| 摘要 |
PURPOSE:To secure the accuracy information for each data and to ensure accuracy of each data by using a field to store the accuracy information on the exponent part variable length floating point data and a means to calculate accuracy of the arithmetic result. CONSTITUTION:A 1st data is set to a register 201 from a floating point register or a memory 50 via a signal line L101 with a 2nd data set to a register 202 via a signal line L102 respectively. The exponent part variable length data is sent to an exponent part variable length data computing element 204 for calculation via signal lines L105 and L106 out of those data set to registers 201 and 202. The result of this calculation is set to a register 205 via a signal line L109. The accuracy information on the data set to registers 201 and 202 are sent to an accuracy arithmetic circuit 203 via signal lines L103 and L104 and calculated there. The result of this calculation is set to the register 205 via a signal line L108. |
