Carl Ebeling

RaPiD - Reconfigurable Pipelined Datapath

Configurable computing has captured the imagination of many architects who want the performance of application-specific hardware combined with the reprogrammability of general-purpose computers. Unfortunately, onfigurable computing has had rather limited success largely because the FPGAs on which they are built are more suited to implementing »ndom logic than computing tasks. This paper presents RaPiD, a new coarse-grained FPGA architecture that is optimized for highly repetitive, computation-intensive tasks. Very deep application-specific computation pipelines can be configured in RaPiD. These pipelines make much more efficient use of silicon than traditional FPGAs and also yield much higher performance for a wide range of applications.

Seeking solutions in configurable computing

Configurable computing offers the potential of producing powerful new computing systems. Will current research overcome the dearth of commercial applicability to make such systems a reality? Unfortunately, no system to date has yet proven attractive or competitive enough to establish a commercial presence. We believe that ample opportunity exists for work in a broad range of areas. In particular, the configurable computing community should focus on refining the emerging architectures, producing more effective software/hardware APIs, better tools for application development that incorporate the models of hardware reconfiguration, and effective benchmarking strategies.

Architecture design of reconfigurable pipelined datapaths

This paper examines reconfigurable pipelined datapaths (RaPiDs), a new architecture style for computation-intensive applications that bridges the cost/performance gap between general purpose and application specific architectures. RaPiDs can provide significantly higher performance than general purpose processors on a wide range of applications from the areas of video and signal processing, scientific computing and communications. Moreover, RaPiDs provide the flexibility that does not come with application-specific architectures. A RaPiD architecture is optimized for highly repetitive, computationally-intensive tasks. Very deep application-specific computation pipelines can be configured that deliver very high performance for a wide range of applications. RaPiDs achieve this using a coarse-grained reconfigurable architecture that mixes the appropriate amount of static configuration with dynamic control. We describe the fundamental features of a RaPiD architecture, including the linear array of functional units, a programmable segmented bus structure, and a programmable control architecture. In addition, we outline the floorplan of the architecture and provide timing data for the most critical paths. We conclude with performance numbers for several applications on an instance of a RaPiD architecture.

Placement and routing tools for the Triptych FPGA

Field-programmable gate arrays (FPGAs) are becoming an increasingly important implementation medium for digital logic. One of the most important keys to using FPGAs effectively is a complete, automated software system for mapping onto the FPGA architecture. Unfortunately, many of the tools necessary require different techniques than traditional circuit implementation options, and these techniques are often developed specifically for only a single FPGA architecture. In this paper we describe automatic mapping tools for Triptych, an FPGA architecture with improved logic density and performance over commercial FPGAs. These tools include a simulated-annealing placement algorithm that handles the routability issues of fine-grained FPGAs, and an architecture-adaptive routing algorithm that can easily be retargeted to other FPGAs. We also describe extensions to these algorithms for mapping asynchronous circuits to Montage, the first FPGA architecture to completely support asynchronous and synchronous interface applications.

Mapping applications to the RaPiD configurable architecture

The goal of the RaPiD (Reconfigurable Pipelined Datapath) architecture is to provide high performance configurable computing for a range of computationally-intensive applications that demand special-purpose hardware. This is accomplished by mapping the computation into a deep pipeline using a configurable array of coarse-grained computational units. A key feature of RaPiD is the combination of static and dynamic control. While the underlying computational pipelines are configured statically, a limited amount of dynamic control is provided which greatly increases the range and capability of applications that can be mapped to RaPiD. This paper illustrates this mapping and configuration for several important applications including a FIR filter, 2-D DCT, motion estimation, and parametric curve generation; it also shows how static and dynamic control are used to perform complex computations.

An FPGA for implementing asynchronous circuits

Field-programmable gate arrays are a dominant implementation medium for digital circuits, especially for glue logic. Unfortunately, they do not support asynchronous circuits. This is a significant problem because many aspects of glue logic and communication interfaces involve asynchronous elements, or require the interconnection of synchronous components operating under independent clocks. We describe Montage, the first FPGA to explicitly support asynchronous circuit implementation, and its mapping software. Montage can be used to realize asynchronous interface circuits or to prototype complete asynchronous systems, thus bringing the benefits of rapid prototyping to asynchronous design. Unfortunately, implementation media for asynchronous circuits and systems have not kept up with those for the synchronous world. Programmable logic devices do not include the special non-digital circuits required by asynchronous design methodologies (e.g., arbiters and synchronizers) nor do they facilitate hazard-free logic implementations. This leads to huge inefficiencies in the implementation of asynchronous designs as circuits require a variety of seperate devices. This has caused most asynchronous designers to focus on custom or semi-custom integrated circuits, thus incurring greater expense in time and money. The net effect has been that optimized and robust asynchronous circuits have not become a part of typical system designs. The asynchronous circuits that must be included are usually designed in an ad-hoc manner with many underlying assumptions. This is a highly error- prone process, and causes implementations to be unnecessarily delicate to delay variations. Field-programmable gate arrays, one of todays dominant media for prototyping and implementing digital circuits, are also inappropriate for constructing more than the simplest asynchronous interfaces. They lack the critical elements at the heart of todays asynchronous designs. Unfortunately, resolving this problem is not just a simple matter of adding these elements to the programmable array. The FPGA must also have predictable routing delay and must not introduce hazards in either the logic or routing. Futhermore, the mapping tools must also be modified to handle asynchronous concerns, especially the proper decomposition of logic to fit into the programmable logic blocks and the proper routing of signals to ensure that required timing relationships are met. Ideally, we need an FPGA that can support both synchronous and asynchronous circuits with comparable efficiency. As a step in this direction we present Montage, an integrated system of FPGA architecture and mapping software designed to support both asynchronous circuits and synchronous interfaces. The architecture provides circuits with hazard-free logic and routing, mutual exclusion elements to handle metastability, and methods for initializing unclocked elements. The mapping software generates placement and signal routing sensitive to the timing demands of asynchronous methods. With these features, the Montage system forms a prototyping and implementation medium for asynchronous designs, providing asynchronous circuits with a powerful tool from the synchronous designers toolbox.

The Triptych FPGA architecture

Field-programmable gate arrays (FPGAs) are an important implementation medium for digital logic. Unfortunately, they currently suffer from poor silicon area utilization due to routing constraints. In this paper we present Triptych, an FPGA architecture designed to achieve improved logic density with competitive performance. This is done by allowing a per-mapping tradeoff between logic and routing resources, and with a routing scheme designed to match the structure of typical circuits. We show that, using manual placement, this architecture yields a logic density improvement of up to a factor of 3.5 over commercial FPGAs, with comparable performance. We also describe Montage, the first FPGA architecture to fully support asynchronous and synchronous interface circuits.

SubGemini: Identifying SubCircuits using a Fast Subgraph Isomorphism Algorithm

The problem of finding subcircuits in a larger circuit arises in many contexts in computer-aided design. This is a problem currently solved using ad hoc techniques that rely on the circuit technology and implementation details. For example, channel graphs and signal flow are often used to extract simple gates from a transistor layout. Such techniques, however, do not generalize to different subcircuit structures and do not transfer to other technologies. We present a technology-independent algorithm for this problem based on a solution to the subgraph isomorphism problem. Although this problem is known to be NP-complete, our solution is very fast in practice for real circuits. We describe the algorithm, which uses an extension of the graph partitioning algorithm used by Gemini [3] for graph isomorphism, and present experimental results that show that the typical running time for large CMOS circuits is approximately linear in the total number of devices within the subcircuits being matched.

Specifying and compiling applications for RaPiD

Efficient, deeply pipelined implementations exist for a wide variety of important computation-intensive applications, and many special-purpose hardware machines have been built that take advantage of these pipelined computation structures. While these implementations achieve high performance, this comes at the expense of flexibility. On the other hand, flexible architectures proposed thus far have not been very efficient. RaPiD is a reconfigurable pipelined datapath architecture designed to provide a combination of performance and flexibility for a variety of applications. It uses a combination of static and dynamic control to efficiently implement pipelined computations. This control, however, is very complicated; specifying a computations control circuitry directly would be prohibitively difficult. This paper describes how specifications of a pipelined computation in a suitably high-level language are compiled into the control required to implement that computation in the RaPiD architecture. The compiler extracts a statically configured datapath from this description, identifies the dynamic control signals required to execute the computation, and then produces the control program and decoding structure that generates these dynamic control signals.

Hardware Acceleration of Short Read Mapping

Bioinformatics is an emerging field with seemingly limitless possibilities for advances in numerous areas of research and applications. We propose a scalable FPGA-based solution to the short read mapping problem in DNA sequencing, which greatly accelerates the task of aligning short length reads to a known reference genome. We compare the runtime, power consumption, and sensitivity of the hardware system to the BFAST and Bowtie software tools. The hardware system demonstrates a 250X speedup versus BFAST and a 31X speedup versus Bowtie on eight CPU cores. Also, the hardware system is more sensitive than Bowtie, which aligns approximately 80% of the short reads, as compared to 91% aligned by the hardware.