Maxeller, with the MaxJ compiler, changes FPGAs into Data Flow Engines, with a dataflow programming language. As shown in the FlexHH paper (doi: 10.1109/ACCESS.2020.3007019), Dataflow Computing fits reconfigurable neural simulation extremely well. As an abstraction over large populations of single cells, neural-mass modeling retains most of the computational properties of neural modeling. As such, the Maxeler approach to neural-mass simulation could very well bring great speedups to this clinically relevant computational problem.
Job Category: MSc Thesis topic
Job Type: Student Thesis