Scalable Softcore Vector Processor for Biosequence Applications

Abstract

Abstract-In this paper we describe the SVP, a Softcore Vector Processor targeted toward Computational Biology and streaming applications. The SVP is a software programmable architecture constructed from predefined hardware building blocks. We leverage the flexibility and power of an FPGA to enhance a streaming vector processor design. Each functional unit includes an instruction controller, parallel processing elements with shared registers, and a memory unit which provides access to both local and streaming data. We target the Smith-Waterman sequence alignment algorithm and report estimated performance numbers based on a Virtex-4 FPGA implementation. I. INTRODUCTIONThe Softcore Vector Processor was developed in response to the need for accelerating a wide range of computational biology applications. The design philosophy of the SVP is based on two important goals: adaptability and high performance.The SVP is designed to be a software programmable and hardware customizable platform. Instruction based execution allows support for a large number of algorithms in computational biology. However, the fact that different classes of applications require different subsets of hardware resources argues for a customizable hardware design built from primitives. The second goal of the project is to achieve programmability without sacrificing performance. The SVP was designed to perform competitively with full custom solutions available today.Currently available genome databases are growing exponentially in size [1], making it difficult for software analysis tools to keep up. A number of hardware solutions utilizing special purpose VLSI or reconfigurable hardware such as Field Programmable Gate Arrays (FPGAs) have been proposed to bridge this gap. Systems such as the BISP, and the Kestrel [2], built on custom asics support high performance designs but are expensive to manufacture and cannot be adapted for applications that require differing hardware resources. FPGA solutions tend to be specialized, with support for new applications requiring a laborious re-design of the hardware.The SVP is a vector processor targetting algorithms that can be easily mapped to a systolic array of processing elements (PEs). Programmable instructions execute on PEs that operate on different units of data achieving fine grained parallelism. Sequence analysis tools commonly use the streaming paradigm, where a large genomic database is streamed through

Mapping a VLIW×SIMD Processor on an FPGA: Scalability and Performance

“…At the same time vector processing techniques on FPGAs were investigated mostly for purpose of specific applications like video processing, algebraic equation solving etc. [3,4,5]. This work investigates the possibilities and limitations of combined VLIW/SIMD processing on Xilinx Virtex-4 FPGA family concerning its performance and scalability.…”

ATCS: Auto-Tuning Configurations of Big Data Frameworks Based on Generative Adversarial Nets

“…parallelism is large, the performance of Spark does not significantly improve, or even decreases, because there is not enough memory to run many tasks while the program is running. Also, every big data application requires a specific subset of hardware resources [6]. For example, Word-Count requires a subset of resources based on CPU [7], and TeraSort requires a subset of resources based on CPU and memory [5], [7], so the optimal configuration of different workloads (e.g., WordCount, TeraSort) is nonconformity.…”