| 摘要 |
PURPOSE:To improve the precision of floating-point arithmetic by providing a first storage part where input data and the intermediate result and the final result of arithmetic are stored and a second storage part where inputted/ outputted protection digits are stored. CONSTITUTION:When arithmetic results (intermediate results) are outputted from floating-point computing elements 1a and 2a, protection digits are outputted also, and arithmetic results are stored in a first storage part 5 through a data bus, and protection digits are stored in a second storage part 8. When these results will be used for arithmetic, arithmetic results from the first storage part 5 and protection digits from the second storage part 8 are taken into computing elements 1a and 2a through the data bus to execute floating-point arithmetic. Consequently, the practical number of effective digits is increased because information of protection digits for each arithmetic is not lost. Thus, the precision of arithmetic results is improved. |
