AMIDAR Project: Lessons Learned in 15 Years of Researching Adaptive Processors
Dennis Leander Wolf, Lukas Johannes Jung, Tajas Ruschke, Changgong Li, Christian Hochberger · 2018
This paper is about AMIDAR, a processor concept for an adaptive processor that uses a Coarse Grained Reconfigurable Array as a configurable accelerator to dynamically speed up hot spots of arbitrary applications. We describe the concepts and implementation details and we summarize some central findings of more than 15 years of research around this topic. Eventually, we demonstrate a fully working hardware implementation and show the potential effect of a configurable hardware accelerator.