Design and implementation of dual-mode configurable memory architecture for CNN accelerator

With the rapid development of deep learning algorithms,the computational complexity and functional diversity are increasing rapidly.However,the gap between high computational density and insufficient memory bandwidth under the traditional von Neumann architecture is getting worse.Analyzing the algorithmic characteristics of convolutional neural network(CNN),it is found that the access characteristics of convolution(CONV)and fully connected(FC)operations are very different.Based on this feature,a dual-mode reronfigurable distributed memory architecture for CNN accelerator is designed.It can be configured in Bank mode or first input first output(FIFO)mode to accommodate the access needs of different operations.At the same time,a programmable memory control unit is designed,which can effectively control the dual-mode configurable distributed memory architecture by using customized special accessing instructions and reduce the data accessing delay.The proposed architecture is verified and tested by parallel implementation of some CNN algorithms.The experimental results show that the peak bandwidth can reach 13.44 GB·s^(-1)at an operating frequency of 120 MHz.This work can achieve 1.40,1.12,2.80 and 4.70 times the peak bandwidth compared with the existing work.

Design and implementation of instruction-driven and data-driven self-reconfigurable cell array

The reconfigurable chip,which integrates the advantages of high performance,high flexibility,high parallelism,low power consumption,and low cost,has achieved rapid development and wide application.Generally,the control part and the computing part of algorithm is accelerated based on different reconfigurable architectures,but it is difficult to obtain overall performance improvement.For improving efficiency of reconfigurable structure both for the control part and the computing part,a hybrid of instruction-driven and data-driven self-reconfigurable cell array is proposed.On instruction-driven mode,processing element(PE)works like a reduced instruction set computer(RSIC)machine,which is mainly for the control part of algorithm.On data-driven mode,data is calculated by flowing between the preconfigured PEs,which is mainly for the computing of algorithm.For verifying the efficiency of architecture,some high-efficiency video coding(HEVC)video compression algorithms are implemented on the proposed architecture.The proposed architecture has been implemented on Xilinx FPGA Virtex UltraScale VU440 develop board.The same circuitry is able to run at75 MHz.Compared with the architecture that only supports instruction-driven,the proposed architecture has better calculation efficiency.

Cooperative Optimization of Reconfigurable Machine Tool Configurations and Production Process Plan

The production process plan design and configurations of reconfigurable machine tool （RMT） interact with each other. Reasonable process plans with suitable configurations of RMT help to improve product quality and reduce production cost. Therefore, a cooperative strategy is needed to concurrently solve the above issue. In this paper, the cooperative optimization model for RMT configurations and production process plan is presented. Its objectives take into account both impacts of process and configuration. Moreover, a novel genetic algorithm is also developed to provide optimal or near-optimal solutions： firstly, its chromosome is redesigned which is composed of three parts, operations, process plan and configurations of RMTs, respectively; secondly, its new selection, crossover and mutation operators are also developed to deal with the process constraints from operation processes （OP） graph, otherwise these operators could generate illegal solutions violating the limits; eventually the optimal configurations for RMT under optimal process plan design can be obtained. At last, a manufacturing line case is applied which is composed of three RMTs. It is shown from the case that the optimal process plan and configurations of RMT are concurrently obtained, and the production cost decreases 6.28% and nonmonetary performance increases 22%. The proposed method can figure out both RMT configurations and production process, improve production capacity, functions and equipment utilization for RMT.

GERO: a general SCA-based readout ASIC for micro-pattern gas detectors with configurable storage depth and on-chip digitizer

The paper presents GEneral ReadOut (GERO), a general readout ASIC based on a switched capacitor array for micro-pattern gas detectors. It aims at providing general readout electronics for low-to-medium event-rate gas detectors with high sampling frequency, configurable storage depth, and data digitalization. The first prototype GERO chip integrates 16 channels and was fabricated using a 0.18-lm CMOS process. Each channel consists of a sampling array working in a ping-pong mode, a storage array with a 1024-cell depth, and 32 Wilkinson analog-todigital converters. The detailed design and test results are presented in the paper.

Configuration Selection for Reconfigurable Manufacturing Systems by Means of Characteristic State Space

The configuration selection for reconfigurable manufacturing systems（RMS） have been tackled in a number of studies by using analytical or simulation models. The simulation models are usually based on fewer assumptions than the analytical models and therefore are more wildly used in modeling complex RMS. But in the absence of an efficient gradient analysis method of the objective function, it is time-consuming in solving large-scale problems by using a simulation model coupled with a meta-heuristics algorithm. In this paper, a new approach by means of characteristic state space is presented to improve the efficiency of the configuration selection for an RMS. First, a characteristic state equation is set up to represent the input and the output resources of each basic activity in an RMS. A production process model in terms of matrix equations is established by iterating the equations of basic activities according to the resource flows. This model introduces the production process into a characteristic state space for further analysis. Second, the properties of the characteristic state space are presented. On the basis of these properties, the configuration selection in an RMS is considered as a path-planning problem, and the gradient of the objective function is computed. Modified simulated annealing（SA） is also presented, in which neighborhood generation is guided by the gradient to accelerate convergence and reduce the run time of the optimization procedure. Finally, several case studies on the configuration selection for some actual reconfigurable assembly job-shops are presented and compared to the classical SA. The comparison shows relatively positive results. This study provides a more efficient configuration selection approach by using the gradient of the objective function and presents the relevant theories on which it is based.

Organization-Driven Business Process Configurable Modeling for Spatial Variability

Nowadays,enterprises need to continually adjust their business processes to adapt to the changes of business environments,especially when one business needs to be deployed in different application scenarios,which is called spatial variability in this paper.In the field of BPM(Business Process Management),configurable business process models have demonstrated their effectiveness in aspects of process modeling and model reuse.Yet,we found that the existing techniques lead to complex configurable models,and are inadequate for model reuse especially for the spatial variability issue because they neglect the root impact of organizations on control flow.S-BPM(Subject-oriented Business Process Management)models provide a solid foundation for dealing with complex applications and help to bridge the gap between business and IT for process execution.In this paper,we propose an organization-driven business process configurable modeling approach for spatial variability by integrating both restriction and extension operations based on the S-BPM paradigm,in which business objects are also included.Our approach is validated with a general business process developed for the Real Estate Administration(REA)in a certain province of China.The resulting configurable modeling framework can express the heterogeneous activity sequences for one business and has the potential to generate process models for uncertain environments in a new organization structure.

Research on the Structural Rigidity Characteristics of a Reconfigurable TBM Thrust Mechanism

To improve the adaptability of TBMs in diverse geological environments,this paper proposes a reconfigurable Type-V thrust mechanism(V-TM)with rearrangeable working states,in which structural stiffness can be automatically altered during operation.Therefore,millions of configurations can be obtained,and thousands of instances of working status per configuration can be set respectively.Nonetheless,the complexity of configurations and diversity of working states contributes to further complications for the structural stiffness algorithm.This results in challenges such as difficulty calculating the payload compliance index and the environment adaptability index.To solve this problem,we use the configuration matrix to describe the relationship between propelling jacks under reconfiguration and adopt pattern vectors to describe the working state of each hydraulic cylinder.Then,both the dynamic compatible equation between propeller forces of the hydraulic cylinders and driving forces,and the kinematic harmonizing equation between the hydraulic cylinder displacements and their deformations are established.Next,we derive the stiffness analytical equation using Hooke’s law and the Jacobian Matrix.The proposed approach provides an effective algorithm to support structural rigidity analysis,and lays a solid theoretical foundation for calculating the performance indexes of the V-TM.We then analyze the rigidity characteristics of typical configurations under different working states,and obtain the main factors affecting structural stiffness of the V-TM.The results show the deviation degree of structural parameters in hydraulic cylinders within the same group,and the working status of propelling jacks.Finally,our constructive conclusions contribute valuable information for matching and optimization by drawing on the factors that affect the structural rigidity of the V-TM.

A Study of Performance Testing in Configurable Software Systems

Customizing applications through program configuration options has been proved by many open-source and commercial projects as one of the best practices in software engineering. However, traditional performance testing is not in synch with this industrial practice. Traditional performance testing techniques consider program inputs as the only external factor. It ignores the performance influence of configuration options. This study aims to stimulate research interest in performance testing in the context of configurable software systems by answering three research questions. That is, why it is necessary to conduct research in performance testing, what are the state-of-the-art techniques, and how do we conduct performance testing research in configurable software systems. In this study, we examine the unique characteristics and challenges of performance testing research in configurable software systems. We review and discuss research topics on the performance bug study, performance anti-patterns, program analysis, and performance testing. We share the research findings from the empirical study and outline the opening opportunities for new and advanced researchers to contribute to the research community.

Configurable Enterprise:A New Trend in Manufacturing Enterprise Evolution

Along with the popularity of enterprise applications,the management reform of manufacturing enterprise(ME) relies on the business process reengineering is gradually invalid.Reviewing and comparison the former structure of different enter- prise such as extend enterprise,virtual enterprise etc.from the perspective of link,unit,and focus.Through comparison the evo- lution of ME structure the configurable enterprise(CE)was put forward which based on the information and communication tech- nology providing a new mechanism for the elements configuration and the resources allocation.Based on the reference model of en- terprise business architecture,the framework of CE was raised,which includes enterprise business architecture,business model and component.Then modeling the dynamics of CE,the mechanism of key success factors interactions was analyzed and verified the key factors that de decided the evolution of CE which could forecast the development trend of manufacturing enterprise.

Powering Next-Generation Broadband Networks: ZTE's World-First Flexible, Configurable Router

16 September 2013, Shenzhen--ZTE today unveiled the world＇s first flexible, reconfigurable terabit router that allows customers to build the highest-performance broadband networks. The terabit router supports the deployment of multiple line cards with processing capabilities of 10 Gbps to 1 Tbps. It also supports the deployment of modules that can scale throughput from 200 Gbps to 18 Tbps. For easy installation in a range of environments, the router interfaces are flexible and the component design is loose-coupled. This allows customers to customize networks to their needs and promotes adaptability, consistency, and continuity.

Disassemblability Modeling Technology of Configurable Product Based on Disassembly Constraint Relation Weighted Design Structure Matrix(DSM)

The current research of configurable product disassemblability focuses on disassemblability evaluation and disassembly sequence planning. Little work has been done on quantitative analysis of configurable product disassemblability. The disassemblability modeling technology for configurable product based on disassembly constraint relation weighted design structure matrix （DSM） is proposed. Major factors affecting the disassemblability of configurable product are analyzed, and the disassembling degrees between components in configurable product are obtained by calculating disassembly entropies such as joint type, joint quantity, disassembly path, disassembly accessibility and material compatibility. The disassembly constraint relation weighted DSM of configurable product is constructed and configuration modules are formed by matrix decomposition and tearing operations. The disassembly constraint relation in configuration modules is strong coupling, and the disassembly constraint relation between modules is weak coupling, and the disassemblability configuration model is constructed based on configuration module. Finally, taking a hydraulic forging press as an example, the decomposed weak coupling components are used as configuration modules alone, components with a strong coupling are aggregated into configuration modules, and the disassembly sequence of components inside configuration modules is optimized by tearing operation. A disassemblability configuration model of the hydraulic forging press is constructed. By researching the disassemblability modeling technology of product configuration design based on disassembly constraint relation weighted DSM, the disassembly property in maintenance, recycling and reuse of configurable product are optimized.

DESIGN OF RECONFIGURABLE MANUFACTURING SYSTEMS WITH STRONGLY COUPLED NATURE

Today＇s manufacturing cnvironmem forces manufacturing companies to make as many product variations as possible at affordable costs within a short time. Mass customisation is one of most important technologies for companies to achieve their objectives. Efforts to mass customisation should be made on two aspects： （1） To modularize products and make them as less differences as possible; （2） To design manufacturing resources and make them provide as many processes variations as possible. This paper reports our recent work on aspect （2）, i.e. how to design a reconfignrable manufacturing system （RMS） so that it can be competent to accomplish various processes optimally; Reconfignrable robot system （RRS） is taken as an example. RMS design involves architecture design and configuration design, and configuration design is further divided in design analysis and design synthesis. Axiomatic design theory （ADT） is applied to architecture design, the features and issues of RRS configuration design are discussed, automatic modelling method is developed for design analysis, and concurrent design methodology is presented for design synthesis.

Reconfigurable Modeling Technology of Similar-structure Products Configuration Design

The product configuration modeling is a key technology to realize configuration design. Most of the researches were mainly conducted focusing on how to establish a product configuration model, but were short of the consideration about the application scope and dynamic change features. In this paper, the reconfigurable modeling technology of similar-structure products configuration design is studied to enhance the generality and diversity of product configuration models, and to facilitate the dynamic upgrade of configuration models. The configuration components of products with similar structures are divided into sharable configuration components, essential configuration components and optional configuration components, so the construction of reconfigurable model of configuration design can be converted into the set operation of configuration components. The horizontal adjusting reconstruction of configuration model is carried out through the adding, replacing and deleting of configuration components, the vertical adjusting reconstruction of configuration model is carried out through the parametric deformation of configuration components, and the variable topological configuration model reconstruction can be realized through the transformation of partial structure of configuration components. Finally, NC machine products are used as an example to analyze and verify the reconfigurable modeling technology of products with similar structures. The speed of NC machine configuration design is found to be improved, and the number of NC machine configuration design models is decreased. And the reconfigurable design system of NC machine products is developed. The proposed modeling technology provides references for the processes, methods and steps of model reconstruction.

Configurable Product Design Considering the Transition of Multi-hierarchical Models

The current research of configurable product design mainly focuses on how to convert a predefined set of components into a valid set of product structures. With the scale and complexity of configurable products increasing, the interdependencies between customer demands and product structures grow up as well. The result is that existing product structures fails to satisfy the individual customer requirements and hence product variants are needed. This paper is aimed to build a bridge between customer demands and product structures in order to make demand-driven fast response design feasible. First of all, multi-hierarchical models of configurable product design are established with customer demand model, technical requirement model and product structure model. Then, the transition of multi-hierarchical models among customer demand model, technical requirement model and product structure model is solved with fuzzy analytic hierarchy process （FAHP） and the algorithm of multi-level matching. Finally, optimal structure according to the customer demands is obtained with the calculation of Euclidean distance and similarity of some cases. In practice, the configuration design of a clamping unit of injection molding machine successfully performs an optimal search strategy for the product variants with reasonable satisfaction to individual customer demands. The proposed method can automatically generate a configuration design with better alternatives for each product structures, and shorten the time of finding the configuration of a product.

Two-dimensional complementary gate-programmable PN junctions for reconfigurable rectifier circuit

The unique features of ambipolar two-dimensional materials open up a great opportunity to build gate-programmable devices for reconfigurable circuit applications,e.g.,PN junctions for rectifier circuits.However,current-reported rectifier circuits usually consist of one gate-programmable PN junction as the rectifier and one resistor as the load,which are not conductive to voltage output and large-scale integration.Here we propose an approach of complementary gate-programmable PN junctions to assemble reconfigurable rectifier circuit,which include two symmetric back-to-back black phosphorus(BP)/hexagonal boron nitride(h-BN)/graphene heterostructured semi-gate field-effect transistors(FETs)and perform complementary NP and PN junction like complementary metal-oxide-semiconductor(CMOS)circuit.The investigation exhibits that the circuit can effectively reconfigure the circuit with/without rectifying ability,and can process alternating current(AC)signals with the frequency prior 1 KHz and reconfiguration speed up to 25μs.We also achieve the reconfigurable rectifier circuit memory via complementary semi-floating gate FETs configuration.The complementary configuration here should be of low output impedance and low static power consumption,being beneficial for effective voltage output and large-scale integration.

Dynamical Self-Reconfigurable Mechanism for Data-Driven Cell Array

The utilization of computation resources and reconfiguration time has a large impact on reconfiguration system performance. In order to promote the performance, a dynamical self-reconfigurable mechanism for data-driven cell array is proposed. Cells can be fired only when the needed data arrives, and cell array can be worked on two modes: fixed execution and reconfiguration. On reconfiguration mode, cell function and data flow direction are changed automatically at run time according to contexts. Simultaneously using an H-tree interconnection network, through pre-storing multiple application mapping contexts in reconfiguration buffer, multiple applications can execute concurrently and context switching time is the minimal. For verifying system performance, some algorithms are selected for mapping onto the proposed structure, and the amount of configuration contexts and execution time are recorded for statistical analysis. The results show that the proposed self-reconfigurable mechanism can reduce the number of contexts efficiently, and has a low computing time.

Configuration analysis of a chain-type reconfigurable modular robot inspired by normal alkane

Normal alkane is an unbranched alkane whose structural formula is H–CH2–CH2–…–CH2–…–CH2–H,which can be regarded as a reconfigurable chain-type structure composed of–CH2–modules.Inspired by normal alkane,a normal-alkane-like reconfigurable modular robot (NAR) is proposed.The module consists of two differential gear trains mounted orthogonally.Each differential gear train contains two input degrees of freedom and two output degrees of freedom.Due to the genderless interface design,multiple modules can be assembled into chain-type configuration.With the genderless interfaces and flexible degrees of freedom,NAR can be reconfigured into different dimensions of spatial configuration.The bond matrix is used to describe the configuration,which represents the bond attitude of the adjacent connected modules.In addition,full interconnected geometric feature (FIGF) algorithm is proposed for non-isomorphic configuration enumeration and judgment.The configurations with three modules are simulated and the results verify the feasibility of the algorithm.Finally,a prototype with three modules is fabricated and the configuration motion sequence is demonstrated.

Network-based reconfiguration routes for a self-reconfigurable robot

This paper presents a network-based analysis approach for the reconfiguration problem of a self-reconfigurable robot. The self-reconfigurable modular robot named ＂AMOEBA-I＂ has nine kinds of non-isomorphic configurations that consist of a configuration network. Each configuration of the robot is defined to be a node in the weighted and directed configuration network. The transformation from one configuration to another is represented by a directed path with nonnegative weight. Graph theory is applied in the reconfiguration analysis, where reconfiguration route, reconfigurable matrix and route matrix are defined according to the topological information of these configurations. Algorithms in graph theory have been used in enumerating the available reconfiguration routes and deciding the best reconfiguration route. Numerical analysis and experimental simulation results prove the validity of the approach proposed in this paper. And it is potentially suitable for other self-reconfigurable robots＇ configuration control and reconfiguration planning.

Center-configuration selection technique for the reconfigurable modular robot

The reconfigurable modular robot has an enormous amount of configurations to adapt to various environments and tasks. It greatly increases the complexity of configuration research in that the possible configuration number of the reconfigurable modular robot grows exponentially with the increase of module number. Being the initial configuration or the basic configuration of the reconfigurable robot, the center-configuration plays a crucial role in system＇s actual applications. In this paper, a novel center-configuration selection technique has been proposed for re- configurable modular robots. Based on the similarities between configurations＇ transformation and graph theory, configuration network has been applied in the modeling and analyzing of these configurations. Configuration adjacency matrix, reconfirmation cost matrix, and center-configuration coefficient have been defined for the configuration network correspondingly. Being similar to the center-location problem, the center configuration has been selected according to the largest center-configuraUon coefficient. As an example of the reconfigurable robotic system, AMOEBA-I, a three-module reconfigurable robot with nine configurations which was developed in Shenyang Institute of Automation （SIA）, Chinese Academy of Sciences （CAS）, has been introduced briefly. According to the numerical simulation result, the center-configuration coefficients for these nine configurations have been calculated and compared to validate this technique. Lastly, a center-configuration selection example is provided with consideration of the adjacent configurations. The center-configuration selection technique proposed in this paper is also available to other reconfigurable modular robots.

Configuration Reusing in On-Line Task Scheduling for Reconfigurable Computing Systems

Reconfigurable computing systems can be reconfigured at runtime and support partial reconfigurability which makes us able to execute tasks in a true multitasking manner. To manage such systems at runtime, a reconfigurable operating system is needed. The main part of this operating system is resource management unit which performs on-line scheduling and placement of hardware tasks at runtime. Reconfiguration overhead is an important obstacle that limits the performance of on-line scheduling algorithms in reconfigurable computing systems and increases the overall execution time. Configuration reusing （task reusing） can decrease reconfiguration overhead considerably, particularly in periodic applications or the applications in which the probability of tasks recurrence is high. In this paper, we present a technique called reusing-based scheduling （RBS）, for on-line scheduling and placement in which configuration reusing is considered as a main characteristic in order to reduce reconfiguration overhead and decrease total execution time of the tasks. Several experiments have been conducted on the proposed algorithm. Obtained results show considerable improvement in overall execution time of the tasks.