In this study we analyzed some aspects of applying the fieldprogrammable Gate Array (FPGA) technology based on Peripheral Component Interconnect Express (PCIe) bus to create a cloud service. Task classification for a...
详细信息
In this study we analyzed some aspects of applying the fieldprogrammable Gate Array (FPGA) technology based on Peripheral Component Interconnect Express (PCIe) bus to create a cloud service. Task classification for a FPGA as a Service (FaaS) was proposed. We considered various approaches to FaaS deploying and feasible ways of communication between the cloud infrastructure and the FPGA platform. We elaborated and approved cost-effective FaaS architecture, which is based on a set of FPGA boards. The input-oriented task based on brute force search of polynomials for nonlinear feedback shift registers of the second degree was implemented. The approach of creating multiparametrized tasks for a wide range of FPGA resources was proved to be effective.
Quantitative effects of Moore's Law have driven qualitative changes in FPGA architecture, applications, and tools. As a consequence, the existing EDA tools takes several hours or even days to implement the applica...
详细信息
ISBN:
(纸本)9781538622544
Quantitative effects of Moore's Law have driven qualitative changes in FPGA architecture, applications, and tools. As a consequence, the existing EDA tools takes several hours or even days to implement the applications onto FPGAs. Typically, routing is a very time-consuming process in the EDA design flow. While several attempts have accelerated this process through parallelization, they still do not provide a strong parallel scheme for FPGA routing. In this paper we introduce a dependency-aware parallel approach, named Bamboo, to accelerate the routing time for FPGAs. With the dependency detection, Bamboo partitions the nets into multiple subsets, where the nets in the same subsets are independent, and the dependency only exists among different subsets. Specifically, the independent nets in the same subset are routed in parallel, and the subsets are processed in serial according to the original routing ordering. The partitioning problem is solved optimally using dynamic programming, and the parallelization is implemented by speculative parallelism on a single GPU. Experimental results show that our approach achieves an average of 15.13x speedup with negligible influence on the routing quality. Most importantly, it effectively maintains deterministic results and always produces the same results as the serial version.
In contemporary mechatronic applications decision-making is often based on information about the underlying model governing the dynamical evolution, in order to ensure optimal operation with respect to a prioritized o...
详细信息
ISBN:
(纸本)9781509060009
In contemporary mechatronic applications decision-making is often based on information about the underlying model governing the dynamical evolution, in order to ensure optimal operation with respect to a prioritized objective. Modeling errors stemming from parameter uncertainty or varying operational conditions result in inevitable deviations from the theoretical estimate and consequently in suboptimal operation. Intelligent systems need to be equipped with inherent means to compensate for these a priori unknown discrepancies, hereby guaranteeing a robust operation in uncertain environments. In this manuscript, advanced filtering techniques are applied to assess both an optimal model representation and state estimates. An appropriate interconnection between both model and state estimation is determined. The proposed methodology is demonstrated for an electric drive, embodying a DC-source, a voltage source inverter (VSI) and an asynchronous machine, as the presence of discrete switching sequences and physical constraints introduces additional challenges. Results prove that the error on the state estimates can be improved by 92.7-97.2%, outperforming the classical estimation techniques, while the relative model mismatch is scaled down to 0.03%, even in highly demanding scenarios. The introduced strategy thus enables high-fidelity virtual sensing and reliable decision-making procedures for advanced asynchronous drives when modeling errors can be anticipated.
FPGAs can provide high performance and energy efficiency to many applications;therefore, they are attractive computing platforms in a cloud environment. However, FPGA application development requires extensive hardwar...
详细信息
ISBN:
(纸本)9782839918442
FPGAs can provide high performance and energy efficiency to many applications;therefore, they are attractive computing platforms in a cloud environment. However, FPGA application development requires extensive hardware design knowledge which significantly limits the potential user base. Moreover, in a cloud setting, allocating a whole FPGA to a user is often wasteful and not cost effective due to low device utilization. To make FPGA application development easier, firstly, we propose a methodology that provides clean abstractions with high-level APIs and a simple execution model that supports both software and hardware execution. Secondly, to improve device utilization and share the FPGA among multiple users, we developed a lightweight runtime system that provides hardware-assisted memory virtualization and memory protection, enabling multiple applications to simultaneously execute on the device.
Key-value stores (KVS) become critical in many applications because of the data explosion recently. There is a strong demand to improve the throughput and reduce the latency for KVS. FPGA-based parallel architecture c...
详细信息
ISBN:
(纸本)9782839918442
Key-value stores (KVS) become critical in many applications because of the data explosion recently. There is a strong demand to improve the throughput and reduce the latency for KVS. FPGA-based parallel architecture can bring excellent performance and power efficiency. Cuckoo hashing has proven to be an efficient approach to implement KVS with good memory utilization and constant worst case access time. In this paper, an FPGA-based KVS implementation is proposed based on Cuckoo hashing, with a decoupled storage to achieve 81.7% memory utilization, and a pipeline scheme to achieve high performance. The latency of insert, search and delete operations is only 40 ns. And the throughput for search and delete can be 200 million requests per second (MRPS) which is 5x faster than [1]. Even when the load factor becomes 0.9, the throughput for insert can still achieve 147 MRPS.
The genetic engineering of microbial organisms offers benefits to society through biotechnology applications. Traditionally, the "engineering" of microbes to arrive at organisms with desired behavior has not...
详细信息
ISBN:
(纸本)9781509063895
The genetic engineering of microbial organisms offers benefits to society through biotechnology applications. Traditionally, the "engineering" of microbes to arrive at organisms with desired behavior has not been engineering in a strict sense. It has, rather, required months (more often years) of trial-and-error type of experiments, with the undertaking being more akin to art than engineering. Enter synthetic biology, a burgeoning area since the turn of the century that aims to put the engineering into genetic engineering. Here, we provide a short commentary on some advancements in this field. By relating these advances to recent progress in our understanding of extracellular electron transfer in bacteria, we also provide a perspective on synthetic biology having the potential to enable the programming of bacteria for electronics engineering-related applications such as biosensors.
Feedback control of mobile robots guaranteeing preservation of state constraints resulting from obstacles in the environment and input constraints imposed by robot mechanical construction is essential in practical app...
详细信息
ISBN:
(纸本)9781509060009
Feedback control of mobile robots guaranteeing preservation of state constraints resulting from obstacles in the environment and input constraints imposed by robot mechanical construction is essential in practical applications of robotic systems. The safety of motion execution is often ensured by a strategy of driving the robot through a sequence of funnels representing safe, positively invariant subsets of robot configuration space for utilized feedback control laws. In this paper, the VFO (Vector field Orientation) control law is leveraged to develop such a feedback control strategy for a unicycle robot with bounded curvature of motion. The proposed definition of funnels arises naturally from analysis of the VFO control law under curvature constraints. Obstacles in the environment are handled by shrinking the funnels using additional artificial curvature constraints. An exact analytic method for computation of funnels is presented. To make the funnels positively-invariant and guarantee motion safety, the original VFO control law has been modified. In contrast to numerous methods available in the literature, proposed feedback control strategy ensures at least C-1 continuity of the control signals during transitions between funnels. Effectiveness of our approach has been verified by simulations, during which the robot was driven through a sequence of funnels planned in the cluttered environment using the RRT* algorithm.
We demonstrate a novel FPGA-based accelerator architecture that can tackle a range of standard computer vision (CV) problems, with scalable performance and attractive speedups. The architecture relies on multiple pipe...
详细信息
ISBN:
(纸本)9782839918442
We demonstrate a novel FPGA-based accelerator architecture that can tackle a range of standard computer vision (CV) problems, with scalable performance and attractive speedups. The architecture relies on multiple pipelined processing elements (PEs) that can be configured to support various belief propagation (BP) settings for different CV tasks. Inside each PE, innovative implementation of Jump Flooding for efficient computation of BP solves the core configurability challenge. A novel block-parallel memory interface supports parallelization by distributing BP inference workloads across the PEs. Experimental results demonstrate that our accelerator achieves scalable performance with 11-41x speedup over standard sequential CPU implementations across a subset of well-known Middlebury and OpenGM benchmarks, with no compromise in quality of inference results. To the best of our knowledge, this is the first FPGA hardware implementation of BP capable of running a range of standard CV benchmarks with significant speedups.
Combinatorial interaction testing (CIT) stands as one of the efficient testing techniques that have been used in different applications recently. The technique is useful when there is a need to take the interaction of...
详细信息
ISBN:
(纸本)9781538620724
Combinatorial interaction testing (CIT) stands as one of the efficient testing techniques that have been used in different applications recently. The technique is useful when there is a need to take the interaction of input parameters into consideration for testing a system. The key insight the technique is that not every single parameter may contribute to the failure of the system and there could be interactions among these parameters. Hence, there must be combinations of these input parameters based on the interaction strength. This technique has been used in many applications to assess its effectiveness. In this paper, we are addressing the effectiveness of CIT for a real-world case study using model-based mutation testing experiments. The contribution of the paper is threefold: First we introduce an effective testing application for CIT;Second, we address the effectiveness of increasing the interaction strength beyond the pairwise (i.e., interaction of more than two parameters);Third, model-based mutation testing is used to mutate the input model of the program in contrast to the traditional code-based mutation testing process. Experimental results showed that CIT is an effective testing technique for this kind of application. In addition, the results also showed the usefulness of model-based mutation testing to assess CIT applications. For the subject of this case study, the results also indicate that 3 - way test suite (i.e., interaction of three parameters) could detect new faults that can not be detected by pairwise.
The purpose of this paper is to make a case for the value of many-valued mathematics, often called fuzzy mathematics. We believe there may be a difference between many-valued mathematics and fuzziness, as used by thos...
详细信息
ISBN:
(纸本)9781509060344
The purpose of this paper is to make a case for the value of many-valued mathematics, often called fuzzy mathematics. We believe there may be a difference between many-valued mathematics and fuzziness, as used by those who work with fuzzy logic and fuzzy set theory and applications thereof. We think that most, if not all, fuzzy mathematics is many-valued. However, for this paper, the difference between many-valued mathematics and fuzzy mathematics, if a difference exists, is not important. We are, in this paper, content to show that many-valued mathematics can contribute to mathematics. We do understand that for those mathematicians who feel that many-valued mathematics does not have a place in mathematics this paper will not cause them to embrace many-valued mathematics, but we would like them to consider that many-valued mathematics might be able to contribute to mathematics. In this paper, we give an example of a mathematical construction which was created and defined in part to help computer scientists understand and be able to use topological ideas and concepts in their work as computer scientists. Thus, one would think that this construction, called topological systems, would be topological (as defined later). However, it seems that topological systems are clearly not topological. Thus, an interesting question is can topological systems be made topological, or said more mathematically, can topological systems be embedded into something which is topological. We answer this question in the affirmative, and we do it by embedding topological systems into something which is many-valued. It may be the case that someone(s) can some day show that topological systems are topological though this seems unlikely. Or it may be the case that someone(s) can embed topological systems into something which is topological but not many-valued. However, our point is that by using something which is many-valued we have added to mathematics, and thus, we have shown a mathematical us
暂无评论