Many DOD systems require advanced digital signal processing and image processing functions that cannot be efficiently implemented on conventional micro-processors, hence designers have started mapping these applications onto FPGAs. However, most FPGA implementations are manually designed and highly coupled to the hardware, often taking advantage of special hardware features of the target FPGA. The manual design of such highly optimized hardware on FPGAs requires design times of the order of months. In this research, we propose a novel methodology and system level tool to design applications on FPGAs by taking software specifications in the form of binary and assembly implementation on a conventional microprocessor, performing high-level synthesis, and automatically generating Register Transfer Level (RTL) VHDL and Verilog code. The RTL code can be synthesized by commercial backend logic synthesis and physical synthesis tools automatically onto FPGAs This revolutionary methodology can reduce the design times for new hardware designs and hardware upgrades from months to hours. The RTL code that is synthesized can be automatically verified for correctness using a simulation based methodology that creates the testbenches and proves the bit-true correctness of the synthesized hardware. As part of this automated flow, our system-level design tool will provide the user with high-level estimates of area, delay and power consumption using which various design tradeoffs can be rapidly explored by the designer.
Keywords: Behavioral Synthesis, High Level Synthesis, Fpga Synthesis, Binary Translation, Fpga Design Productivity, Embedded Applications.