Due to the fast changing wireless communication standards coupled with strict performance constraints, the demand for flexible yet high-performance architectures is increasing. To tackle the flexibility requirement, software-defined radio (SDR) is emerging as an obvious solution, where the underlying hardware implementation is tuned via software layers to the varied standards depending on power-performance and quality requirements leading to adaptable, cognitive radio. In this paper, we conduct a case study for representatives of two complexity classes of WCDMA channel estimation algorithms and explore the effect of flexibility on energy efficiency using different implementation options. Furthermore, we propose new design guidelines for both highly specialized architectures and highly flexible architectures using high-level synthesis, to enable the required performance and flexibility to support multiple applications. Our experiments with various design points show that the resulting architectures meet the performance constraints of WCDMA and a wide range of options are offered for tuning such architectures depending on power/performance/area constraints of SDR.

No abstract

High Level Synthesis (HLS) and design is gaining traction in commercial and academic circles, as an answer to increasing design complexity and short time-to-market. In this paper, we present a short survey on the HLS landscape and propose modeling concepts to extract and exploit the inherent flexibility for a commercially available high-level design tool, to explore ASIC and CGRA besides native ASIP support. Structural descriptions, representation, flexibility and limitations are discussed. Several case studies help highlight the advantages of the proposed methods, providing a solid framework to aid a broader and faster design space exploration.

A new layered reconfigurable architecture is proposed which exploits modularity, scalability and flexibility to achieve high energy efficiency and memory bandwidth. Using two flavors of Column-wise Givens rotation, derived from traditional Fast Givens and Square root and Division Free Givens Rotation algorithms the architecture is thoroughly evaluated for scalability, speed, area and energy. Combining an efficient mapping strategy of the highly parallel algorithms capable of annihilation of multiple elements of a column of the input matrix and using the new features of the architecture, 9 architectural variants were explored achieving a clean trade-off of execution speed versus area, while keeping relatively constant energy.

No abstract