Abstract
Field-programmable gate array (FPGA) devices boast abundant resources with which custom accelerator components for signal, image, and data processing may be realized; however, realizing high-performance, low-cost accelerators currently demands manual register transfer level design. Software-programmable soft processors have been proposed as a way to reduce this design burden, but they are unable to support performance and cost comparable to custom circuits. This paper proposes a new soft processing approach for FPGA that promises to overcome this barrier. A high-performance, fine-grained streaming processor, known as a streaming accelerator element, is proposed, which realizes accelerators as large-scale custom multicore networks. By adopting a streaming execution approach with advanced program control and memory addressing capabilities, typical program inefficiencies can be almost completely eliminated to enable performance and cost, which are unprecedented among software-programmable solutions. When used to realize accelerators for fast Fourier transform, motion estimation, matrix multiplication, and sobel edge detection, it is shown how the proposed architecture enables real-time performance and with performance and cost comparable with hand-crafted custom circuit accelerators and up to two orders of magnitude beyond existing soft processors.
Highlights
R ECENT years have seen rapid evolution in fieldprogrammable gate array (FPGA) technologies, both in the scale of traditional devices and their extension to systemon-chip FPGA, incorporating heterogeneous multicore processor architectures
By ensuring absolute lowest cost FPGA PE (FPE) structure, the economies of scale produce dramatic reductions in multicore resource cost. This extreme focus comes at the cost of flexibility: once synthesized, the FPE does not exhibit the same degree of flexibility as a general soft processor because the architecture is highly constrained at design time to support the desired operation with highest performance and lowest cost; while it may be reprogrammed after synthesis, it cannot enable general-purpose operation in the manner of a standard softcore
Soft processors for FPGA suffer from substantial cost and performance penalties relative to custom circuits handcrafted at register transfer level
Summary
R ECENT years have seen rapid evolution in fieldprogrammable gate array (FPGA) technologies, both in the scale of traditional devices and their extension to systemon-chip FPGA, incorporating heterogeneous multicore processor architectures. These mark FPGA as ideal hosts to high-performance custom computing architectures for signal, image, and data processing [4]. A novel streaming accelerator element (SAE) is presented which enables software-based accelerator development, while maintaining the performance and cost of custom circuits. To the best of our knowledge, the SAE is the highest performance, lowest cost software-programmable component on record for FPGA and the first to enable signal and image processing accelerators with performance and cost comparable with custom circuits. WANG AND MCALLISTER: STREAMING ELEMENTS FOR FPGA SIGNAL AND IMAGE PROCESSING ACCELERATORS
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have