C-to-silicon Compilation
L.E. Then, R.W. Brodersen · 1992
This Paper Presents a C-to-SiliCon design a numerical A lication Specific 8rocessor. The proces- sor,, called PURA, computes inverse kinematics for the Unimation Puma 560 based On an algorithm described by 500 lines of code. The chig is automatically generated from the high-level C description of the algorithm and a hi h-level, user-cus- tomizable architecture description &e for the target architecture. The rocessor has been fabricated and tested, and is fully gnctional. cia1 md (machine description) language which allows even pages of text. The following is the machine description of the address computation unit of the PUMA datapath, which is shown in Figure 1. #define bus node : delay=o and how the system was a plied to the Of fairly complicated processors to be described in a few