SpaceStudio™ is a C/C++ framework to write or refactor legacy applications in order to achieve performance goal (i.e. speedup) using CPU, GPU or FPGA without knowing the underlying details of those technologies. Applications are intuitively partitioned to seamlessly target heterogeneous platform (CPU, GPU or FPGA). Designers explore, analyze, profile and validate designs using SpaceStudio™.
SpaceStudio™ lets designers explore heterogeneous FPGA architectures of different vendors (e.g., Xilinx and Intel Altera), targeting system-on-board (processor and FPGA into a board) and system-on-chip (processor and FPGA into a chip). Architectures designed with SpaceStudio™ can be exported down to the implementation to either Xilinx Vivado or Intel Altera Quartus II.
SpaceStudio™ offers an extensive non-intrusive monitoring infrastructure that gives qualitative insights about the system. Code-level directives can be added to obtain application-specific informations. Gathered informations is stored in a SQL database where designers can perform queries upon to guide the design exploration process to achieve their performance goal. Designers can rely on the obtained information during simulation and take early key decisions with confidence.
SpaceStudio’s™ automatic code generation process generates quality code for virtual platform, communication interfaces, drivers, bootloader and also project exportation for downstream implementation tools. Designers explore more designs compared to the traditional approach since SpaceStudio™ handles error-prone manipulations while keeping control over key decisions. SpaceStudio™ generates everything designers would have done by hand. Designers focus on their application rather than cumbersome architecture’s details.