Using FPGAs as hardware accelerators that communicate with a central CPU is becoming a common practice in the embedded design world but there is no standard methodology and toolset to facilitate this path yet. On the other hand, languages such as CUDA and OpenCL provide standard development environments for Graphical Processing Unit (GPU) programming. FASTCUDA is a platform that provides the necessary software toolset, hardware architecture, and design methodology to efficiently adapt the CUDA approach into a new FPGA design flow. With FASTCUDA, the CUDA kernels of a CUDA-based application are partitioned into two groups with minimal user intervention: those that are compiled and executed in parallel software, and those that are synthesized and implemented in hardware. A modern low power FPGA can provide the processing power (via numerous embedded micro-CPUs) and the logic capacity for both the software and hardware implementations of the CUDA kernels. This paper describes the system requirements and the architectural decisions behind the FASTCUDA approach.

The rising complexity of modern embedded systems is causing a significant increase in the verification effort required by hardware designers and software developers, leading to the "design verification crisis ", as it is known among engineers. Today's verification challenges require powerful testbenches and high-performance simulation solutions such as Hardware Simulation Accelerators and Hardware Emulators that have been in use in hardware and electronic system design centers for approximately the last decade. In particular, in order to accelerate functional simulation, hardware emulation is used so as to offload calculation-intensive tasks from the software simulator. However, the communication overhead between the software simulator and hardware emulator is becoming a new critical bottleneck. We tackle this problem by partitioning the code running on the software simulator into two sections: the testbench HDL (hardware description language) code that communicates directly with the design under test (DUT) and the rest C-like testbench code. The former section is transformed into synthesizable code while the latter runs in a general purpose CPU. Our experiments demonstrate that the proposed method reduces the communication overhead by a factor of about 5 compared to a conventional hardware emulated simulation

In this paper we discuss and analyze the FPGA-based implementation of an algorithm for the traveling salesman problem (TSP), and in particular of 2-Opt, one of the most famous local optimization algorithms, for Euclidean TSP instances up to a few hundred cities. We introduce the notion of "symmetrical 2-Opt moves" which allows us to uncover fine-grain parallelism when executing the specified algorithm. We propose a novel architecture that exploits this parallelism, and demonstrate its implementation in reconfigurable hardware. We evaluate our proposed architecture and its implementation on a state-of-the-art FPGA using a subset of the TSPLIB benchmark, and find that our approach exhibits better quality of final results and an average speedup of 600% when compared with the state-of-the-art software implementation. Our approach produces, to the best of our knowledge, the fastest to date TSP 2-Opt solver for small-scale Euclidean TSP instances.

Ethernet technology is lo longer used only in Local Area Networks (LANs); it is continuously gaining momentum in the Metropolitan Area Networks (MAN s) and Wide Area Networks (WANs). This paper presents a multi-service access concentrator core that has been designed specifically for multi-service, purely Ethernet, access nodes. In particular the presented system is optimised for Ethernet traffic aggregation over MPLS-based optical backbone networks. Moreover, we also demonstrate the bottlenecks that have been identified when such networking applications are executed in a general-purpose network processing device, and the techniques we used in order to bypass them. Our experiments results, executed in a state-of-the-art FPGA-based platform, strongly support that the combination of general purpose processing units with powerful specialized hardware modules, is the most cost-effective approach for designing systems that (i) can support today's and future network speeds and applications, in purely Ethernet networks, (ii) provide the end-user with the required programmability.

One of the main bottlenecks when designing a network processing system is very often its memory subsystem. This is mainly due to the state-of-the-art network links operating at very high speeds and to the fact that in order to support advanced quality of service (QoS), a large number of independent queues is desirable. In this paper we analyze the performance bottlenecks of various data memory managers integrated in typical network processing units (NPU). We expose the performance limitations of software implementations utilizing the RISC processing cores typically found in most NPU architectures and we identify the requirements for hardware assisted memory management in order to achieve wire-speed operation at gigabit per second rates. Furthermore, we describe the architecture and performance of a hardware memory manager that fulfills those requirements. This memory manager, although it is implemented in a reconfigurable technology, can provide up to 6.2 Gbit/s of aggregate throughput, while handling 32 K independent queues.

High-speed switches and routers internally operate using fixed-size cells or segments; variable-size packets are segmented and later reassembled. Connectionless ATM was proposed to quickly carry IP packets segmented into cells (AAL5) using a number of hardware-managed ATM VCs. We show that this is analogous to wormhole routing. We modify this architecture to make it applicable to existing ATM equipment: we propose a low-cost, single-input, single-output wormhole IP router that functions as a VP/VC translation filter between ATM subnetworks. When compared to IP routers, the proposed architecture features simpler hardware and lower latency. When compared to software-based IP-over-ATM techniques, the new architecture avoids the overheads of a large number of labels, or the delays of establishing new flows in software after the first few packets have suffered considerable latencies. We simulated a wormhole IP routing filter, showing that a few tens of hardware-managed VCs per outgoing VP usually suffice. We built and successfully tested a prototype, operating at 2/spl times/155 Mb/s, using one field programmable gate array (FPGA) and DRAM. Simple analysis shows that operation at 10 Gb/s and beyond is feasible today.