Modern computational neuroscientists are in dire need of powerful computing machines that are able to achieve high-performance simulations of large as much as detailed representations of brain regions. The current state of the art finds them either with simulation platforms that are easy to use but score moderate simulation speeds or with powerful machines which, however, are impractical to harness by not technically savvy researchers. To tackle this problem and bring a powerful tool directly to the hands of neuroscientists, we have developed BrainFrame, a comprehensive software toolflow resting on top of award-winning, FPGA-based computing platforms built by Maxeler Technologies. This toolflow is based on PyNN, a Python-based, simulator-independent language for specification of brain models. PyNN has been highly embraced by the bulk of the neuroscience community as it already offers a common interface to popular simulation platforms such as NEURON and NEST. This report concludes the ERC-PoC project “BrainFrame” (Project-ID no: 641000) whose mission is the commercialization of the BrainFrame toolflow. The report details the BrainFrame concept and the novelty it brings about. It, further, embeds a technical report and validation proof of the new toolflow, demonstrating its impressive capabilities. The discussion is complemented with a market survey and an exploration for a business case for BrainFrame. The report concludes with publicity activities related to BrainFrame in the form of scientific output, news flashes and embedding in new research efforts.
