LARGE 3-D ARCHITECTURE FOR PARALLEL PROCESSORS

THERE EXISTS A NEED FOR ANALYSIS OF NUCLEAR-WEAPON ENVIRONMENTS AND THEIR EFFECTS ON MILITARY SYSTEMS. IMPROVEMENTS IN PERFORMANCE OF COMPUTATIONAL TOOLS ARE NEEDED. HEREIN IN PRESENTED AN INNOVATIVE IDEA FOR PRODUCING SUCH AN IMPROVEMENT; IT WILL TAKE ADVANTAGE OF PARALLEL PROCESSORS. THE IMPORTANCE OF THIS WORK IS TO DEMONSTRATE AN APPROACH TO IMPROVE THE FIDELITY OF CALCULATIONS AND REDUCE THEIR COSTS. THIS PROPOSAL ADDRESSES AN INNOVATION THAT CAN BE APPLIED TO SOLVING THE NUMERICAL DIFFERENCEING EQUATIONS THAT REPRESENT THE DIFFERENTIAL EQUATIONS OF FLUID FLOW PHENOMENA. THIS SOLUTION SCHEME IS EXPECTED TO HAVE WIDESPAREAD APPLICABILITY TO BOTH EXISTING AND FUTURE COMPUTER PROGRAMS (CODES). THE SAVINGS IN COMPUTER TIME, WHEN USING THIS APPROACH WITH TYPICAL VECTOR PROCESSORS, WILL BE QUITE SIGNIFICANT. IT IS THESE SAVING THAT ARE IMPORTANT AND RELEVANT. IF SUBSTANTIAL, THEY WILL, FOR EXAMPLE, ALLOW USE OF BETTER CELL RESOLUTION THAT CAN, IN TURN, IMPROVE SOLUTION ACCURACY. FURTHER, AS DNA SHIFTS ITS WORK LOAD (ASSOCIATED WITH SOLVING THEIR COMPUTATIONAL FLUID DYNAMICS PROBLEMS) TO COMPUTERS WITH ADVANCED PARALLEL PROCESSING CAPABILITIES, THESE SAVINGS SHOULD INCREASE EVEN FURTHER.