Coarse-Grained Reconfigurable Accelerators

The question we’d like to answer in this project is “What needs to be accelerated“.

This question is especially important with the rise of edge processors and accelerators that will be used in them. We propose increasing accelerators’ workload coverage by designing Coarse-Grained Reconfigurable Accelerators.

                      Fgure1: instead of designing two separate accelerators for two workloads, one CGRA can accelerate both workloads

 

Figure2: Our benchmarks are a combination of Machsuite and CortexSuite

Although mapping We are in the process of designing a tool that finds similarities. We are using CLANG’s back-end

Publications:

  • P. Mokri and M. Hempstead, “Fingerprinting Coarse-Grained Reconfigurable
    Accelerators Using Data Movement and Structural Similarities in Applications” Boston Area Architecture Workshop (BARC), Jan. 2016.
  • P. Mokri and M. Hempstead, “Stockpile Of Accelerators: A Methodology To Increase Accelerators’ Coverage”, Boston Area Architecture (BARC) Workshop, January 2016.