| 摘要 |
PROBLEM TO BE SOLVED: To suppress such problem that a long time is required for arithmetic operation or a large scale circuit is necessary in a conventional dividing circuit. SOLUTION: Concerning this method, an initial value is calculated through a computing element by Newton-Raphson method based on a value retrieved from a compact table stored in a memory. Besides, the repeat calculation circuit of Newton-Raphson method itself is packaged as an arithmetic circuit, an inverse calculating circuit capable of shortening the arithmetic time, being made a pipeline and improving the throughput and a dividing circuit using that circuit are packaged and by caring of arithmetic accuracy in that case, one part of a multiplying circuit included in the circuit is omitted so that a compact circuit can be provided. The arithmetic accuracy is easily improved by utilizing the interim result of calculation. The method is provided for calculating the minimum size required for a table to be used for calculating the initial value and a means for designing irreducibly minimum circuits while satisfying the requested arithmetic accuracy is applied as well. |
