粒度的并行浮点K-means算法。设计采用了阵列多PE并行处理的任务划分策略,实现了处理单元间的负载平衡,采用数据驱动的流水线隐藏片外存储访问,设计了一种基于脉动阵列结构的主从多PE并行计算阵列,并在单片FPGA(XC5VLX330)上成功集成了4个PE。实验结果表明,我们提出的K-means算法加速器结构具备良好的可扩展性。通过实验测试,我们的实现方案相对于Pentium 4 2.66 GHz单处理器程序达到了15倍的加速比。     

Priority queues and sorting methods for parallel simulation

The authors examine the design, implementation, and experimental analysis of parallel priority queues for device and network simulation. They consider: 1) distributed splay trees using MPI; 2) concurrent heaps using shared memory atomic locks; and 3) a new, more general concurrent data structure based on distributed sorted lists, designed to provide dynamically balanced work allocation and efficient use of shared memory resources. We evaluate performance for all three data structures on a Cray-TSESOO system at KFA-Julich. Our comparisons are based on simulations of single buffers and a 64×64 packet switch which supports multicasting. In all implementations, PEs monitor traffic at their preassigned input/output ports, while priority queue elements are distributed across the Cray-TBE virtual shared memory. Our experiments with up to 60000 packets and two to 64 PEs indicate that concurrent priority queues perform much better than distributed ones. Both concurrent implementations have comparable performance, while our new data structure uses less memory and has been further optimized. We also consider parallel simulation for symmetric networks by sorting integer conflict functions and implementing a packet indexing scheme. The optimized message passing network simulator can process ~500 K packet moves in one second, with an efficiency that exceeds ~50 percent for a few thousand packets on the Cray-T3E with 32 PEs. All developed data structures form a parallel library. Although our concurrent implementations use the Cray-TSE ShMem library, portability can be derived from Open-MP or MP1-2 standard libraries, which will provide support for one-way communication and shared memory lock mechanisms     

FPGA Based High Performance Double-Precision Matrix Multiplication

We present two designs (I and II) for IEEE 754 double precision floating point matrix multiplication, optimized for implementation on high-end FPGAs. It forms the kernel in many important tile-based BLAS algorithms, making an excellent candidate for acceleration. The designs, both based on the rank-1 update scheme, can handle arbitrary matrix sizes, and are able to sustain their peak performance except during an initial latency period. Through these designs, the trade-offs involved in terms of local-memory and bandwidth for an FPGA implementation are demonstrated and an analysis is presented for the optimal choice of design parameters. The designs, implemented on a Virtex-5 SX240T FPGA, scale gracefully from 1 to 40 processing elements(PEs) with a less than 1% degradation in the design frequency of 373 MHz. With 40 PEs and a design speed of 373 MHz, a sustained performance of 29.8 GFLOPS is possible with a bandwidth requirement of 750 MB/s for design-II and 5.9 GB/s for design-I. This compares favourably with both related art and general purpose CPU implementations.     

一种支持优化分块策略的矩阵乘加速器设计

在许多应用领域中,大规模浮点矩阵乘法往往是最耗时的计算核心之一。在新兴的应用中经常存在至少有一个维度很小的大规模矩阵,我们把具备这种特性的矩阵称为非均匀矩阵。由于FPGA上用以存储中间结果的片上存储器容量十分有限,计算大规模矩阵乘法时往往需要将矩阵划分成细粒度的子块计算任务。当加速非均匀矩阵乘法时,由于只支持固定分块大小,大多数现有的线性阵列结构的硬件矩阵乘法器将遭受很大的性能下降。为了解决这个问题,提出了一种有效的优化分块策略。在此基础上,在Xilinx公司的Zynq XC7Z045FPGA芯片上实现了一个支持可变分块的矩阵乘法器。通过集成224个处理单元,该矩阵乘法器在150 MHz的时钟频率下对于实际应用中的非均匀矩乘达到了48GFLOPS的实测性能,而所需带宽仅为4.8GB/s。实验结果表明,我们提出的分块策略相比于传统的分块算法实现了高达12%的性能提升。     

支持多种传输模式的双通路串行RapidIO设计与实现

传统的串行RapidIO2.1接口支持3种通道模式(1×、2×、4×)。在2×或1×模式下,4条物理链路只有2条或1条在进行数据传输,其余链路被闲置,造成带宽浪费;另外,一个RapidIO接口只能与一个目的端互连。基于传统的串行RapidIO2.1接口协议,设计了一种支持双通路传输的串行RapidIO接口,通过PCS层的可配置交叉开关共实现14种传输模式,双通路模式下可同时和两个串行RapidIO接口互连。双通路RapidIO提高了RapidIO系统互连的灵活性和传输带宽。实验结果表明,在1×或2×模式下,双通路传输的传输带宽是传统设计的两倍;4×模式下,双通路RapidIO的有效带宽与传统单通路RapidIO的相同。     

异构多核图形处理器存储系统设计与实现

提出了异构多核图形处理器(HMGPU)存储管理系统的硬件实现方法,采用固定分区与分页式分区两种方式分别对大片连续数据与小片非连续数据进行管理,使用Verilog语言进行硬件设计和仿真,并在FPGA开发板上进行了验证。实验结果表明,该系统为HMGPU提供了2 021.2 MB/s的有效存储带宽。     

一种多通道并行固态存储系统的设计与实现

鉴于高速数据采集系统对实时数据存储带宽和容量的要求,提出一种基于现场可编程门阵列（FPGA）的高速多通道并行固态存储系统。该系统以现场可编程门阵列器件XCV5LX110T为核心,选用大容量高速闪存芯片作为存储介质,通过采用并行总线拓宽技术和流水线缓冲技术,在FPGA片内搭建高速多通道并行存储硬件架构,从硬件角度提高系统的数据吞吐带宽。设计一种基于超级页的地址映射策略,并使用该策略对闪存转换层算法的请求处理机制进行并行加速优化,从软件角度提高系统的存储并行性。测试结果表明,该系统的最大存储速度达到73MB／s,其性能指标能满足高速实时数据存储的需求,证明多通道存储架构和FTL算法具有良好的并行性和可扩展性。     

基于FPGA实现的高速串行交换模块实现方法研究

采用Xlinx公司的Virtex5系列FPGA设计了一个用于多种高速串行协议的数据交换模块,并解决了该模块实现中的关键问题。该交换模块实现4X模式RapidIO协议与4X模式PCI Express协议之间的数据交换,以及自定义光纤协议与4X模式PCIExpress协议之间的数据交换,实现了单字读写以及DMA操作,并提供高速稳定的传输带宽。