Abstract:
© Springer International Publishing AG, part of Springer Nature 2018. In this work, we present the use of a new high-level synthesis engine capable of generating resource-shared compute accelerators, even from very complex double-precision codes, for cell biology simulations. From the domain-specific CellML description, the compilation pipeline is able to generate hardware that is shown to achieve a performance similar to or exceeding current generation desktop CPUs, and has energy savings of up to 96% even for a single accelerator, which requires just 25–30% area on a mid-sized FPGA.