Low-Power Vector Processors

Many future applications in the mobile and server segments will depend on and will benefit from vectorization and Single Instruction Multiple Data (SIMD) support. Specifically, graphics, gaming, signal processing and certain mining and recognition applications will be able to exploit those vector capabilities well. In this project, we are investigating how to provide complexity-effective vector support for these applications, which are not considered to be in the domain of classical vector applications. While certain vector implementations exist, the requirement of low-power and dynamic reconfigurability presents unique challenges. Besides addressing these challenges, we are also looking at how composable architectures could benefit from vector support.

Project Team: