Optimizing Decimal Fixed and Floating Point Arithmetic for Computers
Shi Xue · 2005
Fixed point arithmetic is very boportant for business computing. And also some embedded chips do not provide float unit because of the limit of cost and power. Under both these conditions,we must find a solution to sup- port decimal fixed or float point arithmetic on Integer ALU. This paper presents a new approach based on decimal en- coded long integer to do decimal arithmetic. Experiments show that our data model and algorithms is superior to Java BigDecimal in performance. We get a speedup of 20% in real commercial applications.