PROJECT CRISP

THE OBJECTIVES OF THE PROPOSED RESAERCH ARE TO DESIGN A PARALLEL PROGRAMMING LANGUAGE, CALLED "CRISP," THAT SUPPORTS EXISTING C PROGRAMS AS A SUBSET OF THE LANGUAGE. CRISP SUPPORTS AUTOMATIC SYNCHRONIZATION, MULTIPLE GRAIN-SIZES OF PARALLELISM, AND A LARGE LIBRARY OF COMMON PARALLEL AND SERIAL PROGAMMING CONSTRUCTS. THE PROPOSED STRATEGY FOR THIS PROJECT IS BASED ON UTILIZING AN EXISTING COMMERCIALLY AVAILABLE PARALLEL COMMON LISP SYSTEM AND DESIGNING A SYSTEM THAT TRANSLATES CRISP INTO COMMON LISP. THIS RESEARCH WILL RESULT IN THE DESIGN OF C-COMPATIBLE SYSTEMS THAT INCLUDE MANY OF THE ADVANCED CAPABILITIES OF A COMMON LISP SYSTEM. FEATURES SUCH AS OBJECTS, POLYMORPHISM, GARBAGE COLLECTION, INCREMENTAL COMPILATION, AND INTERACTIVE DEBUGGING THAT ARE INCREASINGLY BEING INCORPORATED INTO EXISTING C SYSTEMS. IN ADDITION TO THESE CAPABILITIES, CRISP ALLOWS THE USE OF FUTURE OBJECTS WHICH PROVIDE FOR AUTOMATIC SYNCHRONIZATION OF PARALLEL ACTIVITY, AND THE USE OF FINE-GRAINED PARALLELISM ALREADY PRESENT IN THE LIST SYSTEM. THIS PROJECT WILL ALLOW THE LARGE NUMBER OF EXISTING C PROGRAMMERS TO DEVELOP PARALLEL SOFTWARE USING A FAMILIAR, BUT MORE POWERFUL, PROGRAMMING LANGUAGE. ANTICIPATED BENEFITS/POTENTIAL COMMERCIAL APPLICATIONS - THIS PROJECT WILL EXTEND THE POWER OF THE C PROGRAMMINGLANGUAGE BY ENABLING C PROGRAMS TO EXPLOIT THE INCREASED COMPUTE POWER OF PARALLEL HARDWARE SYSTEMS AND BY PROVIDING ADVANCED PROGRAMMING CAPABILITIES FROM THE COMMON LIST LANGUAGE. WITH THE RAPIDLY GROWING AVAILABILITY OF PARALLEL HARDWARE SYSTEMS, THE PARALLEL CAPABILITIES THIS PROJECT WILL BRING TO THE C LANGUAGE WILL RESULT IN SIGNIFICANT COMMERCIAL DEMAND.