Software programs are normally run on some type of computer, whether a mainframe computer, a minicomputer, or some type of microprocessor. This paper introduces a new computational architecture, called flowpaths that can implement software programs directly in an FPGA without the need for a microprocessor core. The resulting FPGA implementation will use fewer FPGA resources and will execute faster than executing the same software program on a microprocessor core within the FPGA. The method will be illustrated by compiling a FORTH program for the greatest common divisor directly to a Xilinx FPGA and the results compared to those using Handle-C. Flowpaths can be partitioned into program subsets allowing an FPGA to be reconfigured in real-time during execution of the program.
展开▼