Diana Göhringer

Fast dynamic and partial reconfiguration data path with low hardware overhead on Xilinx FPGAs

Dynamic and partial reconfiguration of Xilinx FPGAs is a well known technique in runtime adaptive system design. With this technique, parts of a configuration can be substituted while other parts stay operative without any disturbance. The advantage is the fact, that the spatial and temporal partitioning can be exploited with the goal to increase performance and to reduce power consumption due to the re-use of chip area. This paper shows a novel methodology for the inclusion of the configuration access port into the data path of a processor core in order to adapt the internal architecture and to re-use this access port as data- sink and source. It is obvious that the chip area, which is utilized by the hardware drivers for the internal configuration access port (ICAP), has to be as small as possible in comparison to the application functionality. Therefore, a hardware design with a small footprint, but with an adequate performance in terms of data throughput, is necessary. This paper presents a fast data path for dynamic and partial reconfiguration data with the advantage of a small footprint on the hardware resources.

A Design Methodology for Application Partitioning and Architecture Development of Reconfigurable Multiprocessor Systems-on-Chip

Until today, the efficient partitioning and mapping of applications for multiprocessor systems is a challenging task. The deployment of reconfigurable hardware in this domain helps to meet the application requirements more efficiently due to hardware adaptation at design and runtime, which is not applicable in the traditional multiprocessor domain. To exploit this novel degree of freedom in multiprocessor system-on-chip (MPSoC) technology, a novel design methodology is needed, which helps to hide the complexity of the hardware architecture and its realization alternatives from the developer. This paper shows one approach for such a design methodology for the development of the hardware architecture and the application partitioning and mapping. A novel multistep approach based on hierarchical clustering is used for partitioning of the software application and for configuration of a runtime adaptive multiprocessor system. Furthermore, each application module is then partitioned in a Hardware-Software Codesign process in order to achieve a maximum of performance on the local processors and therefore in general for the MPSoC.

High performance reconfigurable multi-processor-based computing on FPGAs

Multi-processor architectures are a promising solution to provide the required computational performance for applications in the area of high performance computing. Multi- and many-core Systems-on-Chip offer the possibility to host an application, partitioned in a number of tasks, on the different cores on one silicon die. Unfortunately, a partitioning of the tasks near to the performance optimum is the challenge in this domain and often a show-stopper for the success story of multi- and many-core hardware. The missing feature of these architectures is runtime adaptivity of the underlying hardware, which offers to tailor the hardware to the application in order to meet the task mapping process coming from top-down development. Especially, this Meet-in-the-Middle solution offers the novel hardware and software approach of RAMPSoC, which is described in this paper.

Operating system for runtime reconfigurable multiprocessor systems

Operating systems traditionally handle the task scheduling of one or more application instances on processor-like hardware architectures. RAMPSoC, a novel runtime adaptive multiprocessor System-on-Chip, exploits the dynamic reconfiguration on FPGAs to generate, start and terminate hardware and software tasks. The hardware tasks have to be transferred to the reconfigurable hardware via a configuration access port. The software tasks can be loaded into the local memory of the respective IP core either via the configuration access port or via the on-chip communication infrastructure (e.g. a Network-on-Chip). Recent-series of Xilinx FPGAs, such as Virtex-5, provide two Internal Configuration Access Ports, which cannot be accessed simultaneously. To prevent conflicts, the access to these ports as well as the hardware resource management needs to be controlled, e.g. by a special-purpose operating system running on an embedded processor. For that purpose and to handle the relations between temporally and spatially scheduled operations, the novel approach of an operating system is of high importance. This special purpose operating system, called CAP-OS (Configuration Access Port-Operating System), which will be presented in this paper, supports the clients using the configuration port with the services of priority-based access scheduling, hardware task mapping and resource management.

RAMPSoCVM: Runtime Support and Hardware Virtualization for a Runtime Adaptive MPSoC

Virtualizing complex hardware, such as heterogeneous multiprocessor systems, enables developers to use standard Application Programming Interfaces (APIs) for application integration. Especially, the supply of an Operating System (OS) is well appreciated since many features such as drivers, the runtime environment and scheduling mechanisms are available and well established. For this purpose, Embedded Linux was used as basis OS and extended in order to be able to manage a Runtime Adaptive Multi-Processor System-on-Chip (RAMPSoC) and to provide the standard Message Passing Interface (MPI). This paper describes the adaptation of the Linux kernel supporting MPI with runtime libraries as well as the integration of the software/hardware drivers which supply the message transfer over a reconfigurable and heterogeneous Network-on-Chip (NoC).

CAP-OS: Operating system for runtime scheduling, task mapping and resource management on reconfigurable multiprocessor architectures

Operating systems traditionally handle the task scheduling of one or more application instances on a processor like hardware architecture. Novel runtime adaptive hardware exploits the dynamic reconfiguration on FPGAs, where hardware blocks are generated, started and terminated. This is similar to software tasks in well established operating system approaches. The hardware counterparts to the software tasks have to be transferred to the reconfigurable hardware via a configuration access port. This port enables the allocation of hardware blocks on the FPGA. Current reconfigurable hardware, like e.g. Xilinx Virtex 5 provide two internal configuration access ports (ICAPs), where only one of these ports can be accessed at one point of time. In e.g. a multiprocessor system on an FPGA, it can happen that multiple instances try to access these ports simultaneously. To prevent conflicts, the access to these ports as well as the hardware resource management needs to be controlled by a special purpose operating system running on an embedded processor. This special purpose operating system, called CAPOS (Configuration Access Port-Operating System), which will be presented in this paper, supports the clients using the configuration port with the service of priority-based access scheduling, hardware task mapping and resource management.

New dimensions for multiprocessor architectures: On demand heterogeneity, infrastructure and performance through reconfigurability — the RAMPSoC approach

Multiprocessor hardware architectures enable to distribute tasks of an application to several microprocessors, in order to exploit parallelism for accelerating the performance of computation. Especially for the application domain of image data processing, where computation performance is a crucial factor to keep the real-time requirements, this approach is a promising solution for the assembly of high sophisticated algorithms e.g. for object tracking. Changing requirements and the necessary implementation of the tasks in terms of modified algorithms, precision and communication needs to be handled by software and hardware adaptation in state of the art architectures. Field programmable gate arrays (FPGAs) enable to exploit the adaptation of hardware cores and the software running on embedded microprocessor cores on an integrated multiprocessor system.

Star-Wheels Network-on-Chip featuring a self-adaptive mixed topology and a synergy of a circuit - and a packet-switching communication protocol

Multiprocessor System-on-Chip is a promising realization alternative for the next generation of computing architectures providing the required data processing performance in high performance computing applications. Numerous scientists from industry and academic institutions investigate and develop novel processing elements and accelerators as can be seen in real devices like IBMs Cell or nVIDIAs Tesla GPU. Nevertheless, the on-chip communication of these multiple processor elements has to be optimized tailored to the actual requirement of the data to be processed. Network-on-Chip (NoC), Bus-based or even heterogeneous communication on chip often suffer from the fact of being inflexible due to their fixed physical realization. This paper presents a novel approach for a NoC, exploiting circuit-and packed-switched communication as well as a run-time adaptive and heterogeneous topology. An application scenario from image processing exploiting the implemented NoC on an FPGA delivers results like performance data and hardware costs.

Dynamic and partial reconfiguration of Zynq 7000 under Linux

Dynamic and partial reconfiguration is a well-known technique to update the configuration of a field programmable gate array (FPGA) at runtime. Xilinx FPGAs support this feature which enables extensive research in this domain. However, until today the usage and exploitation of partial reconfiguration has a hurdle. The complex development process, as well as the required control at runtime keeps this technique away from many applications where it would be beneficial and lead to a reduction of costs and power consumption since a smaller FPGA can host more hardware modules due to a temporal partition and configuration in a time sequence. This paper shows an approach using the novel Zynq FPGA architecture from Xilinx. The partial reconfiguration is usable with a Linux realized on the dual core ARM 9 processor. A reconfigurable area provides space for accelerators which can be loaded and updated at runtime.

Adaptive real-time image processing exploiting two dimensional reconfigurable architecture

Fine grained reconfigurable architectures, like Xilinx field programmable gate arrays (FPGAs) provide a high flexibility through runtime re-programming, called dynamic and partial reconfiguration. This feature allows for runtime adaptation of the system architecture and behavior configured on the FPGA. The exploitation of this feature enables to load video image processing algorithms on-demand in order to adapt the configuration in correspondence to the changing requirements of the application depending on the image content. For high resolution sensor images, this novel computing paradigm can provide a huge benefit in power reduction and performance gain for actual and future embedded electronic systems. This paper presents a two dimensional system approach exploiting dynamic and partial reconfiguration in order to adapt the system architecture to the actual requirements of image processing applications. The methodology of runtime reconfiguration can be exploited beneficially for highly adaptive multiprocessor systems. Such systems, different from the traditional static approach for multi- and many-core architectures have the advantage, for providing computational performance directly linked to the requirements of the application. The architecture presented in this paper allows for adapting the processing elements as well as the communication infrastructure which is a novel 2D switch-based Network-on-Chip. The presented approach follows and extends the actual trend in computer science of using many- and multi-core processors for bridging the gap between required computation performance for future application in the field of image processing.